During the past 15 years, protein modification by ubiquitin (UBQ) and ubiquitin-like (UBL) molecules has been suggested to be an important mechanism in regulating numerous critical cellular processes, such as signal transduction, transcriptional control, protein degradation, epigenetic modification and intracellular localization. UBQ is a 76 amino-acid protein that is covalently linked to lysine residues of target proteins in a multi-step process. The process of UBQ conjugation involves three basic classes of enzymes. UBQ-activating enzymes (E1s) activate the C terminus of UBQ, resulting in the formation of an ATP-dependent thio-ester that is linked to the E1's active site - cysteine residue. Afterwards, UBQ is transferred to the active site - cysteine residue - of an E2 conjugating enzyme. The third class of enzymes, the UBQ-protein ligases (E3s) catalyze the final reaction (the UBQ-conjugation process), in which the UBQ is attached to the lysine residue of the protein substrate (reviewed in [1
]). As a result of the ubiquitylation (ubiquitination) process, the target protein can be modified by a single UBQ (mono-ubiquitylation) or several UBQs (poly-ubiquitylation), which determines the tagging outcome [3
Deconjugation of UBQ- and/or UBL-modified protein substrates is essential to recycle both the substrate proteins and the UBQ or UBL modifiers; thus, this process influences protein-protein interactions, signaling processes and the pool of free UBQ/UBL within the cell. The deconjugation process is managed by deubiquitylating enzymes (DUBs). The DUBs are in a heterogeneous group of cysteine or metalloproteases [4
], which can cleave the scissile peptide bond between the last residue of UBQ (Gly 76) and target proteins. Since their discovery in 1990s, DUBs have been emerging as key regulators in many basic cellular processes. For example, they modulate the ubiquitin-proteasome system (reviewed in [5
]). Recently, numerous studies have shown that DUBs are linked to diseases such as cancer. CYDL is a tumor suppressor [7
]; UCHL1 and USP24 are involved in Parkinson's disease [12
]; ATXN3 influences ataxias [17
Most extensively, DUBs have been studied in the human genome [4
]. The classical DUBs can be divided into five subclasses based on their UBQ-protease domains. These are (i) UBQ-specific proteases (USPs), (ii) otubain proteases (OTUs), (iii) UBQ C-terminal hydrolases (UCHs), (iv) Machado-Joseph disease proteases (MJDs) and (v) JAMM motif proteases (JAMMs). Nijman et al.
have identified 95 DUBs, out of which the overwhelming majority (58) belongs to the USP class. Whereas the DUBs in the JAMM family are metalloproteases, the other four groups belong to the cysteine protease class. Additionally, two more subclasses of DUBs have been suggested based on evidence from sequence homology [19
]. The so-called PPPDE group involves papain-like proteases and the WLMs (Wss1p-like proteases) are metalloproteases. Both are hypothesized to play a role in removal of UBQ or UBL from substrate proteins. Experimentally, DUBs can be identified with several approaches such as using hemagglutinin-tagged (HA) UBQ-derived probes [20
] or biotin-labeled UBQ aldehyde probes [21
] for labeling. These methods had been used to identify several members of the USP, UCH and OTU families [6
]. Alternatively, extension of families of known or suspected representatives of DUB classes is possible with the in silico
collection of families of significantly similar sequences within the homology concept.
It must be noted that the understanding of the DUBs' functions remains very limited despite considerable research efforts. Apparently, the problems are associated with the large number of various DUBs (some of which might compensate for others in parallel pathways) and with their position in the subcellular gene networks where a single DUB can affect several pathways via its protein substrates. One possibility to get new insight into the problem is to involve new model organisms into the analysis and to rely on their specific methodological advantages. The zebrafish (Danio rerio
) system with its morpholino (MO) technology is superbly suited for easily studying the loss-of-function phenotype in vivo
, important information necessary to elucidate the genes' function [22
]. In this work, we use the zebrafish model to identify the MO knockdown phenotypes of potential DUB genes that we have identified with in silico
methods from the genome sequence. The respective experiments have been carried out for 85 putative DUBs. 57 out of the 85 genes were found to be required for normal development and their morphants have exhibited complex phenotypes always with inclusion of notochord deformations. In order to narrow down the functions in development for some of the genes, we chose a neuronal marker, huC
, for grouping the DUBs into subgroups that provide hints for specific pathway involvement. This first loss-of-function study of DUB genes in zebrafish provides a valuable resource for further DUB studies in vertebrates.