Post-translational modification of proteins offers a rapid route to change the activity of crucial factors within the cell. One of the more drastic post-translational modifications, in terms of effect on biochemical properties, is the covalent attachment of the small protein ubiquitin, to a target factor. The labile nature of some post-translational modifications puts obstacles in the path of attempting to detect modified species of most proteins. Indeed, ubiquitination can be rapidly reversed by the action of a large family of DUBs (deubiquitinating enzymes), most of which are cysteine proteases. This, taken together with the rapid proteasomal degradation of some species of ubiquitinated proteins, results in difficulties in detecting modified targets. In this review, practical approaches developed for the detection, purification and characterization of ubiquitinated proteins are reviewed. After a brief appraisal of the use of histidine-tagged ubiquitin, focus is placed on development of UBD (ubiquitin-binding domain)–ubiquitin affinity purification.
The modification of a substrate with ubiquitin may be in the form of mono-ubiquitination (possibly at multiple sites; this is referred to as multiple mono-ubiquitination) or as polyubiquitination. Whereas mono-ubiquitination is the attachment of a single ubiquitin molecule to a substrate at one site, polyubiquitination of a protein is the consequence of recurring ubiquitin modification of ubiquitin itself, on one of its seven internal lysine residues. Different polyubiquitin chain types have been described to mediate different cellular functions – while chains linked through Lys-48 have mainly been reported to lead to proteasomal degradation , chain linkage through Lys-63 is involved in a number of different processes, e.g. signal transduction and endocytosis [2,3]. Other chain linkages exist (seven in total), but available information is still not sufficient to attribute a general function to these. To connect with effector functions, there is a requirement for adaptor proteins to recognize modified substrates, in order to mediate an appropriate response, depending on the nature of the particular ubiquitination event.
UBDs (ubiquitin-binding domains) are a diverse family of structurally dissimilar protein modules that bind mono- and poly-ubiquitin. Two of the best described UBDs are the UBA domain (ubiquitin-associated domain) and the UIM (ubiquitin-interacting motif). The UBA domain family is structurally well conserved as domains consisting of compact three-helix bundles. UBA domains have been classified into four different groups with respect to ubiquitin-binding properties. Classes 1 and 2 are defined as preferentially binding Lys-48 and Lys-63 polyubiquitin chains respectively. Class 3 consists of UBA domains that do not bind polyubiquitin chains, and class 4 does not exhibit any particular binding specificity regarding type of polyubiquitin .
The UIM is a short helical motif that can be found in a wide variety of proteins. The first UIMs to be characterized were found in the proteasomal subunit S5a, which contain two functionally independent regions that can bind ubiquitin . The UIMs of S5a interact preferentially with polyubiquitin, while binding weakly to mono-ubiquitin. Since its discovery, variants of the UIM have been reported, such as the MUI (motif interacting with ubiquitin) and the double-sided UIM .
Both the rapid proteasomal degradation and the dynamic equilibrium between modification and de-modification of substrates by the DUBs (deubiquitinating enzymes) result in technical obstacles when attempting to analyse the ubiquitination status of a particular protein under physiological or pathological conditions. Although proteasomal inhibition, using drugs such as lactacystin or MG-132 (the proteasome inhibitor carbobenzoxy-L-leucyl-L-leucyl-leucinal), can inhibit certain proteolytical activities in the proteasome and thus allow analysis of modified protein, these drugs have been shown to also have undesirable effects on protein synthesis . In order to avoid the action of DUBs and to preserve modified substrate in cell lysates, it is necessary to use general cysteine protease inhibitors, such as the noxious compounds IAA (iodoacetamide) or NEM (N-ethylmaleimide).
Use of histidine-tagged ubiquitin
The use of tagged ubiquitin molecules is perhaps the most well-established strategy when trying to identify whether a factor is regulated by ubiquitination. There are many examples of the use of His6-tagged ubiquitin in combination with Ni2+-chelate affinity chromatography [8–11] for purification, concentration and analysis of a particular protein. The fact that cell lysis and purification may be performed using strong denaturants to preserve modified protein contributed to the popularity of this technique (Figure 1A).
Illustration of approaches based either on tagged ubiquitin or on the purification of untagged endogenous ubiquitin
Several different proteomics approaches have made use of histidine-tagged ubiquitin. Peng et al.  used a yeast strain with His6–ubiquitin as the single source of ubiquitin. LC (liquid chromatography)/LC-MS/MS (tandem MS) analysis of Ni2+-purified ubiquitin conjugates resulted in the identification of 1075 proteins and 110 ubiquitinated lysine residues from a subset of 72 proteins. Interestingly, the authors noted that 33% of the proteins with defined ubiquitination sites were modified on more than one lysine residue, indicating that modification on multiple residues is a frequently occurring event. Further, this study  resulted in the first in vivo identification of Lys-11-, -33-, -27- and -6-linked polyubiquitin chains. His6-tagged ubiquitin was moreover used in a similar study performed in mammalian cells . Here, Kirkpatrick et al.  report on the stabilization of His6-tagged ubiquitin under the control of the endogenous polyubiquitin precursor promoter (Ubc) in HEK-293 cells (human embryonic kidney cells). This leads to physiological expression levels of the tagged molecule, sidestepping any artefacts associated with ubiquitin overexpression. However, the authors remark that the availability of endogenous ubiquitin will lead to a less efficient purification of modified protein. In an attempt to reduce non-specific background intrinsic to the use of histidine tags, Tagwerker et al.  developed an HB (histidine–biotin) tag, which allows tandem purification under fully denaturing conditions in all steps. The HB tag allows very stringent purification conditions, avoiding non-specific background and co-purification of interacting partners. Notably, the LC-MS/MS analysis of purified Saccharomyces cerevisiae proteins resulted in the identification of ubiquitin peptides that were modified on more than one site, supporting the existence of forked ubiquitin chains in vivo. In addition to work done in yeast and cell culture, ubiquitin profiling has also been reported using transgenic mice expressing histidine-tagged ubiquitin [15,16].
Even though proved to be a highly useful research tool, the use of histidine-tagged molecules entails the possibility of an influence of the tag on the physiology of the molecule. Moreover, competition of tagged exogenous ubiquitin with the endogenous wild-type molecule will make it difficult to reach conclusions, which is particularly true for cases where the exogenous ubiquitin has been mutated on specific lysine residues (Figure 1A). With this in mind, strategies based on UBDs have been developed in order to purify modified proteins conjugated to endogenous wild-type ubiquitin (Figure 1B).
UBD-based detection of ubiquitination
Use of UBA domains
The physiological purpose of the UBA domain will necessarily depend on the nature of the protein in which it is found; however, several UBA domain-containing proteins have been suggested to serve as factors shuttling ubiquitinated substrates to the proteasome (e.g. hHR23A, p62 and Dsk2). This particular function would be achieved by binding to ubiquitinated substrates through the UBA domain, and binding to the proteasome via another domain, often a UBL (ubiquitin-like domain). The fact that UBA domains bind polyubiquitin with high affinity has been exploited in the development of tools for the purification of ubiquitinated proteins. The relative ease of production of UBA–agarose conjugates, as compared with e.g. production of ubiquitin antibody, makes these domains an attractive tool for use in ubiquitin pull-down experiments. However, apart from interacting with ubiquitin, some UBA domains interact with UBL domains [18,19,25,34], and other proteins [20–24], raising questions regarding their specificity with respect to ubiquitin.
The affinity of various UBA domains for mono-ubiquitin ranges from the low micromolar to several hundred micromolar , whereas the affinity for polyubiquitin chains of different linkages is consistently higher. For example, the UBA from ubiquilin 1 binds to mono-ubiquitin with an affinity of 27 μM, whereas it shows an affinity of 1.2 and 0.5 μM for Lys-48- and -63-linked tetra-ubiquitin respectively, making this UBA one of the tightest ubiquitin binders. Further, this UBA domain appears to bind strongly also to Lys-29- and -6-linked polyubiquitin . Recently, structural results have indicated that the high affinity of this UBA for ubiquitin is due to unique interaction between this domain and ubiquitin .
In the other extreme of the spectrum is the p62 (Sequestome 1) UBA domain, binding weakly (540–750 μM) to mono-ubiquitin [4,26]. The binding preference of the p62 UBA is not clear – while an apparent selectivity for Lys-63 polyubiquitin in vivo seems to exist, little or no discrimination between chain type linkages appears to be present in in vitro pull-down experiments. It has been suggested that the apparent lack of chain specificity observed in ubiquitin pull-down experiment using UBA-conjugated Sepharose is due to interactions of multiple UBAs with a single polyubiquitin chain, masking any subtle differences in chain binding specificity [4,26,27].
In a proteomic study reported by Pridgeon et al. , an expressed cDNA library from human adult brain was screened for proteins interacting with the p62 UBA domain. Here, rabbit reticulocyte in vitro transcription/translation of pooled cDNAs was argued to allow ubiquitination of newly synthesized proteins, which were subsequently purified by use of p62 UBA–agarose. In this manner, the authors identified 11 proteins presumably ubiquitinated. Although a powerful method, the described in vitro system necessitates subsequent validation of positive clones to confirm the in vivo relevance of any observed ubiquitination.
Furthermore, the use of p62 UBA-conjugated agarose is exemplified by a study of ischaemic tolerance in neurons. Highlighting the flexibility of the UBA–agarose, the authors performed studies both assessing the ubiquitination of a single protein (Bim), as well as a subsequent proteomic study, coupling pull down experiments to MS/MS analysis, thus allowing the identification of the ubiquitinated proteins involved in neurological ischaemia [29,30]. Yet another example of the use of p62-conjugated agarose is provided by Drake et al. , reporting evidence supporting a role of ubiquitination of the B-cell receptor in the control of antigen processing.
The use of UBDs in conjunction with MS has allowed a variety of proteomic studies of ubiquitinated proteins – or the ‘ubiquitome’. To date, it appears that the general UBDs of choice are the p62 UBA domain or the S5a UIMs (discussed below); however, several other alternative UBA domains are well described, and may indeed constitute excellent tools for the purification of modified protein.
Use of UIMs
The UIM has been used for enrichment of ubiquitinated protein in several different studies. Regarding the specificity of the UIMs, several reports have shown that the second UIM in S5a binds strongly to the UBLof hHR23A/B [32–34]. As hHR23A/B contains two UBA domains, the presence, absence or relative abundance of this protein may have an effect on the isolation of ubiquitinated material.
In a study by Layfield et al. , full-length S5a was used for enrichment in ubiquitinated protein. In this study , an attempt was made to resolve ubiquitinated proteins by 2DE (two-dimensional electrophoresis); however, the authors note that due to the limited success of this, deubiquitination of purified ubiquitinated protein may be needed for sufficient resolution of spots. This idea was implemented by Ventadour et al. , who also constrained the affinity purification to the use of the region of S5a containing the UIMs. This work resulted in data indicating that several proteasomal subunits are substrates for ubiquitination, a fact that corresponds well with results from a study where the purification strategy was based on the use of anti-ubiquitin antibody . The use of the S5a UIMs can further be exemplified by several studies, such as the detection of higher levels of ubiquitinated proteins in dilated cardiomyopathy , the implication of ubiquitin in protein turnover as a control of synapse remodelling and signalling  and to probe the mechanism of action of the anthrax lethal toxin inhibitor, celastrol . In a recent proteomic study performed in plants, regions from two Arabidopsis proteins containing triple UIMs (Arabidopsis S5a orthologue) and double UBAs (Arabidopsis UBP14) respectively were used to pull down ubiquitinated substrates in Arabidopsis thaliana . A noteworthy result from this report is the observation of a mere 14% overlap between proteins identified as ubiquitinated, using either UIMs or UBA domains. This suggests that the use of UBDs for the purpose of identifying ubiquitinated proteins on a proteomewide scale may result in the characterization of a subset of the ubiquitome. Indeed, while the double UBA domains could successfully be used to identify polyubiquitin chain types linked through Lys-48, -63, -11, -33 and -29, no evidence for Lys-33 polyubiquitin interaction with the triple UIMs was obtained . These results suggest that exploiting several types of UBDs may be a strategy allowing the capture of a representative set of ubiquitinated proteins. Moreover, as S5a has been reported to have a preference for longer polyubiquitin chains, whereas it binds weakly to mono-ubiquitin, it is possible that the use of this particular UBD in proteomic approaches will lead to an underestimation of the importance of mono-ubiquitination in the control of vital cellular regulators .
In order to detect, purify and characterize ubiquitinated proteins, preservation and concentration of ubiquitinated material is necessary. This necessity is in part due to the labile nature of this modification, with respect to DUB action, and also due to the rapid and efficient proteasomal degradation of many ubiquitinated cellular factors. The elucidation of post-modification processes will require efficient tools for the capture of transient ubiquitination events, which have important consequences in the regulation of critical factors. The implication of ubiquitin signalling and the ubiquitin proteasome system in human disease has resulted in efforts directed at identifying drugs acting on various levels of this system . The development of improved tools to purify the proteomic fraction modified with endogenous wild-type ubiquitin, combined with already existing methods such as SILAC (stable isotope labelling with amino acids in cell culture)/MS, will contribute to the understanding of how proteins are regulated by ubiquitination, thus permitting the advance of therapeutic strategies.
Third Intracellular Proteolysis Meeting: A joint Biochemical Society and INPROTEOLYS Network Focused Meeting held at Auditorio de Tenerife, Santa Cruz de Tenerife, Canary Islands, Spain, 5–7 March 2008. Organized and Edited by Rosa Farràs (Centro de Investigación Príncipe Felipe, Valencia, Spain), Gemma Marfany (Barcelona, Spain), Manuel Rodríguez (CICbioGUNE, Derio, Spain), Eduardo Salido (La Laguna, Tenerife, Spain) and Dimitris Xirodimas (Dundee, U.K.).
We apologize that many relevant papers could not be cited due to space constraints. This work was funded by the Ramón y Cajal Programme, Ministerio de Educación y Ciencia (Spain) grant BFU 2005-04091, the FIS (Fondo de Investigaciones Sanitarias) CIBERhed, the Department of Industry, Tourism and Trade of the Government of the Autonomous Community of the Basque Country (Etortek Research Programmes 2005/2006) and by the Innovation Technology Department of the Bizkaia Country.