Volume 64, Issue 10 p. 825-834
Critical Review
Free Access

TAK1, more than just innate immunity

Liang Dai

Liang Dai

Division of Molecular and Cell Biology, School of Biological Sciences, Nanyang Technological University, Singapore

Search for more papers by this author
Chan Aye Thu

Chan Aye Thu

Division of Molecular and Cell Biology, School of Biological Sciences, Nanyang Technological University, Singapore

Search for more papers by this author
Xin-Yu Liu

Xin-Yu Liu

Division of Molecular and Cell Biology, School of Biological Sciences, Nanyang Technological University, Singapore

Search for more papers by this author
Jiajia Xi

Jiajia Xi

Division of Molecular and Cell Biology, School of Biological Sciences, Nanyang Technological University, Singapore

Search for more papers by this author
Peter C. F. Cheung

Corresponding Author

Peter C. F. Cheung

Division of Molecular and Cell Biology, School of Biological Sciences, Nanyang Technological University, Singapore

Division of Molecular and Cell Biology, School of Biological Sciences, Nanyang Technological University, 60 Nanyang Drive, 637551, SingaporeSearch for more papers by this author
First published: 03 September 2012
Citations: 140

Abstract

Transforming growth factor β-activated kinase 1 (TAK1) is a key regulator of the innate immunity and the proinflammatory signaling pathway. In response to interleukin-1, tumor necrosis factor-α, and toll-like receptor agonists, it mediates the activation of the nuclear factor κB (NF-κB), c-Jun N-terminal kinase (JNK), and p38 pathways. In addition, TAK1 plays a central role in adaptive immunity, in which it mediates signaling from T- and B-cell receptors. This review will focus on recent developments and also examine the regulation of TAK1 in response to a diverse range of other stimuli including DNA damage, transforming growth factor-β, Wnt, osmotic stress, and hypoxia. © 2012 IUBMB IUBMB Life, 2012, 64(10):825–834, 2012

Abbreviations

AMPK, AMP-activated protein kinase; ATM, ataxia-telangiectasia-mutated kinase; BCR, B-cell receptor; BIR, baculovirus IAP repeat; BMP, bone morphogenetic protein; CaMKII, Ca2+/calmodulin-dependent protein kinase II; Cyld, cylindromatosis; CUE, coupling of ubiquitin conjugation to endoplasmic reticulum degradation; GSK3, glycogen synthase kinase 3; JNK, c-Jun N-terminal kinase; IKK, IκB kinase; IL-1, interleukin-1; IRAK, interleukin-1-receptor-associated kinase; Itch, itchy homolog of E3 ubiquitin ligase; LPS, lipopolysaccharide; LUBAC, linear ubiquitin chain assembly complex; MKK, mitogen-activated protein kinase kinase; MEF, mouse embryonic fibroblast; NEMO, NF-κB essential modulator; NFAT, nuclear factor of activated T cells; NLK, nemo-like kinase; NZF, Npl4 zinc finger; NF-κB, nuclear factor κB; PAMP, pathogen-associated molecular patterns; RCAN1, regulator of calcineurin 1; RIP, receptor-interacting protein; SnoN, Ski-related novel protein N; TAK1, TGF-β-activated kinase 1; TAB1/TAB2/TAB3, TAK1-binding protein-1, -2, and -3; TAO2, thousand-and-one amino acids 2; TCF, transcription factor; TCR, T-cell receptor; TLR, toll-like receptor; TRAF, TNF-receptor-associated factor; TGF-β, transforming growth factor-β; TNF-α, tumor necrosis factor-α; TRAIL, TNF-related apoptosis-inducing ligand; Ubc13, ubiquitin-conjugating enzyme 13; Uev1a, ubiquitin E2 variant 1a; USP4, ubiquitin-specific peptidase 4; XIAP, x-linked inhibitor of apoptosis protein.

DISCOVERY OF TRANSFORMING GROWTH FACTOR-β-ACTIVATED KINASE 1

Transforming growth factor-β (TGF-β)-activated kinase 1 (TAK1, also called MAP3K7) was discovered in 1995 as a protein kinase that can be activated by TGF-β (1). It was subsequently found to be a critical mediator of the inflammatory response as it is robustly activated by lipopolysaccharide (LPS) and the cytokines tumor necrosis factor-α (TNF-α) and interleukin-1 (IL-1). In cells, TAK1 is bound constitutively to the accessory protein TAB1 (1, 2) and to the homologs TAB2 or TAB3 so that TAK1 can exist as heterotrimeric complexes composed of TAK1–TAB1–TAB2 or TAK1–TAB1–TAB3 (3, 4). Whether TAK1 can exist physiologically as heterodimers, for example, TAK1–TAB1, TAK1–TAB2, or TAK1–TAB3, is unclear. The accessory protein's roles with regard to TAK1 function will be explained in the following sections.

INNATE IMMUNITY AND PRO-INFLAMMATORY SIGNALING PATHWAY

The innate immunity pathway is the host's first line of defense against pathogens and is activated in response to infection. Recognition of these microbes is attributed to conserved molecular structures present in pathogens termed pathogen associated molecular patterns (PAMPs), which are detected by host pattern recognition receptors (PRR) expressed in immune cells. Among these PAMPs, TAK1 can be activated by toll-like receptor (TLR) agonists such as LPS, which is found on the cell wall of Gram-negative bacteria. TAK1 plays a key role in relaying the signal that results in the synthesis of cytokines, which recruit and mobilize immune cells to eliminate microbes and fight infection and act on endothelial cells to promote blood flow. The cytokines IL-1 and TNF-α use similar regulatory mechanisms to activate TAK1. Other TLR agonists such as CpG DNA (5), PolyIC (5), and muramyl dipeptide (MDP) (6), a nucleotide oligomerization domain-like receptor (NLR) agonist, have also been shown to activate the TAK1 pathway. Thus, in addition to TLR ligands, TAK1 is activated by and induces the expression of cytokines. TAK1 is first and foremost associated with inflammation, and studies in this area have provided most of the knowledge regarding TAK1 regulation and function.

ACTIVATION OF TAK1 BY K63-LINKED POLYUBIQUITIN CHAINS

Binding of IL-1 and LPS to their receptors activates TAK1 through a common pathway whereby the kinases interleukin-1-receptor-associated kinase 1 (IRAK1) and IRAK4 recruit the E3 ubiquitin ligase TNF-receptor-associated factor 6 (TRAF6) and its accessory factors ubiquitin-conjugating enzyme 13 (Ubc13)/ubiquitin E2 variant 1a (Uev1a) (Fig. 1). TNF-α activates TAK1 via a similar mechanism, which is mediated by receptor-interacting protein kinase 1 (RIP1) and TRAF2. Activation of the ubiquitin ligase activities of TRAF2, and TRAF6 with accessory proteins Ubc13/Uev1a leads to the generation of K63-linked polyubiquitin chains, which bind to the adaptor proteins TAB2 or TAB3. Although TAB2 and TAB3 possess coupling of ubiquitin conjugation to endoplasmic reticulum degradation (CUE) and Npl4 zinc finger (NZF) ubiquitin binding domains at the N- and C-terminus, respectively (3), K63-linked polyubiquitin binds with much higher affinity to the NZF domain (7). The functional significance of the CUE domain is currently unclear as its deletion does not affect IκB kinase (IKK) activation by TAK1 (8). A coiled coil domain is also found in the C-terminus of TAB2 and TAB3, which harbors the TAK1 binding region (Fig. 2), whereas TAB2 and TAB3 bind to residues 479–547 in the C-terminus of TAK1 (9). Once activated, TAK1 transduces the signal to nuclear factor κB (NF-κB), c-Jun N-terminal kinase (JNK), and p38 via phosphorylation of IKK, mitogen-activated protein kinase kinase 4/7 (MKK4/7), and MKK3/6, respectively. Ultimately, NF-κB and other transcription factors downstream of p38 and JNK are activated, resulting in the transcription of genes important for inflammatory and immune responses.

Details are in the caption following the image

Activation of TAK1 by TNF-α, IL-1, and TLR agonists. Engagement of the receptors with ligands leads to activation of the ubiquitin ligases TRAF2 and TRAF6. In conjunction with Ubc13/Uev1A, K63-linked polyubiquitin chains are synthesized that bind to TAB2 and lead to the autoactivation of TAK1. JNK can be activated by MKK4 or MKK7, whereas p38 can be activated by MKK3 or MKK6. For simplicity, TAB3 is not shown but can form a heterotrimeric complex with TAB1 and TAK1.

Details are in the caption following the image

Domain structure and post-translational modifications of TAK1, TAB1, TAB2, and TAB3. Residues that are phosphorylated, ubiquitylated, or cGlcNac are denoted by (P), (Ub), and (oGlcNAc), respectively. Binding domains are denoted by BD. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

The seminal discovery that TAK1 is activated by K63 polyubiquitin chains was pivotal as it was the first time that ubiquitin had been shown to play a direct role in signal transduction (1). In contrast, proteins covalently tagged with K48-linked polyubiquitin chains are recognized by the proteasome for destruction. The mechanism by which TAK1 is activated when unanchored K63 polyubiquitin chains bind to TAB2 and TAB3 remains unclear (10). However, it triggers conformational changes in TAK1 that lead to autophosphorylation primarily on Thr187 and also on Thr178, Thr184, and Ser192 (2, 11, 12). There is evidence to suggest that K63 polyubiquitin chains sequester TAK1 complexes, facilitates their oligomerization, and may also function as a scaffold that aids TAK1 phosphorylation of the IKK complex, which consists of IKKα, IKKβ, and IKKγ [also known as NF-κB essential modulator (NEMO)].

Structural studies have shown that the NZF domain of TAB2 binds to two molecules of ubiquitin via two interdependent binding sites (7, 13). Although they can bind to monoubiquitin, the affinity for K48 and K63 polyubiquitin chains is 50- and 150-fold higher, respectively (7). Thus, K63 polyubiquitin binding to TAB2 and TAB3 is the modus operandi by which TAK1 is activated. The identity of individual E3 ligases responsible for activating TAK1 by TGF-β, DNA damage, and other stimuli will be discussed in later sections.

TAB1 AND TAK1 ACTIVITY

TAB1 is an adaptor that does not possess enzymatic activity but is constitutively bound to TAK1 in cells. Early reports had shown that TAK1 activity is greatly increased when overexpressed with TAB1 (2). In fact, the C-terminal 68 residues of TAB1 are sufficient to activate TAK1 through relieving autoinhibition by the N-terminal domain of TAK1 (Fig. 2) (2). This interaction is shown by the crystal structure of a TAK1–TAB1 fusion protein (14). Although TAB1 does not display homology to any other protein in the human proteome, the crystal structure of the individual TAB1 protein revealed that it contains folds with most similarity to the protein phosphatase PP2Cα; however, importantly, TAB1 is devoid of phosphatase activity because critical residues are missing (15). Several reports indicate that the requirement for TAB1 in TAK1 activation is stimulus dependent. TAB1 is indispensable for TGF-β or osmotic stress-induced activation of TAK1 (16, 17). In TAB1−/− cells, IL-1 and TNF-α activation of TAK1 as measured by Thr187 phosphorylation is normal (16). However, another study in which TAK1 was assayed by in vitro phosphorylation of MKK6 showed that TAB1 is required for IL-1 and TNF-α activation of TAK1 (18). Using RNAi knockdown of TAB1, IL-1-stimulated cytokine secretion was decreased; however, activation of the NF-κB and MAPK pathways were unaffected (19). Therefore, further studies are needed to clarify the exact role of TAB1. The possibility remains open that TAB1 acts as a link with upstream regulators of TAK1 in a stimulus-dependent manner. In fact, x-linked inhibitor of apoptosis protein (XIAP) is an upstream mediator of TAK1 signaling by DNA damage, TGF-β, and hypoxia (Figs. 3B, 3D, and 3F), and the crystal structure of a XIAP–TAB1 fragment has been solved (20). Inaddition, post-translational modification of TAB1 by phosphorylation can modulate TAK1 activity (21). Recently, O-GlcNAcylation of TAB1 on Ser395 has been reported to be important for the activation of TAK1 induced by IL-1 and osmotic stress (22).

Details are in the caption following the image

Activation of TAK1 by different stimuli. Refer to main text for details.

KNOCKOUT MICE OF TAK1, TAB1, AND TAB2

The TAK1−/−, TAB1−/−, and TAB2−/− knockout mice are embryonic lethal, demonstrating that they are all required for proper development of the embryo (23). The TAK1 knockout mice survive until E9.5–E10.5, and derived TAK1−/− mouse embryonic fibroblasts (MEFs) display severely compromised activation of NF-κB and JNK showing that TAK1 plays a critical role in the immune response and cytokine synthesis (23). However, TAK1-mediated activation of the p38 pathway is cell type dependent. The TAB1−/− mouse dies at late stage of gestation between E15.5 and E18.5 (23) with abnormalities in both cardiac and lung development (23), whereas the TAB2−/− mouse dies from liver defects at E12.5 (23). The role of TAB2 remains unclear as it is thought that TAB3 may play a redundant role by compensating for some of TAB2′s function in its absence. This is underlined by the fact that although TAB2 ablation affects embryonic development, IL-1 and TNF-α signaling pathways were unaffected in TAB2−/− cells (23). Therefore, readers must bear in mind that effects may only become apparent when both TAB2 and TAB3 are absent because of possible compensatory mechanisms.

ADAPTIVE IMMUNITY

The adaptive immune system is composed of T and B cells that are activated in response to T-cell receptor (TCR) and B-cell receptor (BCR) antigens displayed by antigen-presenting cells. Postactivation, the cells undergo clonal expansion and proliferation. In contrast to the TLR/IL-1 pathway, TAK1 is activated by upstream molecules comprising protein kinase C theta (PKC theta), CARD-containing MAGUK protein 1 (CARMA1), and the ubiquitin ligase mucosa-associated lymphoid tissue lymphoma translocation protein 1 (MALT) (Fig. 3A). TCR ligation leads to TAK1-mediated activation of both NF-κB and JNK (24), which is required for the development and maturation of T cells (25). There is some discrepancy over TAK1 function in B cells. One study found that TAK1-deficient B cells develop relatively normally; however, BCR-mediated proliferation is severely impaired (5). Interestingly, BCR activation of TAK1 in B cells only signals to JNK but not NF-κB (5, 24-26). However, Schuman et al. reported that TAK1 is critical for B-cell maturation and BCR-induced activation of both JNK and NF-κB (27). This discrepancy may be due to the use of mice that were generated using different targeting constructs. One study expressed a truncated form of TAK1 in which exon 2 was deleted so that it lacked kinase activity (5), whereas in another study, TAK1 expression was abolished due to deletion of exon 1 (27). Nevertheless, both T and B cells lacking TAK1 are prone to apoptosis on stimulation with TCR and BCR ligands, which is consistent with the prosurvival role of TAK1 due to its ability to activate the NF-κB pathway (24, 26, 27).

DNA DAMAGE AND TAK1

Although the function of TAK1 is well known with regard to its role in innate immunity, a hitherto unknown role of TAK1 in the DNA damage response has been uncovered. In cells, the DNA repair and cell cycle checkpoint pathways are highly coordinated. DNA damage by ionizing radiation or genotoxins leads the cell to undergo cell cycle arrest to allow DNA repair to take place prior to DNA replication or commit to apoptosis if the damage is beyond repair. Activation of the NF-κB pathway by genotoxic stress is a critical event as it promotes cell survival and is a factor in tumor resistance to chemotherapy and radiotherapy.

Three studies have shown that TAK1 is required for NF-κB activation in response to genotoxic insult (Fig. 3B) (28, 29, 30). However, the actual upstream events and E3 ubiquitin ligase responsible for activating TAK1 vary in these reports, possibly a reflection of the different cell types and stimuli used, that is, ionizing radiation as opposed to genotoxic drugs. Different pathways may operate as the magnitude and duration of TAK1 activation have been shown to differ in response to ionizing radiation or drugs such as doxorubicin (29). In response to genotoxic drugs, ataxia-telangiectasia-mutated (ATM) kinase is exported from the nucleus, and an ATM–NEMO–XIAP–Ubc13 complex is assembled that covalently modifies the ELKS (a protein rich in glutamate, leucine, lysine, and serine) adaptor with K63-linked polyubiquitin chains (28). Downstream activation of TAK1 proceeds when TAB2 or TAB3 binds to K63 polyubiquitin chains linked to ELKS (28). This is in contrast to cytokine or TLR ligand activation of TAK1 in which TAB2 and TAB3 bind to unanchored K63 polyubiquitin chains (10). IKK phosphorylation by TAK1 is also facilitated by binding of NEMO to K63 polyubiquitin chains conjugated to ELKS (28). A separate study has shown that an ATM–TRAF6–Ubc13 module stimulates TAB2-dependent TAK1 phosphorylation in response to ionizing radiation (30). Whether this module functions exclusively in response to ionizing radiation is unclear. Yang et al. showed that ubiquitin-modified RIP1 may also be a component of the TAK1 activating complex as TAK1 is not activated in RIP1−/− cells by etoposide treatment (29). Recently, another E3 ligase called linear ubiquitin chain assembly complex (LUBAC) has been shown to be important for genotoxic activation of TAK1 (31). In contrast to XIAP and TRAF6, LUBAC synthesizes linear polyubiquitin chains that link the C-terminal glycine of one ubiquitin with the N-terminal methionine of another ubiquitin molecule (31). Therefore, it appears that DNA damage activates TAK1 by similar mechanisms to those found in the IL-1/TNF-α/TLR pathways, that is, via ubiquitin ligase synthesis of polyubiquitin chains.

TAK1 IN THE Wnt SIGNALING PATHWAY

The Wnt pathway is involved in growth, proliferation, and differentiation of cells. Deregulation of Wnt signaling can cause developmental defects and cancer. In the canonical Wnt1 pathway, inhibition of glycogen synthase kinase 3 (GSK3) leads to the stabilization and accumulation of β-catenin. In the nucleus, transcriptional activation of Wnt1 target genes occurs when β-catenin combines with coactivators such as TCF. In contrast, the noncanonical pathway does not signal through GSK3 and can antagonize the canonical pathway.

The first indication that TAK1 may function in Wnt signaling came with its identification of a role in cell specification in C. elegans where it functions upstream of nemo-like kinase (NLK) (32). In mammalian cells, TAK1 was initially found to be activated by a noncanonical Wnt pathway to antagonize canonical Wnt1 signaling (32). In this setting, a Wnt5a-activated Ca2+/calmodulin-dependent protein kinase II (CaMKII)–TAK1–NLK pathway phosphorylates and suppresses β-catenin transcriptional activity by phosphorylation of its coactivator TCF (Fig. 3C) (32). It is unclear whether CaMKII directly activates TAK1 or if an intermediate E3 ubiquitin ligase is involved. At the present time, whether TAK1 directly activates NLK remains unresolved. The same pathway is also used to specify osteoblasts from bone marrow mesenchymal stem cells (33), in which the TAK1 pathway inactivates peroxisome proliferator-activated receptor (PPAR) while simultaneously inducing runt-related transcription factor 2 (RUNX2) (33).

In mammalian cells, Wnt1-activated TAK1 forms part of a feedback mechanism to limit activation of the Wnt1 pathway (34). TAK1 also participates in Wnt1-mediated inhibition of the A-myb gene product (A-Myb) (35) and the proteasomal degradation of the c-myb proto-oncogene product (c-Myb) (36), a transcriptional activator that regulates the proliferation and apoptosis of hematopoietic cells.

Given the well-understood mechanism by which TAK1 is activated by K63 polyubiquitin chains in DNA damage and the proinflammatory signaling pathway, it will be interesting to examine whether Wnt signaling is compromised in cells deficient in the machinery involved in the synthesis of these chains.

TAK1 IN THE TGF-β AND BMP SIGNALING PATHWAY

The TGF pathway has wide-ranging effects on cells, including proliferation, differentiation, and survival. It is also involved in a variety of processes including development, inflammation, cardiac hypertrophy, tissue patterning, bone formation, tissue homostasis, and repair.

Binding of TGF-β to its receptor activates Smad-dependent and -independent pathways in which TAK1 functions. It should be remembered that TAK1 was originally identified as a kinase in the TGF-β pathway by complementation and rescue of a yeast MAPKK mutant (37). In mouse mesangial cells and cardiomyocytes, TAK1 is rapidly activated within 5 min by TGF-β and mediates the activation of p38 and JNK (Fig. 3D) (17, 38-40). MEF cells in which TAK1 is deleted have reduced NF-κB and JNK activation by TGF-β (23). Interestingly, the phenotype of a conditional TAK1 knockout resembles that of the ALK I (a TGF-β type 1 receptor) knockout in which vascular development is abnormal (41), providing further proof of an epistatic relationship between TAK1 and TGF-β signaling. More recently, the mechanism by which TAK1 is activated by TGF-β has been uncovered. TRAF6 is activated when its TRAF homology domain at the C-terminus binds type 1 and 2 TGF-β receptors (38, 42), following which TRAF6 ubiquitylates itself and TAK1 on K34 with K63-linked polyubiquitin chains (40, 42). It has also been reported that K158 on TAK1 is ubiquitylated by TRAF6, an event that is critical for TAK1-mediated IKK, JNK, and p38 activation by TGF-β (43).

There has been much interest in the role of TGF-β-activated TAK1 with regard to cardiac remodeling/hypertrophy. In cardiomyocytes, TAK1 can activate calcineurin–nuclear factor of activated T cells (NFAT) through regulator of calcineurin 1 (RCAN1) as part of a TGF-β-induced hypertrophic growth response (44). Calcineurin is activated when its inhibition by RCAN1 is relieved by TAK1 phosphorylation (44). Subsequently, calcineurin dephosphorylates NFAT and allows its translocation to the nucleus (44). Evidence suggests that TGF-β–TAK–p38 may act in a cardiac stress signaling pathway resulting in cardiac hypertrophy, fibrosis, and heart failure (38). A transgenic mouse expressing a dominant negative TAK1 was found to suffer from cardiac hypertrophy by artificially increasing mechanical load and is concomitant with increased secretion of TGF-β (45).

The E3 ligase XIAP previously identified to be a TAB1-interacting protein has also been implicated in TGF-β activation of TAK1 (46). Injection of XIAP (or TAK1) into Xenopus embryos causes the ventralization of cells, thus mimicking TGF-β stimulation (46, 47). The baculovirus IAP repeat (BIR1) domain of XIAP in complex with TAB1 has been solved by X-ray crystallography, and this interaction is required for XIAP induction of TAK1 activation (20). In fact, overexpression of the BIR1 domain of XIAP can lead to NF-κB activation by TAK1 (20). Whether XIAP and TRAF6 have redundant functions in the TGF-β-TAK1 pathway is unclear.

Bone morphogenetic protein (BMP) is a TGF-β superfamily member and has roles in cell growth, differentiation and apoptosis, pattern formation, and skeletal genesis (47). Studies of the BMP–TAK1 pathway are well established with early reports showing that BMP-induced ventralization of Xenopus embryos requires activation of TAK1 (47). In dorsal cells, TAK1 is inactive because of inhibition by BMP antagonists such as noggin. However, heterologous expression of TAK1 in the dorsal cells causes their ventralization, whereas a kinase-dead dominant negative mutant blocked ventralization (47). In the Xenopus embryo, BMP–TAK1 signaling is also required for the induction of epidermal tissue from the ectodermal germ layer while simultaneously repressing the neural fate (48). In the absence of BMP, for example, when antagonized by noggin, the ectodermal layer develops into neural tissue instead (48).

Although the involvement of TAK1 in BMP was demonstrated in Xenopus (47), the first indication that TAK1 is activated by BMP in mammals was a study that showed its involvement in cardiomyocyte differentiation (49). Mice with chondrocyte-specific deletion of TAK1 have shown that BMP–TAK1 is important for the morphogenesis, growth, and maintenance of cartilage and bone formation in which BMP target genes are inactivated (50). The phenotype of these mice resembles that of BMP receptor-1-deficient mice (50).

The identification of direct interaction between the mad homology 2 domain (MH2) of Smads and TAK1 indicates crosstalk between TGF-β-activated Smad and TAK1 pathways (51, 52). In some studies, TAK1 is required along with the Smads for optimal activation of TGF-β target genes through activating transcription factor 2 (ATF2) (39). Others have provided data to show that TAK1 activation downregulates Ski-related novel protein N (SnoN) protein (an inhibitor of TGF-β signaling) to enhance Smad signaling (53). When TAK1 is inactivated, stabilized SnoN prevents the induction of TGF-β-responsive genes (53). Interestingly, TAK1 may mediate the phosphorylation of C-terminal serine residues in Smad1, which are important for its activation. This result is supported by two conditional knockouts of TAK1 in which BMP activation of Smad1/5/8 was impaired (54). In certain cell types, TGF-β can inhibit TNF-α activation of NF-κB by binding of Smad7 to TAB2, which disrupts the formation of TAK1–TAB1–TAB2 complex (55).

TAK1 IN OSMOTIC SHOCK, HYPOXIA, AND OTHER STIMULI

TAK1 is also activated by osmotic shock, for example, by treatment of cells with sorbitol or sodium chloride (21). Intriguingly, osmotic stress does not lead to NF-κB activation even though TAK1 is activated (56). It was shown that in response to osmotic shock, TAO2 prevents TAK1 from activating IKK (Fig. 3E). This shows that TAK1 activation is not always accompanied by IKK activation. Although p38 activation by osmotic stress was unchanged in TAK1−/− keratinocytes and slightly decreased in TAK1−/− MEFs, JNK activation was drastically reduced in both cell types (56). The upstream molecules that mediate the activation of TAK1 in this pathway are currently unknown, and whether ubiquitylation is a mechanism for TAK1 activation in osmotic shock is also unclear. However, TAB1 was found to be necessary for TAK1 activation in osmotic shock (16).

Under hypoxia, TAK1 mediates the activation of NF-κB (Fig. 3F) (57). Interestingly, activation of the hypoxia pathway requires an upstream kinase CAMKII that is reminiscent of the Wnt-5A-stimulated TAK1–NLK pathway (32). As XIAP and Ubc13 have been identified as key molecules in hypoxia-induced NF-κB activation (58), ubiquitylation may play some part in TAK1 activation. Although the requirement for TAB1 and TAB2 in hypoxia activation of the NF-κB pathway by TAK1 was discounted, TAB3 may compensate for TAB2. Another group has also shown TAK1 to be activated by hypoxia albeit by a different mechanism that involves c-Jun-amino-terminal kinase-interacting protein 1 (JIP1) and vaccinia-related kinase 1 (VRK1); however, its physiological significance remains unclear (59).

TAK1 AND AMP-ACTIVATED PROTEIN KINASE

In times of cellular energy deprivation, the energy sensor AMP-activated protein kinase (AMPK) is activated by elevated levels of AMP so that ATP-generating processes such as lipolysis and glycolysis are switched on while simultaneously ATP-depleting processes such as lipogenesis and gluconeogenesis are shut down. Thus, AMPK plays an important role in cellular homogenesis. TAK1 was identified as an upstream kinase of AMPK in a screen using yeast mutants deficient in upstream kinases of Snf1 (the yeast ortholog of mammalian AMPK) (60). In vitro, TAK1 has been shown to directly activate AMPK (11). Around the same time, a study was published in which a transgenic mouse expressing a dominant negative TAK1 had cardiac abnormalities resembling the Wolff-Parkinson-White syndrome, which is caused by mutations in AMPK (61). On first inspection, it is surprising that TAK1 should regulate AMPK as there is no obvious connection between them. However, TAK1 has been shown to be activated by certain stresses such as osmotic shock and hypoxia, which can also activate AMPK. Another study demonstrated that TRAIL can activate AMPK via TAK1 to regulate autophagy, the mechanism by which proteins are broken down so that cellular protein homostasis is maintained (Fig. 3G) (62). Nevertheless, when compared with liver kinase B1 (LKB1) and CaMKII, the major activators of AMPK, TAK1 may only play a much more specialized and minor role in activating AMPK to certain stimuli. Thus, the physiological role of the TAK1–AMPK pathway remains uncertain and outstanding questions remain.

NEGATIVE REGULATION OF TAK1

TAK1 is transiently activated within 20 min in response to IL-1 and LPS (21), and within 5 min by TNF-α (21) and TGF-β (38), following which, TAK1 is downregulated to basal levels as sustained and prolonged activation of the signaling pathway can lead to undesirable outcomes. Several different phosphatases have been identified as regulators of TAK1 under different stimuli and characterized by their ability to dephosphorylate Thr187 and to bind TAK1 (63-66). Two PP2C family members PP2Cβ and PP2ε regulate TAK1 activity in the IL-1 pathway (63, 64). Binding of PP2Cβ to TAK1 is also inducible by TNF-α (64). PP2ε can bind to and suppress TAK1 activity under basal conditions, which is abrogated on stimulation with IL-1 (64). The type 2A phosphatase PP6 also regulates TNF-α- and IL-1-stimulated TAK1 as siRNA knockdown of PP6 enhanced Thr187 phosphorylation (65, 66). Intriguingly, polyubiquitin chains bound to TAB2 act as a scaffold for PP6 to dephosphorylate TAK1, an interaction that is induced by TNF-α (66).

In response to TGF-β, PP2A binds to and negatively regulates TAK1 in mesangial cells (67). The phosphatase calcineurin has also been shown to dephosphorylate Thr187 on TAK1 in response to TGF-β in cardiomyocytes (44). Interestingly, calcineurin binds to TAB2 via RCAN1 in cardiomyocytes on stimulation with TGF-β (44). These data suggest that TAB2 can act as a hub for both activation (as a receptor for ubiquitin) and deactivation (as an interactor of phosphatases) of TAK1.

In addition to the action of phosphatases, TAB1 phosphorylation can negatively modulate TAK1 activity. In response to LPS, TNF-α, IL-1, and sorbitol, p38 can phosphorylate TAB1 on Ser423 and Thr431 to downregulate the activity of TAK1 in a negative feedback loop (3, 12, 21). It should be remembered that common signaling pathways within the cell can be activated by various different stimuli simultaneously. Therefore, it is possible that p38 activated by stimuli other than as a result of TAK1 activation can also regulate TAK1, thus representing an example of crosstalk between different pathways.

Both TAB2 and TAB3 are phosphorylated in response to IL-1, TNF-α, and LPS (3). On IL-1 stimulation, TAB3 is phosphorylated in cells on Ser60 and Thr404 by p38 (18). Interestingly, phosphorylation of these two residues is greatly reduced in TAB1−/− cells, suggesting that TAB1 facilitates the phosphorylation of TAB3 by binding p38 (18). IL-1 also stimulates TAB3 Ser506 by MKK2 or MKK3, whereas TAB2 is phosphorylated on Ser372 and Ser524 by an as yet unidentified protein kinase (18). The functional significance of these phosphorylation sites on TAB2 and TAB3 is unknown. However, it cannot be ruled out that TAK1 activity may be modulated by phosphorylation of TAB2 and TAB3.

Another mechanism by which TAK1 activity can be downregulated is by the cylindromatosis (Cyld) deubiquitinase, which has been demonstrated in Cyld−/− T cells in which TAK1 is constitutively activated (68). In response to TNF-α, a Cyld–itchy homolog of E3 ubiquitin ligase (Itch) complex functions in tandem to terminate TAK1 signaling. Although Cyld cleaves K63 polyubiquitin chains bound to TAK1, Itch tags TAK1 with K48 chains, which is then sent to the proteasome for degradation (69). Thus, in TNF-α-stimulated Itch−/− and Cyld−/− bone marrow-derived macrophages, activation of TAK1 is sustained (69). Both Cyld−/− and Itch−/− mice suffer from chronic inflammatory disorder corresponding to increased cytokine release, which could be mediated by prolonged TAK1 activation (69).

Other deubiquitinases such as ubiquitin-specific peptidase 4 (USP4) can downregulate TNF-α-induced TAK1 activation (70) by cleavage of K63 polyubiquitin chains conjugated to K158 on TAK1. Knockdown of USP4 in HeLa cells enhanced TNF-α-induced TAK1 polyubiquitylation (70). It is unclear whether Cyld and USP4 can target the K63-linked polyubiquitin chains bound to TAB2 or TAB3.

FUTURE PERSPECTIVES

Since the discovery of TAK1 as a critical regulator of innate immunity, TAK1 has been found to be involved in other signaling pathways within the cell. Several ubiquitin ligases have been shown to be direct upstream activators of TAK1 by the synthesis of polyubiquitin chains, which bind to TAB2 and TAB3. There is also evidence to suggest that TAK1 can be activated by linear polyubiquitin chains. However, the role of TAB1, TAB2, and TAB3, and their individual contribution to TAK1 function in the cell remains unclear. The mechanism by which TAK1 is activated by TAB2 and TAB3 may be clarified by crystal structures of TAK1 bound to TAB2 and TAB3. Importantly, as TAK1 is involved in many different signaling pathways, it is essential to examine how specificity is achieved as different stimuli lead to distinct responses in the cell while using common components such as TAK1.

Acknowledgements

Because of strict journal guidelines, we apologize to colleagues whose work has not been adequately cited in this review. The work performed in Cheung's laboratory was supported by ASTAR-BMRC (Grant 07/1/22/19/523) and NTU (Grant RG47/10). The authors have declared that no competing interest exists.