3.1. Acetylation and deacetylation of FOXP3

FOXP3 is known to be acetylated by KATs. KATs can add acetyl groups to the lysine of their target proteins. Among KATs, Tip60 (KAT5) and cyclic adenosine monophosphate response element-binding protein (CREB)-binding protein (CBP)/p300 (KAT3A/3B) directly bind to FOXP3 and acetylate it, which enhances DNA binding and stability [38], [50], [51].

Since acetylation sites in FOXP3 overlap with ubiquitination sites, acetylation by KATs prevents ubiquitin-dependent proteasomal degradation (as described below) and induces the stability of FOXP3 protein.

We have found that Tip60 and p300 acetylate each other and further induce FOXP3 acetylation [37]. A structure model suggests that the single acetylation of Tip60 at K327 by p300 would favor FOXP3 binding to the Tip60 cleft. Genetic deletion of Tip60 in Tregs demonstrated severe autoimmune problems similar to those of scurfy mice, suggesting that Tip60 is essential for Treg function.

We have also reported that the natural p300 inhibitor garcinol induces FOXP3 degradation via a lysosome-dependent pathway, which suggests that p300 plays a role in lysosome-dependent FOXP3 degradation [52]. However, single deletion of p300 [38] or CBP [39] only causes relatively mild autoimmunity. Since CBP and p300 have high homology and are functionally conserved [53], single deletion is not effective to limit p300 or CBP function. This has been confirmed by double deletion of p300 and CBP, which causes severe autoimmunity on a level similar to the deletion of Tip60 [39].

Lysine deacetylase, which is sometimes known as histone deacetylase (HDAC), also regulates FOXP3 functions. HDAC consists of four different classes based on the cellular location, size, number of catalytic pockets, and homology to yeast prototypes [54]:

• •

  Class I: HDAC1, 2, 3, and 8;

• •

  Class II: HDAC4, 5, 6, 7, 9, and 10;

• •

  Class III: SIRT1–7;

• •

  Class IV: HDAC11.

Class I HDACs appear to enhance Treg functions. Conditional deletion of HDAC3 in Tregs results in decreased Treg function and severe scurfy-like phenotypes [40]. Other Class I HDACs, namely HDAC1 and HDAC2, also bind FOXP3 at the N terminal to regulate the expression of many genes [55]. These Class I HDACs do not seem to alter FOXP3 PTM; rather, they seem to be recruited by FOXP3, which in turn counteracts the hyper-acetylation of the promoter, and thereby switches off inflammatory genes.

Li et al. [55] have found that FOXP3 binds directly to Tip60, HDAC7, and HDAC9 at the N-terminal region to form a chromatin-remodeling complex. Although the association of HDAC7 with FOXP3 and Tip60 seems to be important in order for FOXP3 to repress the targeted gene [56], the precise mechanism is as yet unknown. Since HDAC7 has little enzymatic activity and needs to form a complex with other factors such as HDAC3 in order to function [40], HDAC7 may be important in recruiting other FOXP3 cofactors. However, deletion of HDAC6 [41], [57], HDAC9 [42], [57], sirtuin-1 (SIRT1) [44], [57], [58], [59], or HDAC11 [43] increases the acetylation level and overall expression of FOXP3, in addition to enhancing Treg functions.

3.2. Interaction of FOXP3 with kinases

Phosphorylation of FOXP3 was first described by Samanta et al. [60]. Since then, several kinases, such as cyclin-dependent kinase (CDK)2, Pim1, Pim2, and lymphocyte-specific protein tyrosine kinase (LCK), have been reported to phosphorylate FOXP3 [18], [36]. Phosphorylation of FOXP3 by CDK2 [46], [61], Pim1 [62], and Pim2 [45] decreases FOXP3 function. CDK-deficient Tregs show higher suppressive functions than wild-type cells. Mutation of the CDK binding site on FOXP3 increases FOXP3 expression and function.

Pim1 phosphorylates FOXP3 at S418, which reduces the chromatin binding of FOXP3 to targeted genes, and prevents further phosphorylation at S422 of FOXP3, which is reported to be important for FOXP3 function [63]. Pim2 is primarily identified as a FOXP3-dependent gene and is involved in Treg expansion in vitro [64]. We have found that Pim2 interacts with FOXP3 directly, and phosphorylates it at the N-terminal site [45]. Pim2 KO mice have Tregs with higher FOXP3 expression and enhanced suppressive function, and are resistant to dextran sodium sulfate (DSS)-induced colitis; this suggests that phosphorylation by Pim2 also affects FOXP3 stability and function. LCK phosphorylates FOXP3 at Y342 in cancer cells and upregulates its function [65]. However, the mechanism by which LCK interacts with FOXP3 and affects subsequent phosphorylation is as yet unknown.

Phosphorylation of FOXP3 at S418 by an unknown kinase augments the DNA binding of FOXP3 to chromatin [63]. Dephosphorylation of this site by tumor necrosis factor (TNF)-induced protein phosphatase I (PP1) has been observed in arthritis patients, whose Tregs show markedly impaired suppressive function [63].

3.3. Ubiquitination and degradation of FOXP3

Protein ubiquitination is a multistep process that is regulated by the E1, E2, and E3 ligase families. Ubiquitination is caused by the binding of ubiquitin peptides on the lysine residue of the targeted protein. On its own, the ubiquitin peptide has seven lysine residues, including K48 and K63. At least one of these lysine residues on each ubiquitin peptide must be utilized in order to form a polyubiquitin chain [66]. Lysine-48-linked (K48) polyubiquitination is the dominant form that marks proteins for degradation by the 26S proteasome, while non-K48 polyubiquitinations are usually implicated in regulating the targeted protein’s function, rather than in protein degradation [67]. The majority of FOXP3 ubiquitination is related to K48 ubiquitination and degradation [18], [36].

Several pathways induce FOXP3 ubiquitination. HIF1α, which is induced under hypoxic conditions and directly binds to FOXP3, induces FOXP3 polyubiquitination and degradation [29]. Chemokine (C–C motif) ligand 3 (CCL3) signaling also induces polyubiquitination by activating protein kinase B alpha (PKBα) pathways in the psoriatic microenvironment; however, the precise mechanism is still unclear [68]. During T cell receptor (TCR) stimulation, selective inhibition of the phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K)-AKT signaling pathway was found to suppress the proliferation of Tregs, but not of conventional T cells to any significant degree; this suggests that Tregs are more dependent on PI3K-AKT pathways, and that AKT inhibitors can be used to limit Treg infiltration into tumors and to induce immunity against tumors, if they are able to be delivered specifically to Treg [69].

The enzymes that ubiquitinate FOXP3 for degradation are the C terminus of HSC70-Interacting Protein (CHIP) and Casitas-B-lineage lymphoma protein-b (Cbl-b). Chen et al. [70] have shown that STUB1 causes K48 type ubiquitination of FOXP3 at residues K227, K250, K263, and K268 in a heat shock protein (HSP)70-dependent manner. Specific knockdown of STUB1 or HSP70 increases FOXP3 expression, and STUB1 overexpression in Tregs decreases FOXP3 expression level and results in reduced suppressive Treg function in both in vitro studies and in vivo mouse models. Cbl-b also binds to FOXP3 upon TCR stimulation, and, in collaboration with STUB1, ubiquitinates and subsequently degrades FOXP3 [47].

As polyubiquitination reduces FOXP3 stability, it is expected that deubiquitinases will increase FOXP3 stability. Ubiquitin-specific peptidase(USP)7 [48], [71] and USP21 [49] have been reported to directly interact with FOXP3 to induce stability by degrading ubiquitin. In a demonstration of the importance of USP7 and USP21, Treg-specific deletion of either of these peptidases was shown to cause autoimmune reaction in mice [48], [49]. It is interesting to note that although both of these deubiquitinases cause autoimmunity, the deletion of USP7 resulted in a more severe phenotype. In those Tregs, Tip60 expression was also dramatically reduced, indicating that this severity is also partially due to Tip60 malfunction.

3.4. Methylation of FOXP3

A recent mass spectrometric approach has shown that FOXP3 can be di-methylated. Using isomethionine and specific arginine methylation antibody, Geoghegan et al. [72] identified 2502 arginine methylation sites from 1257 tissue-specific and housekeeping proteins in human T cells. From these peptides, they identified a FOXP3 peptide that is methylated at the R51 position. Working separately, we identified protein arginine methyltransferase (PRMT)5, which symmetrically di-methylates arginines [73], as a novel cofactor of FOXP3†. Conditional deletion of PRMT5 in Tregs resulted in a severe scurfy-like phenotype, suggesting that FOXP3 methylation is also critical for FOXP3 functions.