Paul Ko Ferrigno Avacta
A Complex Role in Cellular Process & Growing Tool Set to Elucidate It
It is increasingly clear that the ubiquitin-proteasome-system (UPS) is misnamed. Ubiquitylation is not just about targeting proteins for proteasome-mediated degradation, but plays roles in the regulation of transcription, DNA damage response, cell cycle control, and in a host of other cellular processes. Although originally identified as a system that used ubiquitin as a tag for proteins to be degraded by the proteasome, it is now clear that the eight distinct ubiquitin linkages, together with the related “ubiquitin-like” modifiers SUMO and NEDD, form the basis for a complex and nuanced communication network that controls a wide range of cellular processes.
Proteomics studies imply that protein ubiquitylation may be more pervasive in cellular signalling than protein phosphorylation, but the lack of tools for the study of individual modifications and modifiers means that our ability to confirm or refute this concept remains limited. There have been several reviews drawing parallels between the development of these related research areas and, until recently, protein phosphorylation was seen as the faster-moving field, with multiple drugs already in the clinic.
Here we will argue that this is set to change, due to the broader diversity of enzymes involved in the UPS that may be targeted with greater specificity, and the availability of new tools that will enable the development of the assays needed for high-throughput drug screening.
Ubiquitin was identified in 1975 as a small protein that was probably “represented universally in living cells,” hence its name (Goldstein et al., 1975). It was subsequently found by Ciechanover and colleagues to be the heat-stable ATF-1 polypeptide involved in the “ATP-dependent proteolysis system” (Ciechanover et al., 1980). Ubiquitin is 100% conserved across eukaryotic evolution, in plants and animals. Our understanding of its role, though, is continually evolving.
Ubiquitin’s role in biology is now thought to be the regulation of protein behaviour by targeting modified proteins to specific locations in the cell, whether these be proteasomes, membrane compartments, or chromatin domains. Evidence from human proteome-wide mass spectrometry shows that ~4,600 human proteins are modified at more than 23,000 sites—and this is likely to be an underestimate.
At the chemical level, ubiquitin is covalently bound to target proteins through a linkage between the C-terminal glycine of activated ubiquitin and an amino group on the target. This amino group may be the amino terminus of the target or the ε-amino side chain of surface-exposed lysine residues. The covalent modification of lysine draws parallels to protein phosphorylation, where serine, threonine, or tyrosine residues are covalently modified through the addition of a phosphate group.
The exciting twist in the UPS is that ubiquitin itself is a protein with an exposed amino terminus and seven surface-exposed lysine residues, meaning that potentially long chains of ubiquitin molecules using eight different linkages can be formed. These may, in theory, be uniform (all linked via the same residue) or mixed chains where linkages between successive molecules use the amino groups of different residues from one unit to the next. In this respect, at least in its potential complexity, the UPS is perhaps more reminiscent of post-translational modification by glycosylation than by protein phosphorylation.
Glycosylation involves 500 glycoenzymes, including 209 glycosyltransferases, 76 glycosidases, 114 enzymes involved in sugar metabolism and transport, 54 sulfation-related enzymes, and 31 enzymes regulating additional lipid and GPI-anchor biosynthesis pathways (Neelamegham and Mahal, 2016). Similarly, the UPS involves two E1s, ~ 38 E2s, and more than 600 E3 ligases, as well as approximately 100 deubiquitylases (DUBs).
As with glycosylation, these enzymes are able to add a single molecule of ubiquituin to a substrate protein, to add multiple molecules of mono-ubiquitin per substrate, to grow uniform chains of poly-ubiquitin linked using a single linkage type (all K6, all K48, etc.) or to grow mixed chains where successive molecules of ubiquitin are joined by different linkages.
However, whereas glycosylation appears to predominantly affect extracellular proteins, ubiquitylation and the related processes of SUMOylation and NEDDylation are predominantly intracellular processes. Finally, to add to the complexity of the system, ubiquitin and ubiquitin-like modifiers, as well as their substrates, can also be acetylated on the same sites or phosphorylated on a further 11 sites.
There are several key moments when our understanding of protein phosphorylation was enabled by a new tool. One of these involved a new assay— for example, tyrosine phosphorylation was discovered when Anthony Hunter, Ph.D., used an electrophoresis buffer that resolved phospho-tyrosine from phospho-threonine residues (Eckhart, et al., 1979), its intensive study was enabled by the development of a second new tool: a sequence-independent anti-phospho-tyrosine antibody (Druker, 1989).
Here, we briefly consider the tools available for the study of the UPS and some of the discoveries they have enabled.
There are many papers (Kim et al., 2011; Wagner et al., 2011; Ordureau et al., 2015) that use a proteome-wide mass spec approach to attempt to describe the “ubiquitylome.” This technique typically relies on an antibody that binds to the “KGG,” a tri-peptide typically created when the activated di-Gly at the C-terminus of ubiquitin binds to the ε-amino group of a target lysine.
In this approach, cell lysates are subjected to trypsin digestion (which cleaves after lysine and arginine residues), liberating peptides ending with a lysine. The peptide digests are then subjected to affinity enrichment using the anti-KGG antibody before identification using LC-MS, where the peptides are resolved by size and charge by liquid chromatography and analysed by mass spectrometry.
A variation on this approach would be to first immuno-precipitate a protein of interest, and then use the trypsin digest/KGG enrichment/LC/MS process. The problem, of course, is that the protease digestion step dissociates any given chain from its target, so that it is not possible to differentiate between, for example, a mixed K6/K11 chain on a target protein and a mixture of two populations of target protein, one modified with K6 and one with K11 chains. Mass spec can be used to map subcellular proteomes, but the process is limited to whole organelles and lacks the resolution of, for example, immunofluorescence.
In the absence of linkage-specific antibodies for all eight modifications, attempts have been made to use the specificity of natural ubiquitin binding proteins to create engineered recognition tools with the desired specificity. These binders rely on the fact that dimers of an individual ubiquitin-associated (UBA) domain will have much higher affinity for ubiquitin chains than for monomeric ubiquitin.
Indeed, in the first description of tandem ubiquitin binding elements (TUBEs) it was shown that the affinity of a single UBA for ubiquitin was in the µM range, but dimerized UBAs showed nM affinity for specific ubiquitin chains (Herjpe et al., 2009). This has been successful for the M1, K48, and K63 linkages, and generic binders (TUBE1 and TUBE2) with affinity for all eight linkages have also been made.
The utility of commercially available TUBEs (LifeSensors) is to allow the investigation of individual linkage types. However, they are only available against linkages where good antibodies already exist, and have only been validated for mass spec and, more recently, a version of Western blotting using biotinylated TUBE2 (Palazón-Riquelme and López-Castejón, 2016). The sheer amount of work that has been put into these endeavors shows both how difficult it has been to make antibodies with the required linkage specificity, and how widely such tools are needed.
An alternative use of the natural specificity of the UPS is to use the deubiquitinase enzymes that remove the various linked chains. There are over 100 different DUBs encoded by the human genome, but the Komander lab has shown that it is possible to select from among 14 members of the OTU sub-family 8 (OTUB1, AMSH, OTUD3, Trabid, Cezanne, OTUD2/YOD1, Otulin, USP2) that are sufficient to specifically analyse all eight chain types (Mevissen et al., 2013). This assay, marketed by Boston Biochem, requires SDS-PAGE and staining or Western blotting, and also requires that the protein of interest be purified from any cell lysate which might contain contaminating DUBs and free ubiquitin.
There is a potential problem in that longer K48 chains may be more resistant to DUB-mediated hydrolysis. On the plus side, the assay does have the advantage that it monitors deubiquitylation in three ways: the removal of the ubiquitylated substrate, the addition of mono-ubiquitin or poly-ubiquitin chains, and the addition of the non-ubiquitylated target protein.
An alternative approach was taken by Trost et al., who used simple di-ubiquitins and mass spec detection of these the mono-ubiquitins produced by DUB-mediated hydrolysis to measure specificity of 42 different DUBs, as well as the potency of available DUB inhibitors.
Antibodies and Alternative Binding Proteins
The introduction of an anti-K48-linked ubiquitin by Genentech has allowed the coupling of multiple examples of specific protein ubiquitylation events to specific cellular responses or outcomes. These antibodies are now commercially available from a number of suppliers. More recently, antibodies specific for K63-linked chains have become available, and there are rumors of a K11 binder, too. In addition, antibodies specific for Serine-65 phosphorylated ubiquitin should allow greater insights into the role this plays in mitophagy and possibly Parkinson’s disease.
Avacta Life Sciences and their collaborators are developing non-antibody protein binders (called Affimer binders) specific to K6, K33, and K48 linked di-ubiquitins, as well as to the SUMO-1 and SUMO-2 proteins (see, for example, http://www.fbs.leeds.ac.uk/staff/profile.php?tag=Tomlinson_D).
Proteins of the UPS have been implicated in a very wide range of human diseases, including: inflammatory diseases through NF-KB signalling; cancer, through the ubiquitin ligase MDM2 and its substrate, p53; neurodegenerative diseases including Parkinson’s disease, through the ubiquitin ligase Parkin; Huntington’s disease, where aggregates are ubiquitylated; and ALS, through multiple links.
Unsurprisingly the biotechnology and pharmaceutical industries are taking notice, especially given the success of the proteasome inhibitor Velcade? (bortezomib). There is little doubt, given the apparent druggability of DUBs in particular, that the development of high-throughput assays based on novel reagents such as linkage specific antibody and Affimer binders will shortly usher in a broad range of novel therapeutics for a wide variety of indications.
Ciechanover, A., Elias, S., Heller, H., Ferber, S., and Hershko, A (1980) Characterization of the Heat-stable Polypeptide of the ATP-dependent Proteolytic System from Reticulocytes. J. Biol. Chem. 255, 7525–7528
Druker, B, 1989, cited in Kanakura, Y., Druker, B., Cannistra, S. A., Furukawa, Y., Torimoto, Y., and Griffin, J. D. (1990) Signal Transduction of the Human Granulocyte-Macrophage Colony-Stimulating Factor and Interleukin-3 Receptors Involves Tyrosine Phosphorylation of a Common Set of Cytoplasmic Proteins. Blood, 76, 706-715
Eckhart, Hutchinson and Hunter, Cell. 1979 Dec;18(4):925-33
Goldstein, G., Scheid, M., Hammerling, U., Boyse, E. A., Schlesinger, D. H., and Niall, H. D. (1975) Isolation of a polypeptide that has lymphocyte-differentiating properties and is probably represented universally in living cells. Proc. Natl. Acad. Sci. U. S. A. 72, 11–15
Herjpe et al, 2009
Iconomou, M and DN Saunders (2016) Biochem J 473 4083-4101
Mevissen et al, 2013
Neelamegham and Mahal, Current Opinion in Structural Biology 2016, 40:145–152
Palazón-Riquelme P and López-Castejón G (2016), Methods in Molecular Biology 1417, 223-229