BioID-based Identification of Skp Cullin F-box (SCF)β-TrCP1/2 E3 Ligase Substrates*

The identification of ubiquitin E3 ligase substrates has been challenging, due in part to low-affinity, transient interactions, the rapid degradation of targets and the inability to identify proteins from poorly soluble cellular compartments. SCFβ-TrCP1 and SCFβ-TrCP2 are well-studied ubiquitin E3 ligases that target substrates for proteasomal degradation, and play important roles in Wnt, Hippo, and NFκB signaling. Combining 26S proteasome inhibitor (MG132) treatment with proximity-dependent biotin labeling (BioID) and semiquantitative mass spectrometry, here we identify SCFβ-TrCP1/2 interacting partners. Based on their enrichment in the presence of MG132, our data identify over 50 new putative SCFβ-TrCP1/2 substrates. We validate 12 of these new substrates and reveal previously unsuspected roles for β-TrCP in the maintenance of nuclear membrane integrity, processing (P)-body turnover and translational control. Together, our data suggest that β-TrCP is an important hub in the cellular stress response. The technique presented here represents a complementary approach to more standard IP-MS methods and should be broadly applicable for the identification of substrates for many ubiquitin E3 ligases.

More than 600 putative ubiquitin E3 ligases are encoded in the human genome (1,2). Although a number of these proteins are known to play critical roles in human health (1)(2)(3)(4), the specific biological functions-and substrates-of most E3s remain poorly characterized. The identification of E3 sub-strates has been difficult in part because: (1) ligase -substrate interactions are often of low affinity (generally in the high nM to microMolar (M) range) and/or of a transient nature; (2) many substrates are subjected to rapid proteasomal degradation and are therefore not available for detection; (3) the human ubiquitome is extremely complex, and; (4) many substrate proteins are localized to poorly soluble cellular compartments, making their isolation and identification by standard immunoprecipitation (IP)-based techniques extremely challenging (1)(2)(3)(4).
Methods such as protein chip (5) and yeast two-hybrid screening (6) have been used to identify a limited number of E3-substrate interactions. However, these methods are not conducted in live mammalian cells, and may not be generally applicable for the identification of substrates of the hundreds of unique multi-protein E3 complexes (e.g. SCF, APC, VHL, etc.) or the identification of E3-substrate interactions that are dependent on specific types of post-translational modifications. Several putative inhibitors of apoptosis substrates were identified using the recently described NEDDylator technique (7), in which the E2 protein UBC12 (UBE2M) is fused to the E3 ligase of interest, allowing for the conjugation of the ubiquitinlike protein NEDD8 to substrates. Global protein stability profiling (8) and quantitative mass spectrometry methods (9) have also been used successfully to identify E3 targets. However, because of their cost and/or complexity, and the challenges posed by the extremely large size of the human ubiquitome, these methods have not been widely adopted to date.
Proximity-based biotinylation, or BioID 1 (10), is a new method developed for the characterization of protein-protein interactions in living cells. Briefly, a protein of interest is fused in-frame with an E. coli biotin conjugating enzyme mutant (BirA R118G, or BirA*). The BirA* moiety can efficiently activate biotin, but exhibits a reduced affinity for the activated molecule (11); biotinoyl-AMP thus simply diffuses away from BirA* and reacts with nearby amine groups -including those present on lysine residues in neighboring polypeptides. Following cell lysis, biotinylated proteins can be affinity purified using streptavidin and identified using mass spectrometry ( Fig. 1A). Since interactors are covalently modified with biotin, robust lysis conditions can be used to solubilize polypeptides localized to poorly soluble cellular compartments (10,(12)(13)(14)(15)(16). Moreover, since this method does not require that proteinprotein interactions be maintained post-lysis, weak and/or transient interactors may also be identified. We reasoned that BioID may be exploited to capture ubiquitin E3 ligase substrates and tested this notion here.
The human beta transducin repeat-containing polypeptides ␤-TrCP1 (FBXW1) and ␤-TrCP2 (FBXW11) are evolutionarily conserved paralogous F-box proteins sharing Ͼ80% amino acid sequence identity and Ͼ90% homology, and act as substrate recognition components of SCF (Skp1-Cullin-Fbox) complexes (17). A number of ␤-TrCP substrates have been well documented, implicating these ligases in numerous biological functions, including regulation of the NFB, Hippo, Wnt and Hedgehog signaling pathways (18 -21). Some ␤-TrCP targets harbor the sequence DSGX(n)S or variants thereof. Phosphorylation of this sequence is required for SCF ␤-TrCP1/2 binding and thus regulates the half-life of these "phosphodegron"-containing proteins (21,22). However, other bona fide SCF ␤-TrCP1/2 substrates contain highly degenerate or non-canonical degrons, which are thought to mediate constitutive turnover (21). A single, simple linear sequence motif that could predict ␤-TrCP binding has thus not been defined.
Here we demonstrate that BioID performed on cells treated with the proteasome inhibitor MG132 can recover many of the previously characterized substrates and stable interactors of ␤-TrCP1/2. Using semi-quantitative mass spectrometry, we identify and validate a number of new substrates, linking these well-studied E3 ligases to several new biological functions. The method used here is simple, scalable, and should be broadly applicable for the identification of substrates for many other E3s.

EXPERIMENTAL PROCEDURES
Cloning-All PCR reactions were performed using Q5 DNA polymerase (NEB) according to manufacturer's instructions. Primers, destination plasmids, cloning sites and DNA templates are listed in supplemental Table S4. PCR products were digested with AscI and NotI (NEB) and ligated into the destination plasmid using T4 DNA ligase (NEB) following manufacturer's instructions. The PPP1R15B S459 -466A phosphodegron mutant was generated by PCR-driven overlap extension (23) using WT PPP1R15B external primers and the overlapping internal primers listed in supplemental Table S4.
FLAG Affinity Purification-For FLAG pulldowns, 10 ϫ 150 cm 2 dishes of FLAGBirA*-␤-TrCP1 or FLAGBirA*-␤-TrCP2 293 T-REx cells were treated with tetracycline and biotin as above (5 plates were treated with 5 M MG132), scraped into PBS, pooled, washed twice in 25 ml PBS and collected by centrifugation at 1000 ϫ g for 5 min at 4°C. Cell pellets were stored at Ϫ80°C until lysis. The cell pellet was weighed and 1:4 pellet weight/lysis buffer (by volume) was added. Lysis buffer consisted of 50 mM HEPES-NaOH (pH 8.0), 100 mM KCl, 2 mM EDTA, 0.1% Nonidet P-40, 10% glycerol, 1 mM PMSF, 1 mM DTT and 1:500 protease inhibitor mixture (Sigma-Aldrich, St. Louis, MO). Upon resuspension, cells were incubated on ice for 10 min, subjected to one additional freeze-thaw cycle, then centrifuged at 27000 ϫ g for 20 min at 4°C. Supernatant was transferred to a fresh 15 ml conical tube and 250U Turbonuclease (Accelagen) plus 30 l packed, pre-equilibrated FLAG-M2 agarose beads (Sigma-Aldrich) were added. The mixture was incubated for 2 h at 4°C with endover-end rotation. Beads were pelleted by centrifugation at 1000 rpm (100 ϫ g) for 1 min and transferred with 1 ml of lysis buffer to a fresh centrifuge tube. Beads were washed once with 1 ml lysis buffer and twice with 1 ml ammonium bicarbonate (ammbic) rinsing buffer (50 mM ammbic pH 8.0, 75 mM KCl). Elution was performed by incubating the beads with 150 l of 125 mM ammonium hydroxide (pH Ͼ11). The elution step was repeated twice more, and the combined eluate centrifuged at 15000 ϫ g for 10 min, transferred to a fresh centrifuge tube and lyophilized. Following overnight trypsin digestion (as above), peptides were resuspended in 0.1% formic acid and 1/6th of the sample was analyzed per MS run.
Mass Spectrometry-Liquid chromatography (LC) analytical columns (75 m inner diameter) and precolumns (150 m ID) were made in-house from fused silica capillary tubing from InnovaQuartz (Phoenix, AZ) and packed with 100Å C 18 -coated silica particles (Magic, Michrom Bioresources, Auburn, CA). LC-MS/MS was conducted using a 120 min reversed-phase buffer gradient running at 250 nl/min (column heated to 40°C) on a Proxeon EASY-nLC pump in-line with a hybrid LTQ-Orbitrap velos mass spectrometer (Thermo Fisher Scientific). A parent ion scan was performed in the Orbitrap, using a resolving power of 60000. Simultaneously, up to the twenty most intense peaks were selected for MS/MS (minimum ion count of 1000 for activation) using standard CID fragmentation. Fragment ions were detected in the LTQ. Dynamic exclusion was activated such that MS/MS of the same m/z (within a 10 ppm window, exclusion list size 500) detected three times within 45 s were excluded from analysis for 30 s. For protein identification, .raw files were converted to the .mzXML format using Proteowizard (24), then searched using X!Tandem (25) against Human RefSeq Version 45 (containing 36113 entries). Search parameters specified a parent MS tolerance of 15 ppm and an MS/MS fragment ion tolerance of 0.4 Da, with up to two missed cleavages allowed for trypsin. Oxidation of methionine was allowed as a variable modification. Data were analyzed using the trans-proteomic pipeline (26) via the ProHits 2.0.0 software suite (27). Proteins identified with a ProteinProphet cut-off of 0.85 (corresponding to Յ1% FDR) were analyzed with SAINT Express v. 3.3 (28,29). For BioID, 24 control runs were used for comparative purposes; eight runs of a BioID analysis conducted on cells expressing the FlagBirA* tag only, eight runs of BioID conducted on untransfected 293 T-REx cells, and eight runs from a BioID analysis conducted on an unrelated bait protein. In each case, four runs were conducted on untreated cells and four runs were conducted on cells treated with MG132, as above. The 24 controls were collapsed to the highest four spectral counts for each hit. Similarly, for Flag IP-MS analysis, 24 control runs were conducted; FLAG IP-MS analyses of three unrelated FLAGBirA*tagged proteins expressed in 293 T-REx cells, in the presence and absence of MG132. The 24 control runs were collapsed to the highest four spectral counts for each hit. All data are publically available and have been uploaded to the MassIVE archive (massive.ucsd.edu) ID: MSV000079032. Annotated spectra are available for inspection via the MS-Viewer tool (30) of Protein Prospector (prospector2.ucsf.edu; see supplemental Table S1 for all associated file keys).
Interactor Classification-Based on four independent MS runs (two technical replicates and two biological replicates for each bait, in the presence and absence of MG132), bona fide interactors were defined as high confidence protein identifications (ProteinProphet p value Ͼ0.85) displaying a SAINT score Ն0.75. Histones were removed manually. Fold-change was calculated as: log 2 (1ϩ spectral count sum in the presence of MG132)/(1ϩ spectral count sum in the absence of MG132). Statistical significance of the difference between the spectral counts detected in samples prepared from cells cultured with and without MG132 was calculated using Student's t test. Hits considered as putative substrates display a log 2 fold change Ն 1 (Table I) and p value Յ 0.05. Gene ontology (GO) category assignments detailed in supplemental Table S3.
Immunofluorescence, Image Acquisition, and Processing-Untransfected HeLa cells or HeLa stable cell lines expressing FLAG-SUN2 were grown on poly-L-Lysine (Sigma) coated coverslips, fixed with 4% formaldehyde for 15 min, and washed in PBS with 0.1% Triton X-100. Cells were blocked in 5% BSA in PBS for 30 min before incubating in PBS with 1:1000 anti-FLAG M2 (Sigma), anti-Lamin B (Abcam) or anti-SUN2 (Sigma) primary antibodies for 1h at RT. Antimouse Alexa 488 and Streptavidin-Alexa594 (Life Technologies) were used at 1:1000 and 1:5000, respectively, and incubated at RT for 1 h. After removing the solution, cells were incubated with 1 g/ml of 4Ј,6-diamidino-2-phenylindole (DAPI) in PBS for 5 min. After washing with PBS 3 ϫ 5 min, coverslips were mounted with ProLong Gold Antifade (Thermo Fischer Scientific). Cells were imaged using a Plan-Apo 60X oil lens, NA 1.40 on an Olympus FV1000 confocal microscope (zoom factor between 3 and 5; Olympus America, Melville, NY). Images were processed using the Volocity Viewer v.6 and assembled using Adobe Illustrator CS5 (Adobe System Inc., San Jose, CA).
Live Cell Imaging, P-body and Stress Granule Quantification-GFP and mCherry constructs were generated by Gateway cloning into an N-or C-terminal pDEST pcDNA5/FRT/TO-eGFP, or an N-terminal pDEST pcDNA5/FRT/TO-mCherry vector. Clone accession numbers are shown in supplemental Table S4. Stable Flp-In HeLa T-REx cell line pools were generated as described in (32). The PATL1, DCP1A, EIF4A1, PABPC1 and G3BP1 genes were cloned from the following clone accession number templates: AM392959, DQ893260, DQ893044, DQ893508, and DQ893058. For live-cell studies, cultured HeLa cells transiently transfected with pcDNA3 Cherry-SUN2 or stably transfected with the GFP or mCherry-tagged P-body or stress granule markers were transfected with siRNA control or siRNA ␤-TrCP1/2 in 24-well plates, incubated 48 h, then seeded into Lab-Tek chambers (Thermo Fisher) and treated for 24 h with 1 g/ml tetracycline to induce marker expression. HeLa cells were imaged without additional treatment. P-body and stress granule marker expressing cells were stained with 1 g/ml of Hoechst 33342 dye (Fisher Scientific) for 15 min at 37°C. Cells were washed twice with warm PBS before incubating 45 min in phenol red-free complete media with 500 g/ml arsenite (Sigma Aldrich). Cells were incubated in a Chamlide TC stage incubator (LCI, Seoul, Korea) during acquisition (37°C; 5% CO 2 ). The Volocity 6 measurement tool was used to calculate the number of P-bodies or stress granules per cell. Size and brightness parameters were adjusted specifically for each bait protein to delineate the relevant structures, and identical acquisition and calculation parameters were applied under all conditions for a given fluorescent marker (i.e. fixed exposure time, sensitivity and gain, and constant size and intensity thresholds for nuclei, P-bodies and stress granules). Volocity 6 software was used to generate the stitched composite image from 4 ϫ 4 adjacent fields acquired with 10% overlap at 60ϫ magnification, or 3 ϫ 3 adjacent fields acquired with 10% overlap at 20X magnification. Four stacks (500 nm z axis distance spread) were acquired for each field. Nuclei were defined as Hoechst/DAPI positive objects with a minimal volume of 150 m 3 (to avoid counting micronuclei). All cells were seeded at a comparable density (to reach 80% at analysis time) and Ն100 cells were acquired per condition. Five independent experiments were conducted for P-body quantitation, and six independent experiments were conducted for stress granules. Cells treated with siRNA control or siRNA ␤-TrCP1/2 were compared and statistical significance assessed using Student's t test. supplemental Figs. S3B and S4 display the stitched field, extended focus images used for the analysis, and indicate a homogeneous distribution of P-bodies and stress granules among siRNA control and siRNA ␤-TrCP1/2-treated cells.
Flag Co-IP-FLAGBirA*-␤-TrCP1 or FLAGBirA*-␤-TrCP2 293 T-REx cells were transfected at 50 -70% confluence with either pcDNA3 3xHA-PPP1R15B wt or pcDNA3 3xHA-PPP1R15B S459 -466A using the Polyjet reagent, following manufacturer's instructions. 16 h posttransfection, each 10 cm plate was split 1:4, cells were treated with 1 g/ml tetracycline for 24 h and 10 M MG132 for 2 h prior to harvesting. Each plate was washed once with PBS, then cells were scraped and re-suspended in 5 ml PBS with 2 mM EDTA. Cells were pelleted at 2000 rpm for 2 min, resuspended in 1 ml PBS, and transferred to an Eppendorf tube, pelleted again at 2000 rpm for 2 min, PBS was removed, and the cell pellet was snap frozen. Cell pellets were lysed in 1 ml TAP lysis buffer (50 mM Hepes-NaOH pH 8, 150 mM NaCl, 10% glycerol, 2 mM EDTA, 0.1% IGEPAL, 2 mM DTT, 1:500 protease inhibitor (Sigma), 10 mM NaF, 0.25 mM sodium orthovanadate, 50 mM ␤-glycerol phosphate) for 1 h on an end-over-end rotator at 4°C, and centrifuged at 20000 rpm for 15 min at 4°C. Five percent of the cleared lysates was set aside to serve as input, and the remaining lysate was incubated with 20 l packed anti-FLAG M2 Affinity Gel beads (Sigma Aldrich A2220) on an end-over-end rotator overnight at 4°C. Beads were pelleted at 2000 rpm for 2 min and washed 5 times with TAP lysis buffer before adding 60 l boiling Laemmli buffer. Samples were boiled and spun at 15000 rpm to pellet beads prior to loading supernatant onto SDS gel.
ER Stress Response Assay-293 T-REx Flp-In cells stably expressing FLAG-PPP1R15B WT or FLAG-PPP1R15B S459 -466A were seeded into a six well plate. Cultures were treated with tetracycline (1 g/ml), along with thapsigargin (1 M) or tunicamycin (1 g/ml) for 24 h. Cells were collected, washed in 1x PBS and lysed in 100 l TAP lysis buffer (see above; n.b. wash and lysis buffers also contained the ER stress drugs at the same concentrations used in culture). Cell lysates were centrifuged at 20,000 rpm for 15 min at 4°C. Supernatant was removed and protein content quantified for normalization. Phospho-eIF2␣ (Ser51) and eIF2␣ antibodies (Cell Signaling Technology, Beverly, MA) were used at 1:1000 dilution in TBST.
Cell Viability Assay-293 T-REx Flp-In cells stably expressing FLAG-PPP1R15B WT or FLAG-PPP1R15B S459 -466A were seeded into a 96-well plate at a density of 4000 cells per well. Tetracycline (1 g/ml, 72 h) was added to the cultures for 24 h to induce protein expression, then cells were treated with the indicated concentration of tunicamycin (48 h). Each condition was performed in triplicate. To assess viability, 20 l of 1 g/ml MTT (Sigma) dissolved in sterile PBS was added to 100 l of media in each well and incubated for 2 h at 37°C until purple precipitate was visible. 120 l of Stop Reagent (0.04N HCl and 10% Triton-X 100 in isopropanol) was added to each well and the media resuspended to dissolve the precipitate. Plates were read at 570 nm and 630 nm; OD 570 -OD 630 values for each well were determined and the triplicate measurement was averaged for each condition. Each average value was normalized to the untreated sample of the corresponding cell line, and standard error calculated. This experiment was repeated three times, with a representative experiment displayed in Fig. 4F.

RESULTS
BioID Identifies Known and Novel SCF ␤-TrCP1/2 Substrates-SCF ␤-TrCP1 and SCF ␤-TrCP2 target substrates for 26S proteasome-mediated degradation. We reasoned that, using the BioID method, stable ␤-TrCP1/2 interacting partners (such as other components of the SCF complex) should be labeled with biotin to a similar extent in the presence and absence of MG132, whereas substrates should be stabilized when proteasome activity is impaired, resulting in increased abundance in the streptavidin pulldown. A simple semiquantitative mass spectrometry approach could then be employed to identify these putative substrates.
Previous reports have indicated that N-terminally tagged ␤-TrCP maintains proper function (22,(33)(34)(35)(36), and our own preliminary BioID analysis indicated that N-terminal FLAG-BirA*-tagged ␤-TrCP1 recovered more stable interactors (e.g. SCF complex components) and previously reported substrates than a C-terminally tagged protein (data not shown). A FLAGBirA* tag was thus fused to the N terminus of ␤-TrCP1 and 2 (to generate FLAGBirA*-␤-TrCP1 and FLAGBirA*-␤-TrCP2), and these proteins were stably expressed in 293 T-REx Flp-In cells. Two independent stable cell pools were created for each bait protein. The FLAGBirA*-␤-TrCP polypeptides were expressed in a tetracycline-regulated manner, and biotinylated protein species were detected by Western analysis following the addition of biotin to the culture media (Fig. 1B). Cells were lysed using a modified RIPA buffer (see Experimental procedures), followed by sonication and nuclease treatment to maximize the solubilization of proteins localized to poorly soluble intracellular locations, or associated with membranes, RNA or chromatin. Biotin-labeled proteins were captured using a streptavidin-Sepharose matrix. After washing, streptavidin-bound proteins were treated with trypsin, and the eluted peptides identified using nanoflow liquid chromatography-electrospray ionization-tandem mass spectrometry (nLC-ESI-MS/MS). Untransfected cells, cell pools expressing the FLAGBirA* tag alone, and cells expressing an unrelated bait protein were subjected to the same analysis to control for endogenously biotinylated proteins and   Using the ProHits (27) system, MS data were analyzed with the X!Tandem database search algorithm (25), and the resulting peptide identifications subjected to ProteinProphet (37) analysis. Finally, the data were subjected to Significance Analysis of INTeractomes (SAINT) (28,29) to identify bona fide SCF ␤-TrCP1/2 interacting partners. Polypeptides identified with a ProteinProphet confidence value Ն0.85 and assigned a SAINT score Ն0.75 are listed in supplemental Table S1.
To determine how the BioID method compares to a standard IP approach, a standard FLAG tag-based immunopurification (FLAG IP-MS; see Experimental procedures) was also conducted on FLAGBirA*-␤-TrCP1/2 (as above, in the pres-ence and absence of MG132) and the associated proteins identified using MS/MS. In this case, 127 high confidence ␤-TrCP1/2 interactors were identified ( Fig. 1D and supplemental Table S1). 41 proteins displayed a substrate profile, 28 of which (ϳ70%) were identified with a similar profile in BioID ( Fig. 1D and Table I). 17 previously reported substrates and substrate interacting partners were identified with this method. Thirteen additional proteins with a substrate profile were detected that were not also identified in BioID (Fig. 1D, Table I and supplemental Table S1). Together, these data suggested that BioID could represent a valuable complement to standard IP-based approaches for the identification of E3 targets.
Consistent with previous reports, many of the proteins displaying a substrate profile in both the BioID and FLAG IP-MS data sets are involved in cell cycle regulation, Wnt, NFB, and Hippo signaling (Fig. 1E). Notably, however, several specific functional categories -all of which are associated with less soluble cellular compartments and not previously linked with ␤-TrCP function -were enriched in the BioID data set, including e.g.; cytoskeleton, vesicle trafficking and RNA regulation (FLAG: 5, 2, 3 versus BioID: 10, 5, and 8 identified polypeptides, respectively; Fig. 1E).
BioID Identifies ␤-TrCP Substrate Candidates-To begin to characterize the new putative SCF ␤-TrCP substrates, a number of candidates (selected to represent a variety of different biological functions, and localized to different intracellular compartments) were expressed as epitope (FLAGBirA*, FLAG or 3xHA)-tagged proteins in 293 T-REx cells. Cultures were treated with cycloheximide (CHX) to inhibit protein synthesis, and protein half-lives were monitored in the presence of: (1) a single siRNA targeting both ␤-TrCP1 and ␤-TrCP2 (31); (2) a combination of two siRNAs, each directed against a single ␤-TrCP paralog, and; (3) a scrambled control siRNA. Although the control siRNA had no effect, the siRNAs directed against the ␤-TrCP1/2 transcripts efficiently reduced FLAGBirA*-␤-TrCP1/2 protein expression levels ( Fig. 2A). ␤-TrCP1/2 knockdown had no apparent effect on ␤-actin expression levels ( Fig. 2A) or the FG repeat-containing nuclear pore complex proteins recognized by mAb414 (38) (Fig. 2B). However, as of the 26S proteasome inhibitor MG132, ␤-TrCP substrates are stabilized. Following cell lysis under stringent buffer conditions, biotinylated proteins are affinity purified using streptavidin coupled to Sepharose beads. Streptavidin-bound proteins are washed and subjected to trypsin proteolysis, and the liberated peptides are identified using tandem mass spectrometry. B, Expression of FLAGBirA*-␤-TrCP1/2 leads to biotinylation of endogenous proteins. 293 T-REx cells expressing FLAGBirA*-␤-TrCP1 or FLAGBirA*-␤-TrCP2 were treated with tetracycline (1 g/ml) to induce protein expression, and with biotin (50 M) to enable proximity-dependent polypeptide labeling. Whole cell lysates were subjected to SDS-PAGE and immunoblotted with an anti-FLAG antibody (top panel) or streptavidin-HRP (horseradish peroxidase; bottom panel). C, ␤-TrCP1/2 interactors displaying a substrate profile. Proteins identified in the BioID analysis with a ProteinProphet score Ն0.85 (corresponding to Յ1% FDR), a SAINT score Ն0.75, and a spectral count ratio (ϩMG132/untreated) log 2 Ͼ1. Circles, polypeptides identified in ␤-TrCP1 BioID; squares, proteins identified in ␤-TrCP2 BioID. Previously identified ␤-TrCP interactors are highlighted in blue. Proteins demonstrated in this study to be stabilized following ␤-TrCP1/2 knockdown (see Fig. 2 (Fig. 3). In contrast, the FNIP1 protein displayed a substrate profile, but did not ap-pear to be stabilized in response to ␤-TrCP knockdown (supplemental Fig. S2B). As discussed below, this protein is therefore likely to be a substrate interacting partner rather than a direct ␤-TrCP substrate.
Together, these findings demonstrate that BioID can be used as an effective complement to standard IP-MS or other techniques for the identification of Ub E3-associated proteins. Indeed, in this experimental setting, BioID not only provided confirmatory data for the majority of FLAG IP-MS hits display-  Fig. 2 and supplementary Fig. S2) are boxed in dark blue, with white font ing a substrate profile (Table I, Fig. 1D), but also identified many additional substrate candidates ( Fig. 1D and 1E). BioID Identifies E3 Substrates in the Context of Protein Complexes-␤-catenin (CTNNB1) is a well-characterized ␤-TrCP substrate (22,(33)(34)(35) identified with high confidence in our analysis (Table I, Fig. 1D). In the absence of Wnt signaling, ␤-catenin is bound and phosphorylated by the ␤-catenin destruction complex (19). The newly created ␤-TrCP phosphodegron recruits SCF ␤-TrCP , which effects its ubiquitylation. Surprisingly, multiple ␤-catenin destruction complex components (APC, AXIN1, CSNK1A1, and FAM123B) also displayed a substrate profile in our BioID data set, even though these proteins are not direct SCF ␤-TrCP targets (Fig. 1C, Table I). However, these findings are consistent with a recent report demonstrating that ␤-catenin is targeted by SCF ␤-TrCP within the context of an intact destruction complex (40): since BioID is a proximity-based labeling method, when ␤-catenin is stabilized by MG132 treatment we may thus expect to observe increased biotinylation of both ␤-catenin and components of the destruction complex. It is therefore important to note that stabilizing a substrate (via MG132 treatment) that is part of a larger protein complex can lead to increased biotinylation of both the direct substrate and its interacting partners. This information can add significant value regarding the context in which a given substrate is targeted, but additional analysis of the complex will sometimes be required to identify the direct substrate(s) contained therein.
␤-TrCP-dependent Degradation of SUN2 is Critical for Nuclear Membrane Integrity-The LINC (linker of nucleoskeleton and cytoskeleton) complex is composed of SUN (Sad1/UNC-84) proteins localized to the inner nuclear membrane, and KASH (Klarischt/ANC-1/SYNE homology) domain-containing polypeptides localized to the outer nuclear membrane. SUN proteins interact with nuclear lamina components, and KASH domain-containing polypeptides interact with cytoplasmic cytoskeletal proteins. Together, the SUN and KASH proteins are critical for the maintenance of proper nuclear morphology and anchorage (41).
3xHA-SUN2 was dramatically stabilized (with an apparent Ͼ4-fold increase in half-life and a Ͼ2.5-fold increase in steady-state protein levels) by ␤-TrCP1/2 knockdown (Fig.  2D). Notably, knockdown of SCF ␤-TrCP1/2 was also associated with defects in nuclear morphology: while the control siRNA had no effect, 27% of cells treated with the ␤-TrCP1/2 siRNA displayed misshapen nuclei accompanied by nuclear membrane blebbing, as detected using antisera directed against endogenous SUN2 (Fig. 3A), FLAG-SUN2 (expressed stably in HeLa T-REx cells) and endogenous Lamin B (Fig. 3B,   supplemental Fig. S3A). Consistent with these observations, over-expression of mCherry-SUN2 (via transient transfection) was also associated with widespread nuclear membrane blebbing and morphological defects, and this effect was ex- 293 T-REx cells were transfected with control or the indicated ␤-TrCP siRNAs, and treated with cycloheximide (CHX) for the indicated times. Whole cell lysates from these cells were subjected to 4 -12% SDS-PAGE, and Western analysis conducted using mAb414, which recognizes FG repeat-containing nuclear pore complex proteins. C, 293 T-REx cells expressing 3xHA-NFE2L2 were transfected with control or ␤-TrCP siRNAs, as indicated, then treated with CHX for the indicated times. Whole cell lysates were subjected to 4 -12% SDS-PAGE, then probed for NFE2L2 protein expression using an anti-HA antibody. The apparent half-life (T 1 ⁄2) of 3xHA-NFE2L2 is indicated below each blot. D, Validation of novel ␤-TrCP substrate candidates. 293 T-REx cells expressing the indicated epitope-tagged proteins were transfected with control or ␤-TrCP siRNAs (as indicated) and treated with CHX for the indicated times. Protein expression levels were characterized using HA or FLAG antibodies. Apparent protein half-lives (in minutes) are indicated below each blot. acerbated by knockdown of ␤-TrCP1/2 ( Fig. 3C; supplemental Movie S1). Together, these data suggest that SCF ␤-TrCP plays an important role in the maintenance of nuclear membrane structure, and reveal a direct link between the control of SUN2 expression levels and the maintenance of nuclear membrane integrity. While further study will be required to understand the precise relationship between SUN2 levels and nuclear membrane structure, the defects associated with SUN2 misexpression are similar to phenotypes previously reported in fibroblasts derived from Emery-Dreifuss Muscular Dystrophy (EDMD) patients bearing mutations in LINC complex-associated nuclear lamina components such as LMNA and STA (42).
␤-TrCP function is Linked to P-body Abundance-Interestingly, many of the new ␤-TrCP1/2 targets identified by BioID are involved in mRNA turnover, silencing and translational control. For example, CNOT1, CNOT8, RQCD1 (CNOT9), CNOT10 and C2ORF29 (CNOT11) are all components of the CCR4-NOT complex, which catalyzes the removal of mRNA poly(A) tails prior to mRNA decay (43) and plays an important role in RNA-mediated gene silencing, along with an additional putative target identified here -TNRC6B (a GW182 protein) (44). All of these proteins localize to processing (P)-bodies, poorly soluble, dynamically regulated cytoplasmic ribonucleoprotein particles enriched in translation factors and mRNA decay proteins (45).
To begin to explore the role of SCF ␤-TRCP1/2 in the regulation of mRNP particles, P-bodies were visualized in live cells expressing GFP-or mCherry-tagged forms of the P-body markers DCP1A (Fig. 4A and supplemental Fig. S4A) or PATL1 (supplemental Fig. S4A). Previous studies have demonstrated that overexpression or depletion of P-body components can affect P-body number (46). Consistent with our observation that CCR4-NOT components and TNRC6B are stabilized in response to ␤-TrCP knockdown (Fig. 2D), P-body number was significantly increased in cells treated with siRNA directed against ␤-TrCP1/2 (3.46 Ϯ 0.37 P-bodies/cell), as compared with cells treated with the scrambled siRNA control 4B, supplemental Fig. S4A). This did not appear to be a general effect on RNA metabolism, as knockdown of ␤-TrCP1/2 had little or no impact on the number of stress granules (SG), another type of cytoplasmic mRNP particle (as measured here with the SG markers G3BP1, PABPC1, and eIF4A1; Fig. 4C-4D, supplemental Fig. S4B). While further study will be required to understand its precise role in the regulation of mRNA metabolism, these data implicate ␤-TrCP in P-body function.
␤-TrCP Targets PPP1R15B to Regulate eIF2␣ Phosphorylation-PPP1R15B (also known as CReP; constitutive reverter of eIF2␣ phosphorylation) (47) is a regulatory subunit of the Ser/Thr protein phosphatase 1 (PP1). In this capacity, PPP1R15B targets PP1 to the eukaryotic translation initiation factor 2 alpha (eIF2␣) protein. eIF2␣ is phosphorylated on Ser51 (by multiple kinases) in response to e.g. protein folding stress and viral infection (48). eIF2␣ phosphorylation levels regulate the rate at which the translation initiation factor eIF2 can form a complex with charged initiator methionyl-tRNA and deliver it to the 40S ribosomal subunit to initiate protein synthesis (48). Increased eIF2␣ Ser51 phosphorylation attenuates translation initiation to curb protein synthesis rates. Under basal conditions, PPP1R15B constitutively targets PP1 to eIF2␣ to maintain low levels of phosphorylated Ser51 (47). Proper regulation of eIF2␣ phosphorylation is critical for the maintenance of cellular homeostasis, as e.g. mice lacking the Ppp1r15b gene display severe growth retardation, hematopoietic system dysfunction and early embryonic lethality (49), and cells lacking PPP1R15B display a defective environmental stress response (47).

FIG. 4. ␤-TrCP modulates P-body formation.
A, HeLa Flp-In T-REx cells expressing the P-body marker GFP-DCP1A were transfected with control or ␤-TrCP1/2 siRNA and treated with tetracycline (1 g/ml, 24 h) and arsenite (500 M, 45 min), as indicated. B, Live cell confocal microscopy was conducted and P-body number quantified using Volocity software (see Experimental procedures). C, HeLa Flp-In T-REx cells expressing the stress granule marker GFP-G3BP1 were transfected with control or ␤-TrCP1/2 siRNA and treated with tetracycline and arsenite, as above. D, Live cell confocal microscopy was conducted and stress granule number quantified, as above. N.S.; not significant. in a Ͼ1.7-fold increase in steady-state expression levels (Fig.  5A). Notably, PPP1R15B contains a putative phosphodegron (SDSSLSDS, aa 459 -466) that could be recognized by SCF ␤-TrCP1/2 (22). The five serine residues in this region were thus mutated to alanines (S459 -466A) and the half-life of the resultant mutant protein characterized. Similar to that observed in response to ␤-TrCP knockdown, PPP1R15B S459 -466A displayed a substantially longer half-life than the wild type (WT) polypeptide (Fig. 5B). In addition, even though it is expressed at higher steady-state levels, the HA-PPP1R15B S459 -466A mutant protein was co-immunoprecipitated with FLAG-␤-TrCP1/2 much less efficiently than the WT protein (Fig. 5C), directly linking the integrity of the phosphodegron to ␤-TrCP binding.
Cells co-expressing WT or S459 -466A FLAG-PPP1R15B with HA-ubiquitin were next lysed under denaturing conditions (see Experimental procedures), the lysate was diluted and FLAG-tagged PPP1R15B proteins isolated by immunoprecipitation. As expected (because of increased steady-state expression levels), increased quantities of the mutant PPP1R15B were immunoprecipitated (see anti-FLAG blot, Fig. 5D, bottom panel). However, anti-HA Western blotting revealed considerably lower ubiquitylation levels of this protein, as compared with its WT counterpart (Fig. 5D, top panel). Together, these data strongly suggest that PPP1R15B is a ␤-TrCP substrate and that Ser459 -466 represents a bona fide ␤-TrCP phosphodegron. eIF2␣ pSer51 phosphorylation levels in cells expressing WT or S459 -466A PPP1R15B proteins were next monitored in the presence and absence of tunicamycin or thapsigargin, compounds that elicit ER stress by interfering with protein glycosylation or depletion of ER calcium stores, respectively. As expected, eIF2␣ Ser51 phosphorylation was increased in cells expressing the WT PPP1R15B protein in response to ER stress (Fig. 5E). In cells expressing the PPP1R15B S459-S466A mutant protein, eIF2␣ phosphorylation was also increased, but to a much lesser extent (Fig. 5E). The proper regulation of steady-state PPP1R15B protein levels is thus critical for the proper regulation of eIF2␣ phosphorylation in response to protein folding stress. Finally, similar to that previously observed in eIF2␣ Ser51Ala mouse embryo fibroblasts (50), expression of the PPP1R15B S459 -466A mutant protein sensitized cells to ER stress, as measured by cell viability in the presence of increasing doses of tunicamycin (Fig. 5F). Together, our data validate PPP1R15B as an SCF ␤-TrCP substrate, and reveal an important role for this E3 ligase in the regulation of translation initiation. DISCUSSION Understanding the targeting landscape of the ubiquitin system represents a major challenge in cell biology. The identification of E3 ligase substrates has been difficult, in part because: (1) ligase -substrate interactions are often of very low-affinity (generally in the high nM to microMolar (M) range) and/or transient in nature; (2) many substrates are subjected to rapid proteasomal degradation, and therefore cannot be detected using standard techniques; (3) the human ubiquitome is extremely complex, and therefore difficult to study and; (4) many substrate proteins are localized to poorly soluble cellular compartments, making their isolation and identification by standard IP-based techniques problematic.
Here we demonstrate that BioID combined with semi-quantitative mass spectrometry, conducted on cells in the presence and absence of MG132, can be used for the identification of E3 substrates. Focusing on the well-studied F-box proteins ␤-TrCP1 and ␤-TrCP2, we identified a number of established targets along with numerous new substrate candidates, implicating ␤-TrCP in several novel biological functions.
An important limitation of IP-based approaches for the characterization of protein-protein interactions is that no particular buffer combination and cell lysis procedure can be expected to both maintain all protein-protein interactions (and especially weak or transient E3-substrate interactions) and simultaneously solubilize all proteins in a cell. Since BioID utilizes a covalent biotin modification to mark bait protein interactors, the maintenance of protein-protein interactions is not required for their identification, and harsh lysis techniques may be used to liberate proteins from poorly soluble cellular fractions that are often not accessed by IP-compatible buffers. The straightforward, scalable technique presented here is thus highly complementary to standard IP-MS approaches, and is likely to be broadly applicable for E3 substrate identification.
A Role for ␤-TrCP in the Maintenance of the Nuclear Membrane-As a component of the LINC complex, the new ␤-TrCP substrate SUN2 plays an important role in the maintenance of nuclear integrity, intracellular nuclear positioning and nuclear membrane structure (51). Mutations in LINC complex components (and LINC complex interacting partners) have been associated with EDMD, a progressive muscular dystrophy characterized by myofiber degeneration and cardiomyopathy. EDMD muscle biopsies and primary fibroblasts derived from EDMD patients display frequent abnormal nuclear lobulation and herniation, accompanied by the presence of micronuclei (52)(53)(54). These reports are consistent with our observations in cells with misregulated SUN2 levels.
Although the SUN1 and SUN2 genes have not yet been directly linked to EDMD, these LINC proteins clearly represent promising candidates for further study. (Indeed, the currently defined gene mutations associated with EDMD have been found in less than half of EDMD patients (51)). Although additional study will also be required to better understand the role of ␤-TrCP in the regulation of nuclear membrane structure and function, our data reveal a critical link between this E3 ligase and control of a LINC complex component.
A Role for ␤-TrCP in the Cellular Stress Response-␤-TrCP is the E3 ligase for IB, which sequesters the NFB complex 293 T-REx cells expressing FLAG-tagged WT or the S459 -466A PPP1R15B phosphodegron mutant were treated with CHX for the indicated times. Cell lysates were subjected to Western blotting with antibodies directed against FLAG and beta actin. Apparent half lives of the FLAG-tagged PPP1R15B proteins are indicated below each blot. C, The ␤-TrCP interaction is disrupted in a PPP1R15B phosphodegron mutant. Co-immunoprecipitation of FLAGBirA*-␤-TrCP1/2 with HA-tagged WT or S459 -466A PPP1R15B proteins. D, Mutation of the phosphodegron disrupts PPP1R15B ubiquitylation. 293 T-REx cells were co-transfected with plasmids coding for FLAG-PPP1R15B WT or S459 -466A and HA-ubiquitin. Lysates were prepared under denaturing conditions, diluted and FLAG IPs conducted. The isolated proteins were subjected to Western blot analysis with anti-FLAG and anti-HA. Whole cell lysates were probed with anti-HA. E, Expression of the PPP1R15B phosphodegron mutant results in decreased eIF2␣ Ser51 phosphorylation levels. Cell lsyates derived from 293 T-REx cells expressing FLAG-tagged WT or S459 -466A PPP1R15B phosphodegron mutant proteins, and treated with vehicle (DMSO) or the ER stress agents tunicamycin or thapsigargin, were subjected to Western blotting, as indicated. F, Expression of the PPP1R15B phosphodegron mutant imparts sensitivity to ER stress agents. 293 T-REx cells expressing WT or phosphodegron mutant PPP1R15B proteins were exposed to the indicated concentrations of tunicamycin for 24 h. Viable cells, as detected by MTT assay (untreated culture numbers set to 1) are displayed (n ϭ 3; a representative experiment is shown). in the cytoplasm. When IB is phosphorylated by the IB kinase (IKK) in response to e.g. viral infection, tissue damage or other types of cellular stress, the newly created phosphodegron recruits SCF ␤-TrCP . Once IB is degraded, the NFB complex can translocate to the nucleus to effect stressadaptive changes in the transcription program.
Our data reveal previously unidentified links between ␤-TrCP and other proteins involved in the cellular stress response. P-bodies play important roles in the regulation of RNA stability, and P-body formation and turnover is highly regulated in response to cellular stress (45,55). eIF2␣ phosphorylation levels are also regulated in response to many types of stress, including e.g. viral infection, protein folding stress and nutrient deprivation (48). ␤-TrCP may therefore represent an important hub for modulating the proteome in response to cellular stress: i.e. in addition to effecting changes in the transcriptome via NFB activation, the regulation of translation via the sequestration or degradation of pre-existing mRNAs in P-bodies, combined with a global reduction in translation initiation rates via increased eIF2␣ phosphorylation levels, could represent a rapid and efficient way to re-program the entire proteome in response to stress. Further study will be focused on understanding the role of SCF ␤-TrCP in the cellular stress response.
BioID as a Complementary Approach for the Identification of E3 Substrates-While our work was in progress, two additional studies were published on the identification of ␤-TrCP substrates (56,57), utilizing more standard IP-MS and TAP (tandem affinity purification)-MS approaches. Notably, despite the use of widely disparate approaches, many of the same ␤-TrCP interactors were detected among all three studies, significantly expanding the number of high-confidence ␤-TrCP substrates and implicating this already well-studied E3 ligase in several new biological processes. Our work highlights the use of BioID as a complementary approach for the identification of E3 targets in living cells.