Autoreactive T cells are involved in the development of most autoimmune diseases. Consequently, the induction and maintenance of T cell tolerance to self-antigens is as important to the normal function of the immune system as is the activation of T cells in the presence of pathogens. Despite the enormous effort that has already been made to understand the biochemical and cellular mechanisms that lead to the development of immune tolerance in model systems, we do not yet understand how to re-institute immune self tolerance in individuals that have already developed autoimmune disease. Therefore, a better understanding of the molecular processes involved in this immunological decision-making offers the possibility of defining new therapeutic targets and designing new agents that can better promote a state of immunological self tolerance and more effectively treat clinical autoimmunity.

In this review, we will discuss a novel form of T cell regulation that involves a post-translational modification of proteins by ubiquitination. This system of protein ubiquitination plays a key role in many cellular processes, such as the regulation of the cell cycle, modulation of cell surface receptors, cellular differentiation, DNA repair, gene transcription, and cellular stress responses. In the innate immune system, ubiquitin-dependent proteasomal degradation of foreign proteins mediates antigen presentation. Furthermore, the activation of the proinflammatory cytokine gene transactivator nuclear factor κB (NFκB) relies on ubiquitin-mediated degradation of the IκBα inhibitory protein at sites of infection and/or inflammation. Recently, protein ubiquitination has been shown to mediate several important molecular functions in T cells that are linked to the avoidance or development of autoimmunity. Below, enzymes important to the regulation of protein ubiquitination in T cells will be described and their roles as negative regulators of autoimmunity will be considered in more detail.

Ubiquitin is a highly conserved 76 amino acid globular protein that is attached to substrate proteins to modify a variety of cellular processes. Although first described as a mechanism for proteolysis of misfolded or damaged proteins, ubiquitination is now appreciated as an important modification for cellular trafficking and transcriptional activation, as well as for proteasomal- and lysosomal-mediated degradation of signaling intermediates in the regulation of normal cell function. Ubiquitination is accomplished through a series of enzymatic steps involving a ubiquitin-activating enzyme (called E1), a ubiquitin-conjugating enzyme (E2), and a ubiquitin ligase (E3), resulting in the transfer of covalently bound ubiquitin from the E2 protein to a lysine residue on the target protein 1. While mammals have only one confirmed E1, there are over 30 E2 enzymes and many more E3 ligases, and this allows for the ubiquitination system to confer substrate specificity on the many cellular processes controlled by ubiquitin modification 2.

Different patterns of covalent attachment of ubiquitin to target proteins provide a further level of specificity to regulation of cellular processes by ubiquitination. E3 ligases may attach one or more ubiquitin polypeptides to lysine residues of the target protein in order to direct degradation, transport, or function. For recognition and degradation by the 26S proteasome, substrates are polyubiquitinated in that four or more ubiquitins form a chain by ligating the carboxyl terminus of free ubiquitin to Lys48 of the previously attached ubiquitin protein 3. In contrast, monoubiquitination of target protein lysine residues results in altered trafficking to the endosome or lysosome 4. Substrate proteins may also be multiubiquitinated with the ubiquitin chains ligating the Lys63 or Lys39 residue of the previously attached ubiquitin, resulting in altered transport or function of the target protein 5.

All E3 ligases are functionally similar in that they contain a domain for recognition and binding of the E2 ubiquitin conjugating enzyme, a catalytic domain for the transfer of ubiquitin from the E2 to the target protein, and one or more protein-protein interaction domains for substrate recognition and binding (Fig. 1) 2. E3 ligases, however, may be either multi-protein complexes or single proteins containing all of these functional domains. There are three types of E3 ligases known to function in the immune system: the really interesting new gene (RING) proteins, homologous to E6-associated protein carboxyl terminus (HECT), and U-box. These enzymes can act either to enhance immunity or dampen T cell responses.

The single chain RING and HECT E3 ubiquitin ligases perform a similar role as the multi-chain E3s, but all of the functional domains are contained within a single polypeptide (Fig. 1b). The catalytic RING domain transfers ubiquitin from the E2 ubiquitin-conjugating enzyme directly to the target protein, whereas HECT proteins themselves accept the ubiquitin polypeptide prior to its transfer to a target protein. Specificity is achieved through the recognition of target substrates via protein-protein interaction domains.

Clonal anergy has been postulated to be one important immune tolerance mechanism that relies on the inducement of mature T cells into an unresponsive state following their initial exposure to a peripheral self-antigen 69. This outcome differs greatly from that seen during a protective immune response. For the case of T cells responding to dangerous pathogens, continued antigen responsiveness is ensured because antigen presentation is restricted to dendritic cells that have detected the presence of the pathogen and its associated toll-like receptor ligands. Consequently, antigen presentation is accompanied by the surface expression of a high level of 'costimulatory' ligands such as CD80 and CD86 on the dendritic cells. CD80 and CD86 specifically bind to the CD28 costimulatory receptor within the immunological synapse that forms between the T cell and the antigen-presenting cell during antigen recognition. The end result of this strong costimulatory interaction is a maintenance of the high level of antigen sensitivity that is required to clear the pathogen.

In contrast to antigen recognition during infection, the delivery of a strong TCR signal as a consequence of self-antigen recognition is normally unaccompanied by sufficient co-stimulatory signaling to maintain a high level of antigen responsiveness 70. This development of clonal anergy results in an inability of these cells to efficiently produce the autocrine growth factor IL-2 and to proliferate upon re-exposure to antigen. Unresponsiveness is actively induced by an increase in intracellular Ca²⁺, and new proteins must be made in order to establish the anergic state 7172. We have also demonstrated that a fusion of anergic T cells to normal cells fails to restore antigen responsiveness, indicating the presence of dominant-acting repressor molecules within anergic T cells that inhibit signal transduction to the Il2 gene 73. Macian et al. 74 reported that Ca²⁺ signaling using the calcium ionophore ionomycin could induce a limited set of anergy-associated genes in a NF-AT dependent manner to render T cells tolerant of antigen. Some of these genes appear to be involved in protein ubiquitination and, consequently, there has recently been great interest in the roles of E3 ubiquitin ligases as anergy factors.

In normal resting T cells, the protein levels of Cbl-b, GRAIL, and Itch are relatively low, and these E3 ligases normally appear not to interfere with signaling cascades leading to IL-2 secretion and proliferation when costimulatory signals are abundant. Within anergic T cells, however, E3 ligase expression is increased and/or E3 enzymes are directed to unique cellular compartments during antigen stimulation. It appears they then cooperate in the ubiquitination of tyrosine-phosphorylated proteins that leads to their degradation in lysosomes.

The exact mechanism by which these E3 ubiquitin ligases maintain antigen unresponsiveness in anergic T cells remains uncertain. Immediately after clonal anergy induction, T cells demonstrate global defects in TCR signaling, including reduced phosphorylation of TCR ζ and ε chains, poor activation of p56^Lck, Zap70, Ras, JNK and ERK, and defective transactivation at the Il2 gene by NFκB, activating protein 1 (AP-1), and NF-AT 70. Following antigen re-stimulation, anergic T cells also demonstrate an aberrant down-regulation of phosphorylated PLCγ1, PKCθ, and RasGAP 78. Remarkably, the activation of Itch^-/- and Cblb^-/- T cells fails to induce a degradation of these signaling molecules even after an anergy-inducing regimen 78. Itch and its HECT family relative NEDD4 have also been observed to translocate into a detergent-resistant membrane fraction following their stimulation of anergic T cells 78. Itch can mono-ubiquitinate PLCγ1, promoting its degradation within an endocytic compartment 78. Taken together, these findings suggest a model in which the E3 ligases GRAIL, Cbl-b, Itch, and NEDD4 ubiquitinate and chaperone critical proximal signaling molecules into an endocytic pathway and direct them away from the immunological synapse and into a lysosomal compartment where they are subject to degradation.

Another plausible substrate for the Itch E3 ligase activity in anergic cells is the AP-1 component molecule JunB. Like GRAIL, Itch localizes to an endocytic pathway during T cell stimulation. Itch appears to specifically recognize JunB, leading to its ubiquitination and degradation 80. Consistent with this, Itch and ubiquitinated JunB have been co-localized within a lysosomal compartment following stimulation 6781. Itch^-/- T cells do, in fact, have a slower rate of JunB turnover, and higher JunB DNA-binding activity 80. In anergic T cells, dysregulated Ras function and deficient activation of the mitogen-activated protein kinases ERK, JNK, and p38, can be expected to result in only a limited induction of JunB protein during antigen stimulation 828384. Therefore, a combination of defective JunB gene transcription and enhanced JunB protein turnover ultimately leads to a deficiency of AP-1-dependent transactivation at the Il2 gene. Interestingly, JNK has been shown to enhance the degradation of JunB through a phosphorylation-dependent activation of Itch itself 68. Whether the defect in JNK activation that exists in anergic T cells tempers the ability of Itch to ubiquitinate JunB and promote its premature degradation in the lysosome remains unknown at this time.

By working cooperatively or sequentially, these E3 ligases appear to target activated signaling complexes in anergic T cells and disrupt the nascent immunological synapse and inhibit the ongoing TCR signaling cascade. Premature turnover of Jun family nuclear factors would also put a brake on TCR signaling and prematurely abort the IL-2 production and proliferative responses of anergic T cells (Fig. 2).

The induction of autoimmunity is a complicated process that generally involves the breaching of multiple checkpoints 85. Nevertheless, the absence of a single E3 ligase activity can in some cases lead to the spontaneous development of autoimmune disease, perhaps via a loss of T cell tolerance to self antigens. Mice lacking Cbl-b are characterized by the production of autoantibodies, infiltration of activated T and B lymphocytes into multiple organs, and resultant parenchymal damage 38. Furthermore, the resistance of Cbl-b-deficient mice to anergy induction during chronic and repeated exposure to antigen puts them at risk for high mortality due to toxic T cell activation 77. The absence of Cbl-b also allows for the development of a destructive autoimmune arthritis that can be induced with type II collagen even in the absence of mycobacterial adjuvants 77. Similarly, Cblb^-/- mutant mice are highly susceptible to the induction of experimental autoimmune encephalomyelitis, a mouse model of multiple sclerosis 43. A Cblb locus point mutation, which leads to the expression of a truncated form of the Cbl-b protein lacking E3 ligase activity, has been detected in Komeda diabetes-prone rats 86. In one human study of patients with type I diabetes plus a second autoimmune disease, a CBLB exon 12 single nucleotide polymorphism was also shown to be significantly associated with disease occurrence 87.

Itchy mice demonstrate diverse immune disorders, including chronic inflammation of the pulmonary interstitia and alveolar proteinosis, inflammation of the glandular stomach tissue, as well as skin inflammation resulting in scarring due to constant itching. These mice also exhibit severe lymphoid hyperplasia and die at age 6 to 8 months 7980. Itch does not appear to be involved in T cell development in the thymus, but Itch^-/- T cells become chronically activated as the mouse ages 80. Similar to Cbl-b, Itch^-/- T cells show resistance to clonal anergy induction 78. No human autoimmune disease has yet been linked to the ITCH locus.

In mice, the homozygous genomic disruption of Socs1 is lethal. However, Socs1^+/- female mice, as well as Socs1^-/- mice made transgenic for a low level of SOCS1 in the lymphoid compartment using a Eμ promoter, survive into adulthood but develop a lupus-like syndrome, including the expression of double-stranded DNA antibodies and immune-complex glomerulonephritis 88. In these animals, CD4⁺ T cells showed enhanced proliferative responses to IL-2. CD4^-/- Socs1^-/- double knockout mice lacking CD4⁺ T cells were protected from autoimmunity. Thus, SOCS1 function in CD4⁺ T cells may prove to facilitate an induction of anergy in response to self-antigen recognition.

Other E3 ligases have been genetically linked to autoimmune disease. A mutation in the autoimmune regulator (AIRE) gene is responsible for the development of autoimmune-polyendocrinopathy-candidiasis-ectodermal dystrophy (APECED), an organ-specific autoimmune disease with autosomal recessive inheritance 8990. Recently, AIRE protein was identified as an E3 ligase and APECED disease-causing mutations abolish its ubiquitin ligase activity 91. Significant association of rheumatoid arthritis has also been observed with an intron 3 single nucleotide polymorphism from the CUL1 gene. CUL1 is important to proliferation and for the induction of IL-8 secretion during T cell activation 92. Interestingly, the intron 3 sequence polymorphism found to be associated with rheumatoid arthritis demonstrates a greater DNA enhancer activity than an intron 3 sequence having no association with rheumatoid arthritis, perhaps suggesting that increased expression of this E3 ligase can contribute to the excessive T cell activation and loss of tolerance observed in this autoimmune disease.

In summary, these data indicate that the aberrant expression or function of any one of several E3 ubiquitin ligases is sufficient to initiate or prolong a T cell response that is directed against a self-antigen. As regulators of T cell activation, E3 ubiquitin ligases normally set an appropriate threshold for T cell activation to allow for a protective immune response against pathogens while preventing the onset of clinically important autoimmune disease. Dysregulation of one or more of these ubiquitination pathways in the human immune system, therefore, may pose the risk of a loss of immune self-tolerance.

AP-1 = activating protein 1; APECED = autoimmune-polyendocrinopathy-candidiasis-ectodermal dystrophy; CBC = Cullin-Elongin BC-SOCS/VHL; CHIP = carboxyl terminus of Hsc70-interacting protein; E1= ubiquitin-activating enzyme; E2 = ubiquitin-conjugating enzyme; E3 = ubiquitin ligase; GRAIL = gene related to anergy in lymphocytes; HECT = homologous to E6-associated protein carboxyl terminus; HIF = hypoxia inducible factor; IFN = interferon; IL = interleukin; NFκB = nuclear factor κB; NF-AT = nuclear factor of activated T cells; PA = protease-associated; PI3K = phosphoinositide 3-kinase; PLC = phospholipase C; RING = really interesting new gene; SCF = Skp1-Cullin-F-box; SOCS = suppressor of cytokine signaling; STAT = signal transducer of activated T cells; TCR = T cell receptor; TGF = transforming growth factor; Treg = T-regulatory; VHL = von Hippel-Lindau.

The author(s) declare that they have no competing interests.

JLB and RZ contributed equally to this manuscript.