Genome-wide approaches to systematically identify substrates of the ubiquitin/proteasome pathway

Abstract

The ubiquitin/proteasome system handles the majority of controlled proteolysis in eukaryotes. Defects in the ubiquitin/proteasome system have been implicated in diseases ranging from cancers to neurodegenerative disorders. However, the precise role of ubiquitin/proteasome-mediated degradation in health and disease is far from clear. A major challenge is to link specific substrates directly to a particular degradation pathway. Here, we review genome-wide approaches that have been developed in recent years to comprehensively identify ubiquitylated substrates of a particular pathway. The components of the ubiquitin/proteasome system are attractive drug targets, as illustrated by the efficacy of some proteasome inhibitors in the treatment of multiple myeloma. Information that has emerged from these studies could reveal novel drug targets and strategies for treating human diseases.

Introduction

Ubiquitin (Ub), an abundant 76-residue protein, is highly conserved from yeast to human. Ubiquitylation, a post-translational modification in which Ub is covalently attached to lysine residues, regulates a myriad of cellular processes, including cell cycle progression, signal transduction, DNA repair and inflammation [1–4]. Not surprisingly, the Ub system has been demonstrated to be a major player in cancers and neurodegenerative disorders [3–5]. Besides its involvement in protein transport and processing [2, 6, 7], Ub is best known as a signal that targets proteins for destruction by a multi-subunit, ATP-dependent protease, termed ‘the proteasome’ [3, 8]. More specifically, substrates are recognized by a Ub-protein ligase (E3), which assembles the Ub chains onto the substrates with the assistance of a Ub-activating enzyme (E1) and a Ub-conjugating enzyme (E2) (Figure 1). Subsequently, Ub-binding proteins (e.g. Rad23, Dsk2, Rpn10, Cdc48) recognize ubiquitylated substrates and facilitate the substrate transfer to the 26S proteasome (Figure 1), which irreversibly cleaves substrates into shorter pieces [2, 8].

Figure 1.

Schematic for ubiquitin (Ub)/proteasome-mediated proteolysis. The Ub/proteasome system consists of two phases [2, 8]: Ub conjugation and substrate transfer. In the Ub conjugation phase, Ub is activated and transferred to the target substrate through several enzymes, including a Ub-activating enzyme (E1), a Ub-conjugating enzyme (E2) and a Ub-protein ligase (E3). The most critical step is the recognition of specific targets by a relevant E3, which triggers substrate ubiquitylation. For substrates targeted for degradation, successive Ub molecules are added to form a Ub chain on the substrates. In the substrate delivery phase, Rad23-like Ub receptors use separate domains to bind ubiquitylated substrates and the proteasome, thereby delivering ubiquitylated substrates to the proteasome for subsequent degradation. The detailed events involved remain unresolved.

The biological significance of the Ub/proteasome system lies in the cellular processes under its regulation. A major obstacle then is to assign specific substrates to a particular degradation pathway. This can be a rather daunting task, considering that the first cellular substrate (Scc1) of the N-end rule pathway - the first genetically defined Ub-dependent proteolytic pathway - was identified fortuitously after over a decade of intense effort [9–11]. Many strategies have been developed in the past to isolate the substrates of a specific Ub/proteasome pathway with varying degrees of success. Here, we will review the strategies that have been employed to systematically identify ubiquitylated substrates of a specific pathway as defined by the E3 used (Table 1). This review will mainly focus on genome-wide approaches that methodically query each gene or protein, leaving out affinity purification-mass spectrometry coupled strategies that have been the subject of several reviews and research papers [12–19], as well as the bioinformatics approach [20] that can provide leads to potential substrates if the consensus sequence has been established for ubiquitylation targets of a particular pathway (e.g. D box, PY motifs) [2, 8, 21].

Table 1.

Selected E3s

E3s	recognition signal	Function/substrates	Disease	Ref.
Ubr1	A destabilizing residue at the N-terminus of a substrate	Angiogenesis, apoptosis, cardiovascular development, genome maintenance. [RGS4, RGS5, Scc1]	Johanson-Blizzard Syndrome	[10,27]
Rsp5	PY motifs	Gene expression, mitochondria inheritance, endocytosis. [Rpb1, Gap1, Ste2 etc.]		[28]
Nedd4-1	PY motifs	Cell growth, protein sorting, heart development, immune response. [LAPTM5, IGF-1R, Cbl-b]		[28,32,33]
Nedd4-2	PY motifs	Endocytosis, protein sorting and trafficking. [Epithelial Na+ channel, surfactant protein C, Sgk1]	Liddle Syndrome	[28,30,31]
SCF	Phospho-degron	Cell cycle control, tumor suppression, transcription regulation, signaling. [c-myc, p27, Sic1, Far1, Wee1, Notch]	Breast and endometrical cancers, lymphoma.	[1, 23–25]
APC	D-box, KEN-box	Control of mitotic progression, checkpoint. [Securin, cyclin B Dbf4, Cdc5]	Colorectal and breast cancers	[22,24,26]

The methods reviewed herein (Table 2) aim to study one or multiple of the following facts: substrates directly bind to the degradation machinery; substrates are ubiquitylated; substrates are accumulated in degradation mutants; or increasing levels of substrates may adversely affect cell growth. The advent of genomics and proteomics tools has facilitated the implementation of these methods on a larger scale. Nevertheless, once a candidate substrate is isolated, it is crucial to further validate its regulation by the specific pathway (e.g., ubiquitylation status, protein stability) using several in vivo and in vitro approaches (Figure 2). The pathways dissected below (Table 1) include the Skp1-Cul1-F box (SCF) and anaphase promoting complex (APC) E3s that are key regulators of cell cycle with critical roles in cancers [1, 22–26], the Ubr1 E3 that has been implicated in Johanson-Blizzard Syndrome and angiogenesis [10, 11, 27], and two closely related E3s, Nedd4-1 and Nedd4-2 [28], which have been linked to Liddle syndrome [29–31] and also regulate heart development [32] and axon growth [33, 34]. Lastly, Rsp5, a yeast E3 enzyme that belongs to the Nedd4 family, regulates numerous cellular events, including endocytosis, gene expression and misfolded protein degradation [28]. The selection of these E3s in genome-wide studies likely reflects their physiological significances that have attracted research interests. Although the emphasis here is on the substrate discovery of a specific pathway, the strategies described can be easily reversed to identify ubiquitylation pathway(s) responsible for a particular substrate.

Table 2.

Comparison of various genome-wide screens

Genome-wide methods	Principles	Assay conditions	Species	Ref.
1. Protein-protein interaction	Ubiquitylation enzymes bind substrates	in vitro	yeast, human	[40–42]
2. Protein ubiquitylation	Ub chains can be attached onto substrates	in vitro	yeast, human	[41–44]
3. GFP-fusion based protein level measurement	Substrate accumulation in degradation mutants	in vivo	yeast, human	[46–48]
4. Synthetic enhancement	Impaired substrate degradation can lead to growth retardation	in vivo	yeast	[53,54]

Figure 2.

Screens with protein arrays. Purified yeast or human proteins are spotted onto glass slides. The arrays can be used for screening (a) the interaction with a particular E3 enzyme [40–42], (b) in vitro ubiquitylation with purified components (e.g., E1, E2, E3, Ub) in the presence of ATP [41–43], or (c) semi-in vitro ubiquitylation with cellular extracts [44]. The candidate substrates are identified after a fluorescent scan. Commonly employed target validation approaches are also listed.

Protein-protein interaction-based methods

Substrate ubiquitylation and degradation are mediated by a series of protein-protein interactions between substrates and various factors (e.g. E3s, Ub receptors) [2, 8, 21]. Although many techniques have been utilized effectively to study protein-protein interactions in vivo (e.g. yeast two-hybrid), without modification to trap the bindings, they tend to identify the regulators of substrate proteolysis, but are not efficient for substrate identification due to the transient nature of the interaction between substrates and degradation components in vivo, which quickly leads to destruction. To increase the success rate of the discovery of true substrates, means of trapping the complex between substrate and degradation components can be devised [18, 35]. For example, one could use the substrate-binding domain of an E3 as bait, or employ mutations that retain substrate-E3 binding but specifically disrupt substrate ubiquitylation or degradation, including deletion of the relevant E2 or mutation in the RING motif or HECT domain that are often essential for substrate ubiquitylation. However, these variations have yet to be adapted for genome-wide gene-gene queries to isolate ubiquitylation substrates in vivo.

The recent development of protein arrays containing purified yeast or human proteins circumvents some of these problems, which occur in vivo, and has facilitated the identification of E3 substrates (Figure 2). Commercially available arrays (Invitrogen ProtoArray) have been constructed by spotting either 4,088 glutathione S-transferase (GST)- and His6-tagged yeast Saccharomyces cerevisiae proteins or 8,222 human GST fusion proteins in duplicate on nitrocellulose-coated glass slides that are suitable for in vitro binding studies (Figure 2) [36–39]. Individual query proteins are purified in vitro and biotinylated [40]. Fluorophore-conjugated streptavidin is then used to probe the protein-protein association, which is subsequently visualized by fluorescence scanning (Figure 2a). Using this technique, 124 proteins that bind to the substrate-binding domains of Rsp5 E3 have been isolated, eight of which are likely true considering that they were uncovered in other screens for Rsp5 targets [40], although the majority of these were not examined further for the Rsp5-mediated ubiquitylation, likely due to the limitation of manpower or resource required [40].

The protein array-based method has also been used to identify Rsp5 substrates in S. cerevisiae by another group [41]. Full length Rsp5 was labeled with Alexa 647 and 155 Rsp5-binding proteins were isolated. Roughly one-third of these proteins were identified in a separate ubiquitylation-based assay for Rsp5 substrates [41]. Some candidates were also uncovered in the previously mentioned screen [40]. The substrates of mammalian Rsp5 homologues, human Nedd4-1 and Nedd4-2, have also been investigated [42]. Approximately 80 candidates were identified for each, half of which were isolated using an in vitro ubiquitylation-based screen. However, the precise success rate of this type of screen for substrates is not clear because the hits were not pursued further.

Protein microarrays offer obvious advantages since the collections of these proteins are comprehensive, unbiased, and include proteins of low abundance. In addition, these arrays are easy to screen and analyze. Protein microarrays also have limitations: some proteins (e.g. membrane proteins) cannot be easily expressed and purified; other proteins may be co-purified and present on the array; the attachment of epitopes may influence their binding properties; and the substrates that require cofactors for Rsp5-binding would be missed. Furthermore, the in vitro setting may be too permissive and could lead to many false positives. As alluded to earlier, it is crucial to carry out other assays to establish whether the candidate is regulated by the specific E3 in vivo.

In vitro ubiquitylation-based approaches

The outcome of substrate modification is ubiquitylation; therefore, one could monitor the addition of Ub onto a substrate. Several well-defined in vitro ubiquitylation systems have been employed to survey whether a protein can be ubiquitylated and therefore serve as a potential substrate of the ubiquitylation factors (e.g. E2, E3) in the reaction (Figure 2b). This rationale has been applied in screening for substrates of Rsp5 in yeast [41, 43], Nedd4-1 and Nedd4-2 in humans [42] and the APC complex in humans [44].

To uncover Rsp5 targets, yeast protein arrays have been incubated with a ubiquitylation reaction mixture containing budding yeast E1 enzyme Uba1, an E2 (Ubc4), FITC-labeled ubiquitin, Rsp5 (E3) and ATP (Figure 2b) [41]. Proteins ubiquitylated in the reaction were detected by the fluorescent signal over the background. Forty out of 150 proteins identified were selected as the “high confidence” list, including 12 known Rsp5 substrates. Furthermore, 15 candidates that were chosen for further analysis by the in vivo and in vitro ubiquitylation assays appeared to be true Rsp5 substrates because the generation of the ubiquitylated species required Rsp5 [41], thereby validating this strategy as a powerful tool for substrate discovery. Using the same approach, substrates of the human Rsp5 homologs Nedd4-1 and Nedd4-2 [28] and the rat homolog rNedd4-1 have been investigated in a ubiquitylation reaction containing purified E1, E2 (UbcH5), the respective E3, and the human proteome arrays (Invitrogen) [42]. Respectively, 154, 107, and 92 hits for these E3s were obtained. Seventeen candidates were analyzed further by an in vivo ubiquitylation assay, in which ~70% were confirmed as Nedd4 substrates [42]. Although the Nedd4 family is well-conserved, the results suggest that each homolog has a distinct set of substrates.

A similar Rsp5 screen has been performed using yeast protein arrays [43]. Interestingly, there are only 5 common hits in the ‘top 100’ hit list from the two screens that both searched for Rsp5 targets [41, 43]. The largely distinct sets of Rsp5 substrates identified are likely caused by the different E2 enzymes (i.e. UbcH10 instead of Ubc4) employed [41, 43], which are known to influence substrate selectivity [2, 8, 21]. Further analysis of the top 86 candidates has revealed that 28 show genetic interaction with Rsp5. Fifty-six of the top 86 candidates were characterized using the in vitro ubiquitylation system and the in vivo ubiquitylation assay, which uncovered 8 in vivo Rsp5 substrates, only one of which was present in a previous screen [41]. The in vitro ubiquitylation-based screen suffers from many of the same disadvantages (e.g. epitope interference, missing critical co-factors) mentioned previously for protein-protein interaction-based methods. Nevertheless, the ubiquitylation-based screen appears to be more effective in substrate discovery than the interaction-based approach [41, 42], mainly because a low affinity interaction may not be a restriction for ubiquitylation and, further, multi-ubiquitylation might enhance the signal. Although these studies validate the use of in vitro ubiquitylation assay based strategy, they also highlight several unique issues associated with it, such as the requirement for enzymatic activity of various proteins, and the combination of specific components employed in the reaction. Different substrates for the same E3 may use distinct sets of E2s and other co-factors in vivo, which, if not identified, can present a challenge in designing the in vitro reaction.

A variation of this in vitro ubiquitylation approach that partially addresses this dilemma of uncertain co-factors required is a semi-in vitro method developed recently [44], which uses cellular extract that is competent for the ubiquitylation of substrates, such as Securin, in vitro (Figure 2c) [44]. To look for novel targets of the APC E3 that are ubiquitylated upon release from the mitotic checkpoint, the human protoarray (Invitrogen) was incubated with both the checkpoint-released and the APC-inhibited extracts (supplemented with the APC-inhibitor Emi1), and substrate ubiquitylation was subsequently monitored using an anti-polyubiquitin (FK1) antibody and a Cy3-labeled secondary antibody [44]. The differential signals obtained from these extracts identified ~132 candidates, including 11 of the 16 known APC substrates present on the array. Furthermore, 7 out of 10 candidates assayed were subsequently validated by the ubiquitylation or degradation assay [44].

The use of a functional extract instead of a defined ubiquitylation reaction has several advantages. It eliminates the need for laborious protein purification, especially for E3s that are difficult to purify or handle. Additionally, cellular extracts closely reflect the physiological conditions inside the cell, such as specific cell cycle or differentiation stages, or disease states that are hard to replicate by the simple in vitro reactions. However, the nature of this assay demands a clear functional distinction between two extracts, with or without potent E3 activity.

Although they are powerful and effective, these in vitro ubiquitylation assays bear an inherent caveat that the results could be artifacts caused by the in vitro set-ups. An in vivo ubiquitylation detection method has been developed previously to follow real-time ubiquitylation of β–arrestin [45], but this method has not been used in substrate identification, most likely due to the technical challenges involved in capturing substrate ubiquitylation in vivo.

In vivo protein-level assays

Ub-proteasome substrates are accumulated in cells with compromised ubiquitylation or degradation activity. Ideally, then, one strategy for substrate identification is to directly compare the levels of specific proteins in wild-type and degradation mutant cells. Traditionally, an antibody is required to determine the protein level. The discovery of GFP and the available collections of GFP-tagged ORF (open reading frame) libraries, both genomic- and plasmid-based, have paved the way to comprehensively track proteins in vivo under various conditions (Figure 3). To identify substrates of the SCF ubiquitin ligase with the F-box protein Grr1 (SCF^Grr1), an ingenious approach has been used to compare the levels of ~4000 yeast proteins in wild-type vs. grr1 mutant cells (Figure 3a) [46]. A genome-wide collection of 3926 yeast genes was employed, each of which was appended with a carboxy-terminal GFP at its endogenous locus. Using a high-throughput microscope, the GFP signals were compared in the wild-type strain and the SCF^grr1Δ mutant. Of 106 hits obtained from the primary screen, 7 proteins were identified as novel ubiquitylation targets of SCF^Grr1. In-depth analysis further revealed that Grr1-mediated degradation Pfk27 and Tye7 are important for the glycolytic-gluconeogenic switch [46].

Figure 3.

GFP-fusion based screens. (a) Examples of a mixture of wild-type and grr1 (E3) mutant cells expressing the same protein visualized under the microscope. An ingenious strategy was employed to differentiate wild-type (ring colored in green) from E3 mutant cells (red ring) under the microscope [46]. Left panels show that the specific protein expressed is not a target of the E3 because its levels are similar in both wild-type and mutant cells. Right panels present an example of a candidate substrate since its levels are higher in E3 mutants. (b) In a GPS screen [47, 48], mammalian cells transfected with tagged proteins are separated into different pools by FACS sorting, then treated with or without the dominant negative fragment of E3. Candidate substrates are identified by sequencing or microarray deconvolution [47, 48].

For mammalian cells, a method, termed Global Protein Stability (GPS), has been developed to identify SCF^Cul1 (E3) targets (Figure 3b) [47, 48]. To facilitate the large-scale screen, a retroviral reporter library was constructed, in which an enhanced allele of GFP (EGFP) was fused in-frame to ~8000 full-length human ORFs. To provide an internal control, the red fluorescence protein DsRed and the EGFP-tagged protein were under the control of the same promoter, and the separate translation of these proteins was ensured by an internal ribosome entry site. In this setting, a test protein would have an EGFP/DsRed signal ratio that is representative of its stability [47, 48]. GPS profiling combined with either microarray deconvolution or a substrate enrichment scheme based on fluorescence activated cell sorting (FACS) permitted analysis of the cells treated with or without a dominant negative fragment of SCF^Cul1 [48]. Over 300 potential SCF^Cul1 substrates were isolated, and, among 66 candidates that were further analyzed, 6 known and 25 novel SCF^Cul1 targets were identified, thus suggesting that this method is both effective and robust.

One obvious advantage of these screens is that they are performed in vivo, which allows proteins to remain in their physiological environment. The screen can be easily applied to different cell types or to cells subjected to different stresses. These features make GPS screen an effective approach to probe ubiquitylation-related defects associated with human diseases [49]. However, proteins present in low-abundance may result in GFP signals that are below the limit of detection, and as a result, some substrates may be overlooked. In addition, inefficient E3 knockdowns and deficiencies in the ORF libraries may contribute to the rate of false-negatives.

Function-based synthetic enhancement screens

Because large amounts of undegradable substrates could have dire cellular consequences (e.g. cell death, stress sensitivity), the use of a synthetic dosage lethality (SDL) screen [50, 51], which partly depends on the substrate’s in vivo function, for ubiquitylated substrate discovery. The premise of SDL is that increased levels of a protein have little toxic effect on the growth of a wild-type strain but may cause growth retardation or even lethality in a mutant strain. Previous studies suggest that this strategy can be adapted to isolate substrates of the Ub/proteasome system; for example, overexpression of several cell cycle proteins (e.g. Clb2, Scc1) in proteolysis-deficient cells (e.g. apc mutant, ubr1 mutant) have led to extremely slow cell growth [9, 52].

To unravel the function of the Ub-binding proteins Rad23 and Dsk2 [6, 8], more than 5000 yeast genes fused to both GST and His6 tags were separately introduced in wild-type or rad23Δ dsk2Δ mutant cells [53]. The growth of wild-type vs. rad23Δ dsk2Δ cells that expressed each of these yeast genes from a strong, regulatable GAL1 promoter was compared. The expression of these genes was repressed in glucose-containing media, but induced in galactose-containing media. Haploids carrying these plasmids were pinned onto both SD (glucose media) and SG (galactose media) plates. Subsequently, the colony sizes observed on these plates were scored. Eleven genes that cause slower growth upon their overexpression in rad23Δ dsk2Δ cells were confirmed, suggesting that proper functioning of Rad23 and Dsk2 is required for efficient pheromone response, transcription, amino acid metabolism and DNA damage response. Two proteins identified by the screen have been shown to be proteolytic substrates of Dsk2, thus validating the large-scale SDL screen as a new strategy for identifying substrates of a specific degradation pathway [53]. More recently, the SDL screen identified a peroxisomal protein Pex29 as a ubiquitylation substrate of Ufd2 [54], a Ub chain elongation factor [55].

SDL is advantageous as a screening tool for proteolytic substrate identification because the assay is performed in vivo and the protein expression levels are enhanced by a strong promoter (i.e. GAL1). However, like other approaches for target discovery, it has a few limitations: namely, the readout is an indirect measurement of protein degradation, the redundancy of various substrate delivery pathways may mask some substrates, and genes that lead to severe growth defects in wild-type cells with increased dosage are likely to be excluded from such analysis. Although the SDL screen described was performed in yeast [53], the creation of ORF overexpression collections would extend the use of SDL to mammalian cells [56, 57]. Nevertheless, even in the case that the identified proteins are not direct proteolytic substrates, the genetic interactions obtained still pinpoint the cellular events under the regulation of a particular degradation pathway.

Conclusions and outlook

The identification of physiological targets of proteolytic pathways has proven to be quite challenging. Although many strategies have been developed for this purpose, there is no ‘one-size-fits-all’ method, and there is still much to be learned about each degradation pathway. The approaches described above all have their own advantages and limitations that lead to both false-positives and false-negatives. Given different levels of robustness and stringency, these screening strategies are complementary to each other. In several cases, two of these strategies were used in combination and provided some overlapping coverage, which accelerated the pace of discovery of bona fide substrates [41–43]. For example, 86 candidate Rsp5 substrates initially isolated by the Rsp5-dependent in vitro ubiquitylation approach were overexpressed in rsp5 mutant cells to examine SDL or suppression [43]. Whereas 7 of the 28 candidates that exhibited genetic interaction were found to be ubiquitylated in a Rsp5-dependent manner in vivo, only one of the other 28 candidates selected that did not show genetic interaction appeared to be a Rsp5 substrate in vivo [43], indicating that the combination of in vitro ubiquitylation assay and genetic interaction approach (e.g. SDL) leads to a significantly higher success rate of substrate identification. Curiously, although genome-wide measurements of protein half-life or modifications have been carried out [58], they have yet to be adapted to identify specific targets of individual ubiquitylation pathways. A more recent development of targeted protein analysis using mass spectrometry [59, 60] holds promise for determining specific protein levels in cells under various conditions, much more easily than using other currently available methods. The power of this targeted proteomics strategy lies in its ability to quantitatively measure specific proteins that are expressed in vivo at very low levels (<50 copies per cell), which has often escaped the detection by other methods [59, 60].

Besides the technical challenges, another major barrier of genome-wide substrate identification is that the large number of candidates isolated in the initial screen is often overwhelming to a single laboratory that has discovered them. Further validation of many substrates revealed by the genome-wide screens came from other groups studying those proteins and the processes they are involved in. For instance, Sna3, a candidate Rsp5 target isolated in two separate genome-wide screens [40, 41], was later demonstrated to be an authentic Rsp5 substrate in vivo by several groups [61–64]. Therefore, the comprehensive understanding of the Ub system also calls for close collaborations among the laboratories with different expertise in proteomics, specific ubiquitylation pathways, and various cellular events under the regulation of Ub.

Functions of many ubiquitylation pathways remain elusive. For example, what is ubiquitylation target of WWP-1 E3 relevant for longevity extension [65]? What is the specific role of Brca1 E3-mediated ubiquitylation in breast cancer tumorigenesis [66]? Physiological roles of many Ub ligases have not been characterized. Furthermore, cells also employ several Ub-like protein modifications, including sumoylation, ISGylation and urmylation, in a non-proteolytic manner to regulate cellular localization and signaling [2, 7, 67–69]. However, the biological roles of these Ub-like molecules are far from clear because many substrates of these post-translational modifications remain to be identified. The strategies described in this review can be easily adapted to those studies as well. Hence, continuing refinements and future technical innovation for ubiquitylation target identification will be well worth the effort considering we have just started to scratch the surface of the Ub and other Ub-like systems.

Intracellular protein homeostasis is essential for the proper functioning of the cell [2, 3, 10]. The Ub/proteasome system handles the majority of regulated proteolysis and plays an important role in nearly every area of cell biology. Defects in the Ub/proteasome system can lead to diseases ranging from cancers to neurodegenerative disorders [3–5]. For example, mutations in Brca1 and Parkin - two E3 components of the Ub system - are associated with breast cancer and hereditary Parkinson’s disease [66, 70–72]. Importantly, inhibiting the activity of the proteasome has proved to be an effective therapeutic means in the management of multiple myeloma [3, 22], which further suggests that modulating specific pathways may have better therapeutic potentials [1, 3, 5, 24]. Hence, the study of the functional relationship between a degradation pathway and its substrates not only is important for understanding the biological function of proteolysis, but could also potentially uncover novel drug targets or reveal effective approaches to combating human diseases.