br Introduction Protein modification by ubiquitin
Introduction Protein modification by ubiquitin and ubiquitin-like proteins is one of the most common and important regulatory mechanisms in biology (Finley et al., 2004, Pickart, 2004). Ubiquitination is carried out by an enzymatic cascade consisting of three steps. In the first step, ubiquitin is activated by the ubiquitin-activating enzyme, E1, which forms a thioester bond with the C terminus of ubiquitin in an ATP-dependent reaction. In the second step, the activated ubiquitin is transferred to a ubiquitin-conjugating enzyme, E2 (also known as Ubc). Finally, in the presence of a ubiquitin-protein ligase, E3, the C terminus of ubiquitin is conjugated to the ɛ-amino group of a lysine residue on specific protein substrates. So far only one E1 is known to activate ubiquitin in eukaryotic cells, whereas both E2 and E3 form large families. The human genome encodes approximately 50 E2s that contain a core UBC domain of about 140 amino acids. The number of human ubiquitin E3s, which normally contain a RING or HECT domain in the core subunits, is estimated to exceed 700. In addition, human contains approximately 80 deubiquitination bicuculline (DUB; Nijman et al., 2005), which reverse ubiquitination reactions by removing ubiquitin from protein targets. Thus, the human genome dedicates a large number of genes to the ubiquitin pathway, underscoring the importance of this pathway in cell regulation. Several ubiquitin-like proteins have been discovered in the past few years (Kerscher et al., 2006). These include SUMO, NEDD8, ISG-15, and FAT10 as well as proteins involved in the autophagy pathways (ATG8 and ATG12). These ubiquitin-like proteins contain a C-terminal diglycine motif that can be covalently attached to protein targets through reaction cascades similar to that of ubiquitination. The E1s and E2s, and in some cases the E3s, that are involved in the conjugation of SUMO, NEDD8, ISG15, and ATG proteins have recently been discovered. In contrast, none of the enzymes involved in the activation and conjugation of FAT10 is known. FAT10, also known as diubiquitin, contains two ubiquitin-like domains, similar to ISG-15 (Liu et al., 1999). It is encoded by the major histocompatibility (MHC) class I locus, and its expression is induced by tumor necrosis factor-alpha (TNFα) and interferon-gamma (IFNγ). FAT10 is highly upregulated in hepatocellular carcinoma as well as in other gastrointestinal and gynecological cancers (Lee et al., 2003). The biochemical and cellular functions of FAT10 have remained a mystery. It has been shown that FAT10 binds noncovalently to the spindle checkpoint protein MAD2 (Liu et al., 1999), but the significance of this interaction is not clear. FAT10 is also known to form covalent conjugates with cellular proteins through its C-terminal diglycine motif (Raasi et al., 2001). However, the substrates of FAT10 conjugation have not been identified.
Discussion In this report, we have presented strong evidence that E1-L2 forms a thioester with FAT10 both in vitro and in vivo. After our manuscript was submitted, Jin et al. and Pelzer et al. reported that E1-L2/Uba6 could activate ubiquitin, but not FAT10 (Jin et al., 2007, Pelzer et al., 2007). We noticed that these authors used GST-FAT10 in their assays, which in our hands did not form thioester with E1-L2 (Figure S4). We believe it is important to remove the GST tag or replace it with a smaller tag such as His6 in order to observe the activation of FAT10 by E1-L2. Although FAT10 conjugates have been detected in cells and an intact C terminus of FAT10 is known to be important for its conjugation (Raasi et al., 2001), the enzymes and substrates involved in FAT10 conjugation have remained unknown. E1-L2 is the first enzyme shown to function in the FAT10 conjugation cascade, a process herein referred to as “fattenation” (in analogy to ubiquitination and other modifications such as sumoylation and neddylation). We have tested several known E2s, including Ubc3, Ubc5, Ubc13, and E2-25K, for their ability to accept FAT10 from E1-L2 and found that none of these E2s could function as a FAT10 E2 (Figure S7). This result suggests that both E1-L2 and FAT10 contribute to the selection of a cognate E2. The discovery of E1-L2 should facilitate the identification of potential E2 and E3 that are involved in the fattenation cascade.