1. Introduction
1.5 Ubiquitin‐like modifiers
1.5.5. FAT10
1.5.5. FAT10
In 1996 Fan et al. found seven new genes in the human major histocompatibility complex (MHC) class I region at the human leukocyte antigen (HLA)‐F locus on chromosome 6, one of them showing high similarity to a doubled ubiquitin (Fan, Cai et al. 1996). Thus the 18 kDa protein was named diubiquitin or ubiquitin D. Later the nomenclature HLA‐F adjacent transcript (FAT) 10 was asserted.
FAT10 was first found to be expressed in mature dendritic cells and mature B‐cells (Bates,
Ravel et al. 1997), but then it was discovered that FAT10 is strongly upregulated in all cell types after stimulation with the proinflammatory cytokines Interferon (IFN) and Tumour necrosis factor (TNF) (Liu, Pan et al. 1999, Raasi, Schmidtke et al. 1999). Further studies revealed that under non‐inflammatory conditions endogenous FAT10 expression is highest in spleen, thymus, gut and lymph nodes of mice (Lukasiak, Schiller et al. 2008), suggesting a role of FAT10 in the immune system. The mild phenotype of FAT10 knockout mice under laboratory conditions merely can be challenged by lipopolysaccharide (LPS) to induce endotoxin hypersensitivity (Canaan, Yu et al. 2006). This hints to an involvement in specialized immune reactions as FAT10 ‐ in contrast to other parts of the adaptive immune system ‐ is found only
The two domains of FAT10 have 29% and 36% sequence homology to ubiquitin, respectively, but only 20% homology to each other (Bates, Ravel et al. 1997). Although the structure is not yet completely solved (Theng, Wang et al. 2014), both domains are expected to have the typical ubiquitin fold with a five amino acid linker in between which is assumed to allow flexible positioning to each other (Figure 2). Corresponding to the amino acid sequence of ubiquitin human FAT10 harbours four lysines (K27, 33, 48 and 63) at the same positions whereas in mice only lysine 48 is conserved. Different to ubiquitin FAT10 contains two cysteines per domain but it exhibits the C‐terminal diglycine motif as a common feature of all ULMs (Bates, Ravel et al. 1997). Most of them ‐ as well as ubiquitin itself ‐ are produced as precursors and need to be processed. However FAT10 is translated with a free diglycine motif and can be directly attached to lysines in substrate proteins in a ubiquitin‐like manner (Raasi, Schmidtke et al. 2001) by an enzyme cascade sharing even the same enzymes with ubiquitin.
Figure 2: Ribbon model of ubiquitin (left) and FAT10 (right). Both domains of FAT10 display the typical ‐grasp fold of ubiquitin, witha central ‐helix (turquoise) surrounded by ‐sheets (purple) (Groettrup, Pelzer et al.
2008).
FAT10 is activated by the E1 enzyme UBA6 which was found to prefer ubiquitin for thioester formation (Chiu, Sun et al. 2007, Pelzer, Kassner et al. 2007). Nevertheless Gavin et al.
measured a higher binding affinity of FAT10 to UBA6 in vitro (Gavin, Chen et al. 2012). It is assumed that massive expression of FAT10 during infections changes the ratio to ubiquitin increasing the probability for FAT10 to be taken by UBA6 for activation (Chiu, Sun et al. 2007, Gavin, Chen et al. 2012). The E2 conjugating enzyme UBA6 specific enzyme (USE) 1 is bispecific
for ubiquitin and FAT10, too (Aichem, Pelzer et al. 2010), and takes over – as its name indicates
‐ both ULMs only from UBA6. However, FAT10 can’t be activated and conjugated by the UBA6 dependent E2s Ubc5 and Ubc13 or by the second ubiquitin E1 (UBE1) and corresponding E2s (Chiu, Sun et al. 2007). Knockdown of either UBA6 or USE1 massively reduce FAT10 conjugates proposing that these are the main if not only E1 and E2 enzymes for FAT10, respectively (Chiu, Sun et al. 2007, Aichem, Pelzer et al. 2010).
As there are no E3 ligases for FAT10 published yet, one might speculate that in vitro the E1 UBA6 and the E2 USE1 are sufficient for conjugation to substrate proteins likewise it was published for SUMO (Tatham, Jaffray et al. 2001, Werner, Flotho et al. 2012). Recently it could be demonstrated that at least in vitro even the E1 UBA6 is sufficient for conjugation of FAT10 to its substrate protein UBE1 (Bialas, Groettrup et al. 2015), which might be due to high protein concentrations and thus close proximity of all involved proteins. Nevertheless the need for an E3 ligase to allow substrate specific FAT10ylation in vivo can’t be ruled out and several attempts have been done to identify putative candidates, which are still under investigation (J. Bialas, unpublished data).
The incomplete knowledge about the FAT10‐conjugation enzyme cascade arises the question about deconjugating enzymes. Expression of linear ubiquitin‐GFP in cells leads to rapid cleavage of the fusion proteins, whereas FAT10‐GFP stays intact (Hipp, Kalveram et al. 2005).
But keeping in mind that FAT10 does not need to be processed to liberate its diglycine motif it seems to be likely that there is no enzyme cleaving linear FAT10 fusions. On the other hand endogenous FAT10 conjugates are degraded at the same rate as free FAT10 which is very short lived (Hipp, Kalveram et al. 2005), why it might be possible that there are no deconjugating enzymes for FAT10 at all.
The first substrate identified was the E2 enzyme USE1 which auto‐FAT10ylates itself leading to its degradation by the proteasome (Aichem, Pelzer et al. 2010, Aichem, Catone et al. 2014).
This auto‐modification relies on the active site cysteine of USE1 which transfers the activated FAT10 onto its lysine323, however when it’s mutated another lysine is chosen (Aichem, Catone et al. 2014). Interestingly also the ubiquitin E1 enzyme UBA1 (also UBE1) is a substrate of FAT10 and becomes degraded by the proteasome (Rani, Aichem et al. 2012, Bialas, Groettrup et al. 2015), which might be a mechanism to bias the conjugation machinery
Involvement of FAT10 in other degradation mechanisms can be assumed e.g. because of the modification of the autophagy adapter protein p62 (Aichem, Kalveram et al. 2012). It is multi‐
monoFAT10ylated meaning modified with one FAT10 at several lysines simultaneously which marks p62 for degradation via the proteasome. Additionally p62 interacts also non‐covalently with FAT10 (Aichem, Kalveram et al. 2012), but a function for the interaction couldn’t be
established yet.
NEDD8 ultimate buster 1 long (NUB1L; see also in section 1.6), which is inducible by IFN (Kito, Yeh et al. 2001), was shown to interact non‐covalently with the N‐terminal domain of FAT10 accelerating its degradation. As NUB1L interacts also with the proteasome subunit Rpn10 it was proposed to transfer FAT10 to Rpn10 in order to facilitate the degradation of FAT10 (Hipp, Raasi et al. 2004, Tanji, Tanaka et al. 2005, Rani, Aichem et al. 2012).
FAT10 itself was reported to be modified by ubiquitination (Hipp, Kalveram et al. 2005, Buchsbaum, Bercovich et al. 2012) and by at least slight acetylation (Kalveram, Schmidtke et al. 2008). As it was shown that degradation of FAT10ylated substrates is independent of further ubiquitination (Hipp, Kalveram et al. 2005, Schmidtke, Kalveram et al. 2009), the significance of these modifications for the function of FAT10 is still under debate.
Despite the characterization of the degradation of some substrates the role of FAT10 in biological processes is poorly understood. Before substrate proteins or even E1 and E2 enzymes were identified, the spindle assembly checkpoint protein mitotic arrest deficiency (MAD) 2 was found as an non‐covalent interaction partner of FAT10 (Liu, Pan et al. 1999).
Because it’s expressed in B‐cells and dendritic cells the authors speculated that FAT10 modulates cell growth during development and activation. Further investigation by Ren et al.
showed delocalization of MAD2 from the kinetochore upon FAT10 overexpression, leading to missegregation of chromosomes (Ren, Kan et al. 2006), which could be inhibited by siRNA knockdown of FAT10 at simultaneous TNF treatment (Ren, Wang et al. 2011). As FAT10 expression was claimed to be upregulated in hepatocellular, colorectal, ovarian and uterus carcinomas it is implied to promote tumorigenesis (Lee, Ren et al. 2003). The tumour suppressor p53 was first linked to FAT10 as a negative regulator of expression and a loss of p53 leading to transcriptional upregulation of FAT10 contributing to tumorigenesis (Zhang, Jeang et al. 2006). Additionally, FAT10 was described to upregulate transcriptional activity of
p53 and modify inactive p53 (Li, Santockyte et al. 2011). Choi et al. proposed a synergistic activation of FAT10 expression by NF‐B and STAT3 which leads to p53 repression (Choi, Kim et al. 2014). Further FAT10 was suggested, besides mutant p53, to be used as prognostic marker for gastric cancer (Ji, Jin et al. 2009).
Contradictory to the statements mentioned before cytokine‐induced expression of FAT10 could be a response to pro‐inflammatory stimuli of the surrounding tissue to fight the tumour.
The simultaneous upregulation of the immunoproteasome subunit LMP2, changing the subset of MHC class I ligands (see also section 1.7.2) argues for this (Lukasiak, Schiller et al. 2008).
Whether FAT10 expression promotes or suppresses induction of apoptosis is discussed very oppositional. Some groups could demonstrate that FAT10 overexpression induces apoptosis in mouse fibroblasts, HeLa cells and renal tubular epithelial cells (Liu, Pan et al. 1999, Raasi, Schmidtke et al. 2001, Ross, Wosnitzer et al. 2006) and that FAT10 knockdown protects from apoptosis (Ross, Wosnitzer et al. 2006). In a recent study FAT10 was found to interfere with the NF‐B pathway by modifying leucine‐rich repeat Fli‐I‐interacting protein (LRRFIP) 2 (Buchsbaum, Bercovich et al. 2012). FAT10‐modified LRRFIP can’t be recruited to the LPS‐
sensing Toll‐like receptor (TLR) 4 anymore and therefore NF‐B mediated transcription of apoptosis inhibitors is prevented. In contrast, Ren et al. observed a protective function of FAT10 expression from TNF induced apoptosis in the colon cancer cell line HCT116 (Ren, Kan et al. 2006).
Various other interaction partners were found which point to different tasks of FAT10. The histone deacetylase (HDAC) 6 mediates the transport of polyubiquitinated proteins to aggresomes upon proteasome inhibition. It acts as a linker between ubiquitin chains and dynein which brings substrates to these inclusion bodies for degradation via autophagy. Under these conditions FAT10 interacts non‐covalently with HDAC6 and localizes to aggresomes, too (Kalveram, Schmidtke et al. 2008).
A possible role for FAT10 in eye development was suggested recently. The aryl hydrocarbon receptor‐interacting protein‐like 1 (AIPL1) which is mutated in Leber’s congenital amaurosis (LCA) was shown to interact with NUB1 (van der Spuy and Cheetham 2004) and free as well as conjugated FAT10 build a ternary complex (Bett, Kanuga et al. 2012). The authors claimed that
is abolished by pathogenic mutations in AIPL1. Additionally AIPL1 associates with UBA6 which implicates further regulation of FAT10ylation (Bett, Kanuga et al. 2012).
As FAT10 seems to be involved in various processes and its role there is heavily discussed, a lot of investigation is needed to shed light on its actual impact on these pathways.