In this study we have charted the cellular ubiquitin landscape, focusing on the various modes of ubiquitin conjugates, and ubiquitin-in-conjugates. To take a broadest possible snapshot of the landscape, we refined a protocol for rapid lysis of cells under denaturing conditions without involving further purification steps. This approach enabled us to grasp a representative spectrum of conjugated ubiquitin. The ubiquitin signal is not limited to polyUb chains, but encompasses a variety of chain lengths and linkage types distributed throughout the entire MW range, from unanchored molecules to extremely high MW conjugates (, , , supplemental Figs. S1, S2, S4
). Of polyUb modifications, the most prevalent linkages that we detected in whole cell extract were via
Lys48 and Lys63. This was particularly evident for high MW conjugates. Together these linkages make up close to half of all conjugated ubiquitin, with the bulk of the remainder going to monoUb (and chain terminating endcaps). Very few studies have charted the linkage profile in whole cell extract without including any enrichment steps. One recent study (76
) identified Lys48 and Lys63 as the primary and secondary linkages in high MW conjugates of mammalian extract. A range of linkage distributions has been reported when isolating conjugates to enrich for the ubiquitin content. In one such case, Lys48 and Lys63 made up 90% of chains in conjugates isolated from Drosophila
neuronal tissue (110
). In another case, ubiquitin captured from mouse brain showed the highest content to be of Lys48 linkages, followed by lower levels of Lys63 and Lys11 (111
). Conjugates isolated from urea-denatured yeast lysate identified Lys11 as the second most abundant linkage making up over a quarter of chains and nearing the abundance of the majority type, Lys48 (8
). Lys11 of ubiquitin was shown to play a role in ERAD (8
) and in cell cycle progression (51
), suggesting that levels of Lys11 conjugates cyclically fluctuate. The relatively higher abundance of Lys11 found in lower MW samples of unsynchronized cells (supplemental Fig. S2
) or of specifically ubiquitinated samples (76
) indicates that many Lys11 targets may be small proteins or modified by short chains. Considering these studies together, the ubiquitin linkage profile seems to be extremely sensitive to experimental conditions such as culture medium, growth phase, lysis conditions, and sample handling postlysis. Of the conditions tested herein, TCA-SDS lysis provided the most consistent results and highest levels of all linkages captured. Using these conditions we were able to capture high levels of Lys48 and particularly high levels of Lys63 chains (), which may partially explain the low portion (though not necessarily low levels) of “alternative” ubiquitin chains.
In the high-MW region, where we find the highest ubiquitin-to-protein ratio, the most striking observation is the prevalence of nonextended ubiquitin molecules (). These consist of both monoubiquitin modifications and chain terminators (endcap) implying that average chain length is short (either a majority of short chains or a mixture of few long chains among multiple-monoUb modifications). Monoubiquitination on targets may represent intermediate steps in chain elongation, intermediate steps in polyUb disassembly, or a specific stand-alone signal. Additionally, there is also the option of branched chains (61
) which are impossible to analyze using current MS protocols and therefore may account for a portion of ubiquitin-in-conjugates (adding multiple endcaps per chain). Modifications by multiple mono-ubiquitin on a single substrate were shown to sustain proteasomal processing in vitro
). Mono-ubiquitin has even been shown to target substrates to the 26S proteasome for degradation (albeit at lower efficiency) (114
), suggesting that the cell may be able to tolerate a certain level of premature chain-terminations or short chains. Involvement of monoubiquitin has also been documented as a signal for endocytosis and intracellular sorting of proteins (3
), though the relative portion of such monomeric modifications has not been quantified. Moreover, attributing a distinct, unique, signal to monoubiquitin has been tricky as many substrates for which a role of monoubiquitination has been documented, are also found modified by polyUb chains, particularly Lys63 linked chains (29
Ubiquitin mutants in which arginine has been substituted for one or more of the lysine residues on ubiquitin have been used extensively to tilt the ubiquitin landscape by preventing/promoting polymerization of certain chain linkages over others (21
). In the extreme case, nonextendable lysineless Ub (K0 Ub) has been used as a chain terminator to probe the fate of monoubiquitin conjugates (53
). Because decreasing average chain length has generally been assumed to be toxic to cells, most studies using K0 Ub were performed in vitro
, or in cell free extract. In the current study we assessed the overall effect of K0 Ub induction over WT background and correlated changes to the ubiquitin signal with changes to the population of ubiquitinated proteins. We deliberately expressed K0 Ub from a mild promoter so as not to overwhelm the ubiquitin pool with the mutant ubiquitin. Interestingly, even though K0 Ub consisted of more than half of the ubiquitin molecules in the cells (supplemental Fig. S3
), the resulting stress on growth rates and viability was minimal. K0 Ub was utilized by the ubiquitination machinery to modify proteins, increasing the portion of monoubiquitin from 50% to 80% of total signal in high MW conjugates, reflecting multiple modifications per target (). Even when K0 Ub literally replaces all available cellular ubiquitin (supplemental Fig. S4
), the ubiquitination processes continued generating substrates in heavily ubiquitinated forms, which are improperly targeted and hence accumulate. Over time, the accumulation of these mutant ubiquitin-conjugates eventually leads to cell death.
Analysis of conjugates preferentially tagged with K0 Ub drew a correlation between K0 Ub and Lys63 linkages (). The implication being that K0 Ub was not evenly distributed throughout the ubiquitinome, but was more likely to be found on targets that were also modified with Lys63 chains (, ). In particular, K0 Ub interfered more with the nonessential pathways of protein internalization and sorting rather than with the essential pathways of cell cycle, transcriptional regulation, protein synthesis or general metabolism. Such a skewing of ubiquitin between pathways and target populations was nontrivial, as we initially assumed that K0 Ub would serve as a general chain terminator, equally capping chains of all linkage types. Based on results presented in this study, we propose that K0 Ub concentrates primarily on proteins associated with membrane processes for protein sorting and trafficking, and to a lesser extent with rapidly turned-over short-lived proteasome targets (, , , , and supplemental Tables S3 and S5
Understanding the direct targets and relative roles of monoubiquitination versus Lys63-linked polyubiquitination in endocytosis, protein sorting, and vacuolar targeting is an ongoing endeavor. We wish to raise the possibility, subject to further investigation, that monoubiquitin (or multiple monoubiquitin) is a sufficient signal for most steps in endocytosis and vacuolar targeting but does not represent the preferred signal, particularly so for MVB sorting and late steps of vacuolar trafficking. Overexpression of K0 Ub, which covered 80% of conjugated cellular ubiquitin, triggered a slight defect on internalization of plasma membrane proteins (F), yet interfered to a greater extent with MVB sorting leading to mislocalized vacuolar targets failing to enter the vacuole lumen.
Partially impaired Lys63 modifications on MVB cargoes such as Sna3 (E
) might be one reason for their altered fate. Another reason could be a change in the ubiquitinated status of ESCRT proteins. Strikingly, some ESCRT proteins were recovered in the list of enhanced ubiquitinated proteins ( and ). Notable is the case of Vps27 (Hrs in mammals; ESCRT0) and Bro1, the yeast homolog of Alix (119
). Ubiquitination of yeast ESCRT proteins has not yet been studied in as great a detail as their mammalian counterparts. Thus, human Hrs was described to undergo ubiquitination by AIP4, one of the human HECT domain ligases homologous to Rsp5. Ubiquitination was proposed to trigger a conformational change of Hrs, influencing its function in MVB cargo sorting (120
). STAM (in yeast; Hse1), an Hrs partner, was also described to be ubiquitinated, and to be a substrate of the deubiquitinating enzyme AMSH (121
) that specifically disassembles Lys63-linked ubiquitin chains (122
). Accumulation of ubiquitinated STAM in AMSH deficient cells was shown to result in defective MVB sorting (121
). Likewise, accumulation of ubiquitinated yeast ESCRT proteins induced by K0 Ub expression may impair correct functioning and regulation of the ESCRT machinery. Finding Vps28 as the most significantly diminished protein in the proteome upon induction of K0 Ub coincides with defective MVB sorting (supplemental Table S2
). As a member of the ESCRT I complex, low levels of Vps28 would be expected to hinder sorting of proteins into the endosome (MVBs) with some cargo ending up on the outer membrane of the vacuole as seen in .
In a seeming contradiction to the mild effects observed on proteasome-dependent turnover, a large number of ubiquitination machinery components, including proteasome subunits, were found enriched conjugated to K0 Ub (). Particularly notable are Rsp5 adaptors for ubiquitination of plasma membrane proteins (Bul1/2, Arts) and Rsp5 substrates in connection with the actin cytoskeleton, and components of the 26S proteasome (). Rsp5 itself was also recovered among the protein more heavily ubiquitinated in the presence of K0 Ub. This phenomenon suggests that ubiquitin-proteasome components bear the brunt of the stress imposed by nonextendable K0 ubiquitin, possibly as a ubiquitin-landscape sensing response mechanism. In this respect, finding so many Rsp5-associated proteins raises the possibility that phenotypes observed are largely an indirect result of Rsp5 malfunctioning. As detailed above, Rsp5 is heavily involved in regulating trafficking events via Lys63-ubiquitination of endocytic and MVB cargos (41
). Actin cytoskeleton organization and Rsp5 are genetically linked via rsp5
mutant phenotypes (123
). Indeed, several proteins in connection with actin were already described to undergo Rsp5-dependent ubiquitination (90
), but the underlying function is still largely unknown. The actin cytoskeleton is known to play a crucial role in both yeast and mammals in the internalization step of endocytosis (125
). Several proteins located at actin patches and/or involved in actin cytoskeleton assembly/disassembly or in actin nucleation and polymerization (Ent1/2, Sla2, Rvs167, Syp1, Arp2, and Act1 itself) were present in the list of proteins preferentially ubiquitinated upon expression of K0 Ub (, ), and GO analysis identified this category as prone to perturbation (). Of particular interest is the case of Syp1, a protein that interacts with Ede1 and is one of the first proteins arriving at actin patches, the sites of endocytosis (100
). Our study emphasizes that ubiquitination and deubiquitination events play subtle roles in the regulation of protein/protein interactions in the process of actin-mediated endocytosis.
Taking into account ubiquitin in all its forms, the resulting ubiquitin landscape seems to be roughly dissected into two main spheres: the ubiquitin proteasome arm (dominated by Lys48-linked polyUb), and the ubiquitin trafficking arm (where Lys63 linkages and monoUb modifications prevail). The two branches of the ubiquitin system are codependent as they both share the pool of free unconjugated ubiquitin, a large portion of which is recycled from conjugates. A profound implication is that inhibition of one branch could tie up ubiquitin in conjugates leading to a shortage of available ubiquitin, and in turn inhibit other processes indirectly. In this respect, the effectiveness of K0 Ub to dissect the ubiquitin landscape into its two arms may serve as a powerful experimental tool to tweak and study specific cellular processes, which may be encouraging for drug discovery efforts.