Perspective

TRAF1: Lord Without A RING

See allHide authors and affiliations

Science's STKE  21 May 2002:
Vol. 2002, Issue 133, pp. pe27
DOI: 10.1126/stke.2002.133.pe27

Abstract

Tumor necrosis factor (TNF) receptor-associated factors (TRAFs) constitute a family of adaptor proteins that associate with the cytosolic tail of particular TNF-family receptors (TNFR) and regulate cytokine signaling by linking TNFRs with downstream protein kinases, ubiquitin ligases, and other effector proteins. A total of six members of this family (TRAF1-6) have been identified in mammals. TRAF1 is unique among TRAFs because it lacks a RING finger domain present in TRAF2-6 that has been shown to be required for TRAF2- and TRAF6-mediated activities. TRAF1 also has the most restricted expression among TRAFs, and is found almost exclusively in activated lymphocytes, dendritic cells, and certain epithelia. Recent evidence obtained from TRAF1-/- mice shows that TRAF1-deficient T cells are hyper-responsive to TNF-α, having increased T cell receptor (TCR)-dependent T cell proliferation rates in vitro. Also, these TRAF1-/- mice had increased sensitivity to TNF-α-induced skin necrosis in vivo. These results support a role for TRAF1 as a negative regulator of signaling by certain TNF-family receptors. This review summarizes current knowledge about TRAF1, focusing on the new information provided by these TRAF1-deficient mice. Also, the pros and cons of TRAFs as potential targets for drug discovery are discussed.

Tumor necrosis factor (TNF) receptor-associated factors (TRAFs) represent a group of structurally similar adaptor proteins sharing a conserved COOH-terminal region known as the TRAF domain (TD) (Fig. 1). TRAF proteins were originally identified by their ability to interact through their TDs with the cytosolic domains of members of the TNF receptor (TNFR) superfamily (1). TRAFs function as molecular bridges, linking the cytosolic portions of these receptors to downstream protein kinases, ubiquitin ligases, and other adaptor proteins (2-4). Six TRAF family proteins (TRAF1-6) have been described in humans and mice, and all appear to mediate signal transduction by one or more members of the TNFR family TRAF6 also regulates signaling from Toll-interleukin-1 receptor (IL-1R) (5), with orthologous proteins in Drosophila probably playing similar roles in innate immunity by activating NF-κB (6, 7). Four additional proteins encompassing less conserved, more primitive TDs have also been described in humans (Fig. 1) (8), but no evidence supporting their involvement in regulating TNFR signaling has emerged so far.

Fig. 1.

Schematic representation of the members of the TRAF family. A total of six members of the TRAF family (TRAF1-6) have been described in mammals. All share a COOH-terminal fold called the TRAF domain (TD). The TD is a fold consisting of eight anti-parallel β sheets (also denominated TRAF-C domain), which is associated to a NH2 terminal α-helix (TRAF-N domain) in all the members of the TRAF family (38, 39, 41). TRAF2-6 also have RING finger domains and each contains variable numbers of zinc-finger domains in tandem. TRAF3, TRAF3B, and TRAF5 also have a coiled-coil before the TRAF-N domain. TRAF2A is a TRAF2 isoform that contains an insertion of seven amino acids in the RING finger domain, and it has been described as a dominant inhibitor of TRAF2 (42). TRAF3B is an alternative splicing form of TRAF3 with only three zinc-finger domains (43). Four TRAF-encompassing factors (TEF) have been identified in humans (8). TEFs have an ancient TD composed only by the eight anti-parallel β-sheets (TRAF-C domain) that is located either at the NH2-terminus (TEF1, 2 and 4) or in the middle of the molecule (TEF3). TEF1, also known as USP7, is a nuclear ubiquitin-specific protease. TEF2 (SPOP) is a nuclear protein of unknown function that also contains a POZ (poxvirus and zinc finger) domain. TEF3 (also known as MUL or TRIM37) is the product of the causative gene of mulibrey nanism, a rare autosomal recessive disorder affecting muscle, liver, brain, and eye. TEF3 contains an NH2-terminal RING finger domain, followed by a zinc finger B box and a coiled-coil, thus constituting an RBCC or tripartite domain. The TD is found after the RBCC domain and is followed by two polyacidic regions. TEF4 is similar to TEF2 and also contains a C-terminal POZ domain. Its function is still unknown. Meprins are a class of dimeric extracellular metalloproteinases of the astacin family that also contain putative TDs of unknown function (44) (not shown). Online Mendelian Inheritance in Man (OMIM) accession numbers: TRAF1 601711; TRAF2 601895; TRAF3 601896; TRAF4 602464; TRAF5 602356; TRAF6 602355; TEF1 602519; TEF2 602650; TEF3 605073.

TRAF1 is unique among TRAFs because it lacks the "really interesting new gene" (RING) domain found near the NH2-terminal regions of TRAF2-6. Deletion of the RING domain, which contains a Zn2+-binding fold, from TRAF2, TRAF5, or TRAF6, revealed that the RING domain is critical for the activation of NF-κB or Jun NH2-terminal kinase (JNK) (2). Moreover, RING-deleted (ΔRING) mutants of these TRAFs typically inhibit signal transduction by members of the TNFR family. Given the absence of a RING domain in TRAF1, it may not be surprising that a recent paper from Tsitsikov and co-workers has provided in vivo evidence of a negative regulatory role for TRAF1 in mice by demonstrating enhanced TNF signaling in TRAF1-deficient mice (9). However, when compared with other observations concerning TRAF1, one wonders whether there might be more to the story of this unique TRAF family member.

The expression of TRAF1 is the most restricted among TRAFs; TRAF1 is found almost exclusively in activated lymphocytes, dendritic cells, and certain epithelia (10). TRAF1 has been reported to associate directly or indirectly with multiple TNFR family members. Though results are sometimes based on over-expression experiments, TRAF1 reportedly also binds several intracellular proteins, including adaptor proteins such as TNFR-associated death domain protein (TRADD), TRAF-associated and NF-κB activator (TANK, also known as I-TRAF), and TRAF-interacting protein (TRIP); protein kinases such as NF-κB-inducing kinase (NIK), receptor-interacting protein (RIP) and RIP2 (Cardiak); the NF-κB inhibitory protein A20; and the apoptosis-suppressors inhibitor of apoptosis 1 (cIAP1), cIAP2, and FADD-like interleukin-1β converting enzyme (FLICE)-like inhibitory protein (FLIP) (2). Additionally, TRAF1 forms heteromers with TRAF2. TRAF1 can also be recruited to the TNFR1 and TNFR2 through its interaction with TRADD and TRAF2, respectively, where it is also found associated with cIAP1 and cIAP2 (Fig. 2).

Fig. 2.

Multifaceted functions of TRAF1. Possible mechanisms by which TRAF1 might regulate signaling by TNFR2 (left) and TNFR1 (right) are depicted. TNFR2-mediated activation of NF-κB and AP1 transcription factors depends on the association of TRAF2 trimers with the receptor, and the subsequent TRAF2-dependent recruitment to the complex of kinases that activate a signaling cascade resulting in NF-κB and AP1 activation. The RING finger domain of TRAF2 is required for the activation of these kinases. TRAF1 can also associate with TNFR2 through its interaction with TRAF2. TRAF1-TRAF2 heteromers inhibit TNFR2 activities, probably by decreasing TRAF2 binding to the receptor or by interfering with the recruitment or activation of the kinases (red, dashed lines). TRAF1 expression is induced after NF-κB activation, and therefore TRAF1 may participate in a feedback loop to suppress TNFR2-induced NF-κB activity (black, dashed lines). Expression of the caspase inhibitors cIAP2 (and also probably cIAP1) and FLIP, is also NF-κB-dependent. cIAP1 and cIAP2 can also interact with TRAF1 and TRAF2, and through this interaction they can also be recruited to the TNFR2 complex. With respect to TNFR1 signaling, TRAF1 and TRAF2 can associate with TNFR1 through their interaction with TRADD, which directly binds to the cytosolic tail of this receptor. TRAF1 and TRAF2 can also recruit cIAP1 and cIAP2 to the TNFR1 complex. FADD interacts with TRADD and recruits caspase-8. The balance between proapoptotic (FADD/caspase-8) and antiapoptotic (TRAF1/TRAF2/cIAP1/cIAP2) complexes that are recruited to TNFR1 might determine the signal transduction pathways activated through TNFR1, inducing either gene expression or apoptosis. Furthermore, TRAF1 can be cleaved by caspase-8, producing two fragments. The COOH-terminal TRAF1 fragment potentiates receptor-mediated apoptosis by inhibiting receptor-mediated NF-κB activation and possibly by interfering with the recruitment of antiapoptotic complexes to the receptor.

This plethora of interacting partners positions TRAF1 for multiple possible functions and suggests that absence of this protein should produce several defects in cytokine signal transduction networks. Surprisingly, however, Tsitsikov and co-workers find that TRAF1 knockout mice are phenotypically normal in appearance. TRAF1-/- mice have no apparent defects in development of either T cells or B cells, and have normal numbers of lymphocytes in peripheral lymphoid organs, unimpaired antibody responses to both T-dependent and T-independent antigens, and typical kinetics of lymphocyte deletion and expansion in response to superantigen (9). Only the inguinal lymph nodes of TRAF1-deficient mice showed significant differences compared to wild-type littermates, having a larger number of lymphocytes and an increased T/B cell ratio. These results stand in sharp contrast to the those of gene knockouts for TRAF2, TRAF3, TRAF4, and TRAF6; striking developmental defects in these mice are evident (2).

Although no clear difference in lymphocyte proliferation rates were observed in lymphatic tissues, where the total numbers of lymphocytes in TRAF1-/- and wild-type mice were similar, lymphocytes cultured in vitro from TRAF1-deficient mice exhibited a hyperproliferative response when stimulated through the T cell antigen receptor (TCR) complex with antibodies to CD3 (9). Moreover, experiments using blocking antibodies suggested that this enhanced proliferation depended on TNFR2--one of the two cellular receptors for TNF-α with which TRAF1 associates (1) and whose expression in induced in T cells after TCR activation (9, 11). Also, adding TNF-α to cultures of T cells stimulated with antibodies to CD3 enhanced the proliferation of TRAF1-deficient but not wild-type T cells. Consistent with these data, Tsitsikov et al. (9) also showed that TNF-α treatment results in increased amounts of NF-κB and JNK activation [with a concomitant increase in activating protein 1 (AP1) activity] in the TRAF1-deficient T cells. Thus, the observations of Tsitsikov et al. (9) suggest that TRAF1 interferes with TNFR2-mediated events that promote CD3-depedent T cell proliferation.

At first glance, these results from the TRAF1-/- mice seem at odds with data from TRAF1 transgenics that overexpress this protein in T lymphocytes. Speiser and co-workers (12), for example, previously showed that CD3-stimulated proliferation is unchanged in T cells isolated from TRAF1 transgenic mice compared to control littermates, suggesting that overproduction of TRAF1 does not alter CD3-mediated T cell proliferation. However, since TRAF1 expression is strongly induced in T cells in response to TCR activation (10, 13), it is likely that endogenous TRAF1 is sufficient to inhibit TNF-α-mediated TNFR2 co-stimulatory responses in normal activated T cells, thus explaining why TRAF1 transgene overexpression has no effect on CD3-mediated proliferation (12). Interestingly, these TRAF1 transgenic mice (12) also have normal T cell populations, thus strongly arguing against a role for TRAF1 in T cell homeostasis or development.

TRAF1 is not expressed in resting lymphocytes, but rather is induced by mitogenic stimuli (10, 14). In this regard, the promoter region of TRAF1 contains several NF-κB binding elements, and TRAF1 gene expression is greatly increased by activated NF-κB (15). Thus, the RING-less TRAF1 protein may participate in a negative feedback loop that squelches signaling by TNFR2 and might prevent excessive NF-κB activation or ensure that NF-κB activation occurs only in short bursts. How TRAF1 accomplishes this is uncertain. By forming heterocomplexes with TRAF2, TRAF1 might decrease TRAF2 binding to the cytosolic tail of TNFR2. Alternatively, or in addition, TRAF1 might sequester downstream kinases such as NIK, which binds the TDs of the TRAFs but depends on the RING for activation (16). These observations therefore suggest a scenario whereby T cell expansion could be controlled by altering the TRAF1:TRAF2 ratio, with TRAF2 representing the "gas pedal" and TRAF1 the "brakes." Maintaining proper T cell activity might depend on properly balancing these antagonistic molecules in time and space (subcellular location); unbalanced TRAF1:TRAF2 ratios could misdirect lymphocytes toward the promotion of autoimmunity or malignancy.

With regard to proper in vivo ratios of TRAF proteins, Tsitsikov et al. took their observations from cell culture experiments (where addition of exogenous TNF-α enhanced T cell proliferation) and extended them in vivo by injecting TNF-α into the skin of wild-type and TRAF1-null mice (9). In both normal and TRAF1 knockout mice, injection of high doses of intradermal TNF-α induced skin necrosis. In contrast, low-dose TNF-α induced skin necrosis in TRAF1-/- mice but not wild-type littermate controls, revealing that TRAF1-/- mice were hypersensitive to TNF-α. Experiments with Recombinase Activating Gene (RAG)-2-deficient mice that lack lymphocytes revealed that TNF-α-induced skin necrosis is a lymphocyte-dependent response, permitting speculation that TRAF1-deficient mice exhibit hypersensitivity to TNFα because their lymphocytes are hyperresponsive to this cytokine. In this regard, the co-stimulatory role of TNFR2 enhancing the proliferative response of T cells to CD3 activation is well documented (17-19). Accordingly, TNFR2-/- T cells display a marked reduction in their proliferative response to TCR-agonists (20). These results therefore are in concordance with the opposite, hyperresponsiveness to TNFα observed in TRAF1-/- T cells. However, as the authors pointed out, alternative explanations are possible for the TNFα hypersensitivity phenotype seen in the skin of TRAF1-/- mice. For example, TRAF1 is constitutively expressed in epidermal keratinocytes (10); therefore, its absence might sensitize keratinocytes to TNFα-induced cell death.

Indeed, a role for TRAF1 in suppressing TNFα-induced apoptosis mediated by the p55-60 TNFR1 has been suggested previously. TNFR1 recruits TRAF1 to its cytosolic domain through the adaptor protein TRADD. TRADD can interact with two types of proteins through different domains. In one interaction, TRADD binds Fas-associated death domain (FADD), a cysteine aspartase (caspase)-8-binding protein, at its death domain (DD). In the other interaction, TRADD binds TRAF1 and TRAF2 using an NH2-terminal domain (21). TNFR1 recruits a TRADD-dependent complex containing TRAF1, TRAF2, cIAP1 and cIAP2, which, it has been argued, can suppress TNF-α-dependent caspase-8 activation, thereby preventing apoptosis (13, 22) (Fig. 2). Alternatively, recruitment of TRAF1 to some TNFR family members might suppress apoptosis indirectly by enhancing (rather than suppressing) NF-κB activation (23, 24), whereby NF-κB could then directly induce the transcription of several antiapoptotic proteins including cIAP2 (and possibly other IAPs), FLIP, and Bcl-2 family members Bcl-XL and Bfl-1 (22, 25-27). Furthermore, compelling evidence that TRAF1 can suppress apoptosis has come from experiments involving transgenic mice that overexpress TRAF1 in their T cells, showing that TCR-mediated apoptosis is reduced (12). Several reports indicate that apoptosis in activated T cells is regulated by Fas and either TNFR1 or 2 (11, 28). Interestingly, recent evidence shows that TRAF1 can be converted into a proapoptotic protein after cleavage by caspase-8 in the setting of Fas-ligand or TNF-α induction of apoptosis (29-31). Caspase-8 cleaves TRAF1, producing two fragments. Overexpression of the COOH-terminal TRAF1 fragment enhances TNFR1- and Fas-mediated apoptosis (29, 30). On the other hand, apoptosis induced by chemotherapeutic drugs or UV-radiation does not cause TRAF1 cleavage (29), demonstrating specificity for apoptosis pathways that activate caspase-8. Since association of TRAF1 with TRAF2, IAP1 and IAP2 have been claimed to prevent caspase-8 activation in TNF-stimulated cells (13, 22); one wonders whether overexpression of TRAF1 in transgenic mice ensures an adequate reserve of full-length TRAF1 molecules for preventing caspase-8 activation.

Taken together, these findings beg the question: Why is apoptosis of lymphocytes apparently unperturbed in TRAF1-null mice? The percentage of apoptotic cells observed in lymphoid organs is not appreciably altered in these TRAF1-/- mice, and lymphocytes cultured from TRAF1-deficient animals exhibit no propensity for apoptosis in vitro (9). Given that TNF-α induces greater amounts of NF-κB activation in TRAF1-deficient T cells, one would expect the various NF-κB-inducible antiapoptotic genes to reduce apoptosis. On the other hand, TRAF1-deficiency was also associated with greater JNK activation. Because the JNK family kinases are generally associated with the promotion of apoptosis in response to stress (32), perhaps increased JNK activity functions to counterbalance the contributions of excessive NF-κB to apoptosis regulation. Further, why are the overall numbers of T cells in TRAF1-deficient mice normal? If proliferation is enhanced (as is observed in vitro), then maybe apoptosis is also increased so that there is no net gain in the number of T cells. Curiously, TRAF1 transgenic mice also have normal percentages of T cell subsets, suggesting that neither overexpression nor loss of TRAF1 is sufficient to upset the normal homeostatic mechanisms that dictate the total numbers of T lymphocytes in the body (12).

How do we reconcile these disparate observations? Differences in cell context provide at least one convenient, although unsatisfying, excuse. Like the proverbial blind men feeling the elephant, our knowledge is only as complete as the experiments we choose to perform. For instance, in the case of TRAF2 knockout mice, initial reports revealed that TRAF2 is expendable for TNF-α-mediated activation of NF-κB, yet critical for JNK activation (33). More recently, it was reported that B cells from TRAF2-null mice are defective in NF-κB activation in response to CD40 (34), demonstrating that TRAF2 is required for NF-κB in some cellular contexts. Similarly, TRAF2 might dictate whether TNF-α is involved in apoptotic signaling in activated T cells, depending on whether RIP is present or absent (35). When present, RIP binds TRAF2 on TNFR2 and recruits TRADD, which binds FADD, which in turn triggers apoptosis by activating caspase-8. If RIP is absent, then TRAF2 activates NF-κB and apoptosis is avoided (35). One wonders therefore whether TRAF1, like the infamous Dr. Jekyll and Mr. Hyde, can flip-flop its phenotype depending on what proteins it binds or perhaps on whether TRAF1 is intact versus cleaved by caspases. The production of cleavage-resistant TRAF1 knock-in mice could be used to address this question. Another confounding variable concerns the issue of which of the many TRAF-binding TNF-family receptors is engaged by their ligands at any particular moment in a given cell. It could be that TRAF1's interaction with some TNF-family receptors is stimulatory (23), whereas its binding to others is inhibitory (9, 36). Given that TRAF1 associates directly or indirectly with TNFR1, TNFR2, CD30, CD27, 4-1BB (a member of the TNFR superfamily, also termed CD137), TNF-related activation-induced cytokine receptor (TRANCE-R), herpes virus entry mediator (HVEM), CD40, OX-40 (also termed CD134), the Epstein-Barr virus latent membrane protein 1 (LMP1), B cell maturation antigen (BCMA), and activation-inducible TNFR (AITR) (2), the combinatorial possibilities are rather daunting.

We are left wondering how much we have yet to learn about TRAF1. For example, although T cells from TRAF1-null mice exhibit enhanced proliferative responses in vitro, their B cells have normal proliferation rates upon IgM or CD40 binding, and CD40-mediated activation of NF-κB and AP1 is unaltered (9). Furthermore, both T-dependent and T-independent antibody responses are intact in the TRAF1 knockout mice, indicating that antibody production, Ig isotype switching, and T cell helper functions do not require TRAF1. Nevertheless, TRAF1 mRNA and protein levels are typically increased in B cell malignancies, including non-Hodgkin lymphomas (NHL) and chronic lymphocytic leukemias (CLL), as well as the Reed-Sternberg cells of Hodgkin disease--especially Epstein-Barr virus (EBV)-positive cases of Hodgkin disease in which the TRAF-binding viral protein LMP1 is found (10, 37). What is the role of TRAF1 in neoplastic B cells? Could it be that TRAF1 sequesters TRADD, RIP, or other proapoptotic proteins, thus tipping the apoptosis-antiapoptosis balance in favor of cell survival? Perhaps mating the TRAF1 knockouts with oncogene-expressing transgenic mice that develop B-cell malignancies will shed some light on these questions.

The observations derived from the recently described TRAF1 knockout mice have important implications where drug discovery efforts are concerned. The solving of three-dimensional structures of TDs bound to their docking sites on members of the TNFR family and TRADD has generated interest in developing small molecule drugs that compete with 3 to 5 amino-acid peptidyl motifs found in TNFRs for binding to the surface of TDs (21, 38-40). These structural studies and experiments with peptide libraries have revealed a core motif involved in binding to several TDs that might be exploited for making TRAF inhibitors. Sufficient sequence differences between TDs exist, suggesting that the generation of specific TRAF inhibitors might be possible (21, 39, 40). The complexity of the biological roles of TRAF-family proteins however, suggests that it might not be easy to predict which diseases would be most appropriate for applying either broad-spectrum or selective TRAF inhibitors. For example, if TRAF1 is an endogenous inhibitor of TNFR2 signaling that operates in a feedback loop that squelches NF-κB activation, then inhibiting its binding to this TNFR family member is not something one would want to do in patients with inflammatory or autoimmune diseases. Targeting other TRAFs, such as TRAF2, could also be problematic. In this regard, TRAF2-deficient macrophages produce increased amounts of nitric oxide (NO) and TNF-α in response to TNF-α stimulation (34). Thus, although the results obtained from TRAF2-null mice predict benefits based on the requirement of TRAF2 in CD40-mediated B cell proliferation and T-dependent antibody production, enhancing the proinflammatory effects of macrophages is not a desired outcome. Altogether, the collective results suggest multiple positive and negative roles for TRAFs, which differ in highly cell context-dependent manners. Empirical testing of selective and broad-spectrum TRAF inhibitors in animal models is needed to define whether TRAFs represent valid drug-discovery targets.

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.
View Abstract

Stay Connected to Science Signaling

Navigate This Article