Peptidic degron in EID1 is recognized by an SCF E3 ligase complex containing the orphan F-box protein FBXO21

Significance

Many short-lived proteins perform important cellular functions that must be tightly regulated. EID1 is an unstable protein implicated in the control of transcription, differentiation, and genomic integrity. EID1 is particularly unstable in G0 cells. Here we show that an SCF complex containing the orphan F-box only protein FBXO21 is the ubiquitin ligase that recognizes EID1 in both cycling and G0 cells and that it recognizes a modular, peptidic domain within EID1. This degron overlaps with the EID1 region that binds to the retinoblastoma tumor suppressor protein (pRB) and to melanoma-associated antigen (MAGE) proteins, suggesting that multiple proteins compete for binding to EID1 and secondarily influence its ubiquitinylation. This study sheds light on the regulation of EID1, the function of FBXO21, and control of protein turnover by ubiquitin ligases.

Abstract

EP300-interacting inhibitor of differentiation 1 (EID1) belongs to a protein family implicated in the control of transcription, differentiation, DNA repair, and chromosomal maintenance. EID1 has a very short half-life, especially in G0 cells. We discovered that EID1 contains a peptidic, modular degron that is necessary and sufficient for its polyubiquitylation and proteasomal degradation. We found that this degron is recognized by an Skp1, Cullin, and F-box (SCF)-containing ubiquitin ligase complex that uses the F-box Only Protein 21 (FBXO21) as its substrate recognition subunit. SCF^FBXO21 polyubiquitylates EID1 both in vitro and in vivo and is required for the efficient degradation of EID1 in both cycling and quiescent cells. The EID1 degron partially overlaps with its retinoblastoma tumor suppressor protein-binding domain and is congruent with a previously defined melanoma-associated antigen-binding motif shared by EID family members, suggesting that binding to retinoblastoma tumor suppressor and melanoma-associated antigen family proteins could affect the polyubiquitylation and turnover of EID family members in cells.

Regulation of protein abundance through alterations in protein stability plays a critical role in cellular homeostasis. For many proteins, stability is governed by the concerted actions of dedicated ubiquitin ligases, which promote proteasomal degradation, and deubiquitinating enzymes, which oppose them. There are thought to be over 500 potential ubiquitin ligases and ∼100 deubiquitinating enzymes (1, 2).

Many ubiquitin ligases belong to the Skp1, Cullin, and F-box (SCF) family, characterized by the presence of S-phase kinase-associated protein 1 (Skp1), Cullin-1 (Cul1), and an F-box protein, with the latter serving as the substrate recognition subunit (3, 4). F-box proteins contain the collinear motif for which they are named and, frequently, a dedicated protein-binding motif such as a WD40 domain. F-box proteins lacking such protein-binding domains are referred to as F-box only proteins. Archetypal SCF complexes include the SCF complex containing the F-box protein Skp2, which polyubiquitylates proteins such as the cell-cycle regulators p27 and cyclin E, and the SCF complex containing Fbw7, which polyubiquitylates proteins such as c-Jun and c-Myc (1).

EP300-interacting inhibitor of differentiation 1 (EID1) is a short-lived protein with a half-life measured in minutes, especially as cells exit the cell cycle (5, 6). It was originally identified by virtue of its ability to bind to the retinoblastoma tumor suppressor protein (pRB) and contains an LXCXE pRB-binding motif first identified in the viral oncoproteins adenovirus E1A, SV40 large T, and human papillomavirus E7 (5–9). Structural studies suggest that EID1 makes two contacts with pRB, one involving the pRB N terminus and one involving the LXCXE-binding domain within the pRB C terminus (10). EID1 also binds to small heterodimer partner (SHP) [also called NR0B2 (nuclear receptor subfamily 0, group B, member 2)], which is a transcriptional corepressor for various steroid hormone family transcription factors (11–14). In cell-based models, overexpression of EID1 likewise represses transcription factors and blocks differentiation (5, 6, 11, 14–17). These cellular activities correlate with its ability to inhibit p300 histone acetyltransferase in vitro and in vivo (5, 6, 11).

Proteins with very short half-lives often perform important functions that justify the energetic cost of their continuous synthesis and degradation. Previous studies of EID1 function relied heavily on exogenous overexpression, which could introduce artifacts caused by nonphysiological protein abundance. In an attempt to control EID1 abundance in a more physiological manner, and to begin to understand its normal functions, we sought to identify the ubiquitin ligase for EID1.

Results

Previous structure-function studies showed that the EID1 C terminus is required both for its polyubiquitylation and for its ability to bind to the mouse double minute 2 homolog (MDM2) ubiquitin ligase, suggesting the two properties are linked (5). Moreover, exogenously expressed MDM2 polyubiquitylates EID1 in cells (5, 18). On the other hand, we found that EID1 was rapidly polyubiquitylated and degraded in p53^−/−;Mdm2^−/− mouse embryo fibroblasts, strongly suggesting the existence of another EID1 ubiquitin ligase. In addition, two studies showed that EID1 is dramatically induced by MLN4924, which inhibits the NEDD8 activating enzyme and hence cullin-dependent ubiquitin ligases (19, 20). Together, these observations suggested that EID1 is polyubiquitylated by a cullin-dependent ubiquitin ligase instead of, or perhaps in addition to, MDM2.

We tried to define the minimal EID1 degron as a prelude to identifying the EID1 ubiquitin ligase. Toward this end, we transiently expressed various fragments of EID1, fused to the C terminus of green fluorescent protein (GFP), in HeLa cells and monitored their accumulation in the presence and absence of the proteasomal inhibitor MG132 by anti-GFP immunoblot analysis. EID1 contains 187 amino acid residues (Fig. 1A). Consistent with prior work, an EID1 mutant [EID1(1–157)] lacking its extreme C terminus accumulated to very high levels compared with wild-type EID1 and was largely insensitive to MG132 (Fig. 1B). In contrast, the behavior of GFP-fused EID1 residues 151–187 was indistinguishable from that of GFP-fused to full-length EID1 in these assays, being produced at very low levels in untreated cells and induced by MG132 (Fig. 1B). Therefore, the EID1 C terminus is both necessary and sufficient for the proteasomal degradation of EID1.

Fig. 1.

Mapping of EID1 degron. (A) EID1 schematic. (B–G) Anti-GFP immunoblot analysis of HeLa cells transiently transfected to produce the indicated GFP-EID1 fusion proteins and then treated or untreated with 10 μM MG132 overnight as indicated.

We then performed fine-mapping studies using a nested set of C-terminal and N-terminal EID1(151–187) truncation mutants fused to the GFP C terminus. Progressive elimination of C-terminal residues from EID1(151–187) revealed that EID1 residues 173–187 are dispensible for proteasomal degradation of EID1 (Fig. 1C). In contrast, eliminating residues 170–187 stabilized EID1 (Fig. 1C). Conversely, progressive elimination of N-terminal residues from EID1(151–187) showed that residues 151–159 are dispensible for proteasomal degradation of EID, whereas eliminating residues 151–162 stabilized EID1 (Fig. 1D). This latter result needs to be interpreted cautiously as the presence of the GFP moiety could have steric effects on the N terminus of the EID1 degron. Specifically, some N-terminal EID1 amino acids in the fusions could be acting as spacers. This concern, however, is largely mitigated by experiments described below.

Consistent with these results, GFP fused to EID1 160–172 was apparently unstable and induced by MG132, whereas in-frame deletion of EID1 residues 160–172 stabilized otherwise full-length EID1 (Fig. 1E and Fig. S1). Finally, we performed an alanine scan in the context of full-length EID1 focused on residues 160–172, changing three sequential residues to alanine at a time. Converting EID1 residues 158–160, 167–169, or 170–172 to three alanines (AAA) had no discernable effect on EID1 degradation (Fig. 1F). In contrast, converting EID1 residues 161–163, and especially 164–166, to AAA partially stabilized EID1 (Fig. 1F). A more focused alanine scan, changing one alanine at a time, showed that L165 was required for the rapid turnover of EID1 (Fig. 1G).

Fig. S1.

EID1 contains a modular degron. Anti-GFP immunoblot analysis of HeLa cells transiently transfected to produce the indicated GFP-EID1 fusion proteins and then treated or untreated with 10 μM MG132 as indicated. This is the long exposure of Fig. 1E.

To identify the relevant ubiquitin ligase, we stably infected HeLa cells with a retrovirus encoding exogenous EID1 with both a Flag epitope and hemagglutinin (HA) epitope tag at its N terminus (pBABE-Flag-HA-EID1) or the empty vector (vector) (Fig. 2A), treated the cells with MG132 overnight, and performed a preparative immunoprecipitation with anti-Flag Sepharose. Bound proteins were eluted with a Flag epitope peptide and then immunoprecipitated with anti-HA Sepharose, eluted with an HA epitope peptide, and analyzed by silver staining (Fig. 2B) and by mass spectrometry (MS) (Fig. 2C). The presence of EID1 in the sequential Flag and HA eluates was confirmed by immunoblot analysis (Fig. 2A). MS analysis of the final HA eluate from cells expressing Flag-HA-EID1 revealed the presence of multiple proteins, including the F-box Only Protein 21 (FBXO21) (Fig. 2C and Fig. S2). We pursued FBXO21 because F-box proteins often serve as substrate recognition modules in SCF ubiquitin ligases.

Fig. 2.

EID1 copurifies with FBXO21. (A) Anti-EID1 immunoblot analysis of whole-cell extracts (WCEs) of HeLa cells stably infected with a pBABE-based retrovirus encoding Flag-HA-EID1 or with the empty vector (Top), Flag peptide eluate after anti-Flag immunoprecipitation (Middle), and HA peptide eluate after anti-HA immunoprecipitation of the Flag eluate (Bottom). (B) Silver stain of final HA peptide eluates from cells treated as in A. (C) Peptides detected by MS/MS analysis of HA peptide eluates as in B.

Fig. S2.

Identification of FBXO21 in anti-EID1 immunoprecipitates by MS. Shown is the MS/MS spectrum of tryptic peptide derived from FBXO21(555–577) showing the assigned N-terminal (b-type) and C-terminal (y-type) fragment ions. The colored glyphs above and below the amino acid sequence indicate detected fragment ions, with doubly charged ions represented as lighter shades. Position of colored circles represents relative abundance of detected fragment ions.

In cotransfection experiments, we confirmed that exogenous FBXO21 can bind to EID1 in cells. In one set of experiments, 293FT cells were cotransfected to produce HA-EID1 and Flag-FBXO21 and treated with MG132. EID1 was detected, as determined by anti-HA immunoblot analysis, in anti-Flag immunoprecipitates (Fig. 3A). The recovery of HA-EID1 required the presence of Flag-FBXO21 and was specific because binding of HA-EID1(164–166AAA) and HA-EID1(1–157) to Flag-FBXO21 was diminished and absent, respectively, in these assays (Fig. 3A). Similarly, wild-type EID1, but neither EID1(164–166AAA) nor EID1(1–157), bound to FBOX21 in mammalian two-hybrid assays in which EID1 was expressed as the prey protein fused to VP16 and FBXO21 was used as the bait fused to the Gal4 DNA-binding domain (Fig. 3 B and C).

Fig. 3.

FBXO21 binds to the EID1 degron. (A) Immunoblot analysis of WCEs and anti-Flag immunoprecipitates (IP:Flag) prepared from 293FT cells transfected with a plasmid encoding Flag-FBXO21 or the empty vector together with a plasmid encoding the indicated HA-EID1 variants or empty vector. Cells were treated with MG132 for 3 h before lysis. (B) Immunoblot analysis of 293FT cells transfected with a plasmid encoding Gal4-FBXO21 or the empty vector together with a plasmid encoding the indicated EID1–VP16 fusion proteins (or empty vector), a plasmid encoding renilla luciferase, and a reporter plasmid containing Gal4 DNA-binding sites upstream of firefly luciferase. (C) Normalized firefly luciferase values for cells treated as in B 24 h after transfection. (D) Anti-Flag immunoblot (Top) of Flag-FBXO21 in vitro translate bound to the indicated GST-EID1 fusion proteins. Comparable recovery of the GST fusion proteins was confirmed by Coomassie blue staining (Bottom). (E and F) Anti-FBXO21 immunoblot analysis of Flag-FBOX21 produced in 293FT cells and bound to the indicated EID1(151–177) (E) and EID(160–172) (F) biotinylated peptides. The 293FT cells transfected with empty vector were included as controls.

To ask if FBXO21 could bind to recombinant EID1, we performed GST–pull-down assays with recombinant GST-EID1 fusion proteins. Flag-FBXO21 produced by in vitro translation specifically bound to GST-EID1 but bound poorly to GST-EID1(164–166AAA) and undetectably to GST-EID1(1–157) (Fig. 3D). Comparable recovery of the different GST fusion proteins was confirmed by Coomassie blue staining (Fig. 3D).

Finally, we did binding assays with biotinylated EID1 peptides. Flag-FBXO21 produced in 293T cells by transient transfection bound to synthetic EID1 peptides corresponding to EID1 residues 151–177 or the minimal degron 160–172 (Fig. 3 E and F, respectively). Consistent with our degron alanine scanning experiments, replacement of residues 158–163 or residues 164–169 with a flexible linker (NAAIRS) (21) in the EID1 151–177 peptide abrogated FBXO21 binding, as did replacement of residues 161–163 or 164–166 in the EID1 160–172 peptide with three consecutive alanine residues (Fig. 3 E and F).

Next we performed in vitro and in vivo ubiquitylation assays. For the former, 293FT cells were transfected to produce Myc epitope-tagged versions of wild-type FBXO21, an F-box deleted version of FBXO21(FBXO21ΔF), or were transfected with an empty vector. We then performed in vitro ubiquitylation assays with the anti-Myc immunoprecipitates and ³⁵S-labeled EID1 produced by in vitro translation with rabbit reticulocyte lysates. Myc-FBXO21, but not Myc-FBXO21ΔF, immunoprecipitates ubiquitylated EID1 when supplemented with a recombinant E1 ubiqutin-activating enzyme and an E2 ubiquitin-conjugating enzyme (Fig. 4A). This was specific because ubiquitylation of EID1(164–166AAA) was greatly diminished compared with wild-type EID1, and ubiquitylation of EID1(1–157) was undetectable (Fig. 4B). Similarly, wild-type FBXO21 promoted the ubiquitylation of wild-type T7-EID1 in vivo, as determined by the presence of HA-tagged ubiquitin in anti-T7 immunoprecipitates performed under denaturing conditions from cells in which the proteasome was inhibited by MG132 (Fig. 4 C and D). As was true in the in vitro assays, ubiquitylation of EID1(164–166AAA) and EID1(1–157) by FBOXO21 in vivo was greatly diminished and absent, respectively, compared with wild-type EID1 (Fig. 4 C and D).

Fig. 4.

Polyubiquitylation of EID1 by an FBXO21. (A and B) Autoradiogram (Top) of the indicated ³⁵S HA-EID1 in vitro translates after incubation with anti-Myc immunoprecipitates from 293FT cells transfected to produce Myc-FBXO21, Myc-FBXO21ΔF-box, or with the empty vector, in the presence or absence of recombinant E1 and E2, as indicated. Comparable recovery of Myc-tagged proteins was confirmed by immunoblot analysis (Bottom). In B, ubiquitylation assays performed with a mock immunoprecipitate lacking a primary antibody (beads only) were included as an additional negative control. Lanes labeled input equal one-third of the in vitro translate amount used in the ubiquitylation assays. (C) Anti-HA (Top) and anti-T7 (Bottom) immunoblots of anti-T7 immunoprecipitates from 293FT cells transfected to produce Flag-FBXO21, HA-ubiquitin, and T7-EID1 variants, as indicated. (D) Immunoblots of WCEs used in D.

Next we stably infected 293FT cells to produce Flag-HA-FBXO21 and identified FBXO21-associated proteins in the presence of MG132 by sequential preparative immunoprecipitations with anti-Flag and anti–HA-Sepharose followed by MS. MS revealed the presence of peptides from Skp1 and Cul1 in addition to peptides from EID1 and its paralog EP300-interacting inhibitor of differentiation 2 (EID2) (Fig. S3A). Notably, the EID1 degron described above is well conserved in EID2 (Fig. S3B). Coimmunoprecipitation of Skp1 and Cul1 with FBXO21 was confirmed by immunoblot analysis (Fig. 5A). These findings suggested that FBXO21 is the substrate recognition subunit of a canonical SCF ubiquitin ligase complex and strengthened the idea that EID1 is one of its substrates.

Fig. S3.

Identification of FBXO21-binding partners. (A) Proteins identified in preparative immunoprecipitations of epitope-tagged FBXO21. Protein matches, number of unique peptides identified by MS. (B) Sequence alignment of human EID1 and human EID2. Degron recognized by FBXO21 is indicated.

Fig. 5.

SCF^FBXO21 regulates EID1 turnover in cells. (A) Immunoblot analysis of WCEs and anti-Flag immunoprecipitates (IP:Flag) prepared from 293FT cells stably infected to express Flag-HA-FBXO21 or the empty vector. MG132 was added for 3 h where indicated. (B) Immunoblot analysis of U2OS cells treated with MLN4924 for 4 h. (C–E) Immunoblot analysis of 293FT cells transfected with the indicated siRNAs. In E, cells were retrovirally infected to stably produce the indicated HA-EID1 variants. (F) Immunoblot analysis of 293FT cells infected to produce the indicated shRNAs. IP, endogenous FBXO21 immunoprecipitate; WCE, whole-cell extract. (G) Immunoblot analysis of A549 cells infected to produce Cas9 and the indicated sgRNAs. (H) Immunoblot analysis of 293FT cells transfected with the indicated siRNAs and then treated with CHX for the indicated duration. (I) Normalized luciferase values after transfecting 293FT cells as in G with a plasmid encoding EID1 fused to firefly luciferase and renilla luciferase from a bicistronic mRNA.

We used several approaches to ask whether SCF^FBXO21 regulates EID1 turnover in cells. Consistent with two recent studies (19, 20), MLN4924 robustly induced the accumulation of endogenous EID1, implicating an SCF ubiquitin ligase in the control of EID1 turnover (Fig. 5B). Down-regulation of FBXO21 or Cul1 with effective small interfering RNAs (siRNAs) induced the accumulation of endogenous EID1 (Fig. 5 C and D and Fig. S4). These effects were specific because effective siRNAs against other F-box only proteins or cullins did not induce EID1 (Fig. 5 C and D and Fig. S4 A–D). Similarly, siRNAs against FBXO21 and Cul1 induced the accumulation of exogenous EID1 but had little or no effect on EID1(164–166AAA) (Fig. 5E). MDM2 has also been reported to regulate EID1 turnover cells (5, 18), but we did not observe robust induction of EID1 with an effective MDM2 siRNA, either alone or in combination with an FBXO21 siRNA (Fig. 5E and Fig. S4 E and F). Consistent with our siRNA results, stable knockdown of FBXO21 with multiple independent small hairpin RNAs (shRNAs) (Fig. 5F) or with CRISPR-associated protein 9 (Cas9) and multiple independent FBXO21 small guide RNAs (sgRNAs) induced the accumulation of EID1 (Fig. 5G). In contrast, but in keeping with our siRNA results, Cas9-mediated knockdown of MDM2 did not induce EID1 but did induce the canonical MDM2 target p53 (Fig. 5G).

Fig. S4.

siRNA validation. (A and B) Target mRNA levels (A) and EID1 mRNA levels (B) as determined by real-time PCR in 293FT cells transfected with siRNAs against the indicated F-Box only genes. (C and D) Target mRNA levels (A) and EID1 mRNA levels (B) as determined by real-time PCR in 293FT cells transfected with siRNAs against the indicated Cullin genes or FBXO21. (E and F) FBXO21 mRNA levels (E) and MDM2 mRNA levels (F) as determined by real-time PCR in HeLa cells transfected with the indicated siRNAs as in Fig. 5E.

As expected, the accumulation of EID1 in cells lacking FBXO21 was linked to increased EID1 stability, as determined by its rate of disappearance in cells treated with cycloheximide (CHX) (Fig. 5H), and was posttranscriptional, as evidenced by increased firefly luciferase activity in cells transfected with a bicistronic reporter plasmid encoding an EID1–firefly luciferase fusion protein and renilla luciferase (22) (Fig. 5I and Fig. S5).

Fig. S5.

Elimination of FBXO21 using CRISPR. Shown is the immunoblot of 293FT cells expressing Cas9 and the indicated sgRNAs.

EID1 protein abundance falls as cells exit the cell cycle (5, 6). To ask whether cells that have not previously entered the cell cycle regulate EID1 at the G0/1 transition, we stimulated resting T cells to enter the cell cycle and monitored EID1 protein levels. EID1 protein was barely detectable in resting T cells but was induced 12–24 h after T-cell receptor activation, which stimulates cell-cycle entry, without a corresponding change in EID1 mRNA abundance (Fig. 6 A and B). EID1 protein levels were also induced in G0 cells by the proteasome inhibitor MG132 (Fig. 6C). Therefore, EID1 protein levels are regulated upon both entry into and exit from the cell cycle.

Fig. 6.

SCF^FBXO21 is required for degradation of EID1 in G0. (A–C) Immunoblot analysis (A and C) and mRNA levels (B) of resting T cells induced to enter the cell cycle with anti-CD3 (cluster of differentiation 3) and anti-CD28 (cluster of differentiation 28) antibodies. In C, cells were also treated with MG132 for the indicated duration before cell lysis. (D and E) Immunoblot analysis of T98G (D) and WI-38 (E) cells grown under serum-poor and serum-replete conditions. (F and G) Normalized luciferase values (F) and immunoblot analysis (G) of T98G cells stably expressing an EID1–luciferase stability reporter after transfection with the indicated siRNAs and growth in the presence or absence of serum for 72 h.

To ask if SCF^FBXO21 regulates EID1 in both G0 and cycling cells, we analyzed T98G glioblastoma cells and WI-38 human fibroblasts that were serum-starved into G0 or maintained in serum. Under both conditions EID1 was induced by Cas9-mediated elimination of FBXO21 with two different effective sgRNAs (Fig. 6 D and E and Fig. S6 A and B). In a complementary set of experiments, we transiently transfected T98G stably producing a bicistronic EID–luciferase stability reporter with an effective FBXO21 siRNA or a control siRNA and then grew the cells for 72 h in the presence or absence of serum. EID1 stability, as measured by firefly/renilla luciferase ratio, decreased in serum-starved cells treated with the control siRNA but not FBXO21 siRNA (Fig. 6 F and G and Fig. S6C). Notably, FBXO21 is present in both G0 cells and in cycling cells. In some, but not all, cells, we observed a modest increase in FBXO21 levels in G0 compared with cycling cells (Fig. 6 C–E). We have not, under the assay conditions tested to date, detected a decrease in FBXO21 ubiquitin ligase activity in cycling cells compared with G0 cells nor changes in the subcellular localizations of FBXO21 and EID1 under these two conditions. It will be important to determine how, mechanistically, EID1 levels increase in cycling cells.

Fig. S6.

Cell-cycle exit after serum withdrawal. (A and B) Cell-cycle distribution, as determined by propidium iodide staining, of T98G cells (Left) and WI-38 cells (Right) expressing the indicated sgRNAs and grown under serum-poor or serum-replete conditions for 72 h. (C) Normalized luciferase values (Left) and immunoblot analysis (Right) of T98G cells stably expressing the EID1-luciferase reporter depicted in Fig. 6 after growth for 72 h in the presence or absence of serum. Also shown is cell-cycle distribution as determined by FACS.

Discussion

Our studies show that SCF^FBXO21 is a ubiquitin ligase for EID1. The same conclusion has now been reported by another group working in parallel (23, 24). Despite earlier reports (5, 18), we have thus far not confirmed a physiological role for MDM2 in EID1 ubiquitylation, although MDM2 might polyubiquitylate EID1 under conditions not tested here.

EID1 has several paralogs including EID2, EID2B, and EID3 (25–32). The EID1 degron identified here is well conserved among the EID family members, and EID2 coimmunoprecipitates with FBXO21 in cells treated with MG132. It is therefore likely that SCF^FBXO21 regulates other EID family members.

The binding of SCF complexes to their substrates is often regulated by posttranslational degron modifications. The EID1 degron we identified here is not known to be modified and, as a synthetic, presumably unmodified, peptide binds to FBXO21. It remains possible that the EID1 degron is modified under certain circumstances, leading to altered recognition by FBXO21 or that regulatory signals impinge upon SCF^FBXO21 rather than its substrates.

In overexpression studies, EID family members can repress transcription mediated by various DNA-binding transcription factors, perhaps by inhibiting p300 and CBP, and can block differentiation. On the other hand, we have thus far been unable to replicate these EID1-induced phenotypes by inactivating FBXO21, suggesting that they were artifacts caused by EID1 overexpression or that FBXO21 loss has additional effects that mitigate the consequences of increased EID abundance.

Finally, it has been suggested that the EID and melanoma-associated antigen (MAGE) proteins are mammalian counterparts of yeast Nse4 and Nse3, respectively, which form an SMC complex involved in chromatin maintenance and DNA repair (30, 32). Conceivably this function provides a teleological explanation for the induction of EID proteins during the cell cycle. Moreover, the EID1 degron recognized by FBXO21 is essentially congruent with the MAGE/Nse3-binding motif identified by others. Perhaps binding to MAGE/Nse3-like proteins shields EID1 from FBXO21 during the cell cycle. The ability of the MAGE-related protein Necdin to bind to and stabilize EID1 is consistent with this idea (16). Similarly, the EID1 degron overlaps with the EID1 pRB-binding region, suggesting that EID1 stability could also be influenced by pRB and providing a mechanistic explanation for the induction of EID1 in cells observed after exogenous pRB expression (5).

Materials and Methods

Cell Culture.

HeLa, U2OS, T98G, and 293FT cells were maintained in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% (vol/vol) FBS. WI-38 cells were grown in Eagle’s Minimal Essential Media supplemented with 10% (vol/vol) FBS. A549 cells were maintained in RPMI supplemented with 10% (vol/vol) FBS. To induce cell-cycle exit, T98G cells and WI-38 cells were grown in DMEM without FBS and 0.1% FBS, respectively, for 72 h. Where indicated, cells were treated with 10 μM MG132 (474790; EMD Millipore) or 1 μM MLN4924 (A-1139; Active Biochem). Following retroviral or lentiviral infection, stable cell lines were cultured in media containing 1 μg/mL puromycin, 10 μg/mL blasticidin, or 200 μg/mL hygromycin.

Plasmids.

EID1 cDNAs encoding full-length EID1(1–187), the C-terminal truncation mutant EID1(1–157), and the N-terminal truncation EID1 mutant (151–187) were amplified with a 5′ primer that introduced a BamH1 site and a 3′ primer that introduced an EcoRI site. The PCR products were cut with BamHI and EcoRI and subcloned into pcDNA3-GFP, pcDNA3-T7, and pGEX-2T that had been linearized with these two enzymes. The internal deletion mutant EID1(Δ160–172) was made by overlapping PCR-based site-directed mutagenesis and then cloned into pcDNA3-GFP. Mutants including EID1(151–181), EID1(151–178), EID1(151–175), EID1(151–172), EID1(151–169), EID1(154–187), EID1(157–187), EID1(160–187), EID1(163–187), EID1(158–160AAA), EID1(161–163AAA), EID1(164–166AAA), EID1(167–169AAA), EID1(170–172AAA), EID1(F161A), EID1(I162A), EID1(E163A), EID1(E164A), EID1(L165A), and EID1(F166A) were generated by site-directed mutagenesis. To make pBABE-Flag-HA-EID1, pcDNA3-HA-EID1, pcDNA3-HA-EID1(164–166AAA), and pcDNA3-HA(1–157), the corresponding EID1 cDNAs were amplified with a 5′ oligonucleotide that introduced an N-terminal tag (FLAG-HA or HA) and subcloned into pBABE-puro or pcDNA3 (Invitrogen) cut with BamHI and EcoRI. To generate pVP16-EID1 and its mutants, an EID1 cDNA was amplified with primers that introduced a 5′ EcoRI site and a 3′ BamHI site and was subcloned into a pVP16 vector (Clontech). The EID1 stability reporters were generous gifts from Dr. Gang Lu, Harvard Medical School, Boston (22).

An FBXO21 cDNA was recovered from an EST clone SC124348 (Origene) by PCR amplification using primers that introduced an N-terminal epitope tag (Myc, FLAG, or Flag-HA) and then cloned into pcDNA3 cut with EcoRI and XhoI or pRetroX-Tight-Pur (Clontech) cut with NotI and EcoRI. pM-Gal4-FBXO21 was generated by PCR amplification of an FBXO21 cDNA with primers that introduced a 5′ EcoRI site and 3′ SalI site to facilitate subcloning into a modified pM Gal4 DNA-BD cloning vector (Clontech).

Lentiviral pLKO.1 human EID1 shRNA vectors (TRCN0000134015, target sequence, CCTATGGCTTTGACTTTGTTA; TRCN0000138936, target sequence, GCCGACAAGATGTTTCTGAGA; and TRCN0000135114, target sequence, CTCGGCTGTGATGAGATTATT) and human FBXO21 shRNA vectors (TRCN0000034285, target sequence, GCCTTCCCTTATGAAACACTA; TRCN0000034286, target sequence, CCTGGACATCTTTGACTACAT; and TRCN0000034288, target sequence, CGGCAACAGAAGATCTTAAAT) were obtained from the Broad Institute TRC shRNA library. A nontargeting hairpin (target sequence, TGAACTTCAGGGTCAGCTTGC) was used as a control.

Immunoblot Analysis and Immunoprecipitation.

Cells were lysed in buffer A [50 mM Tris pH 7.5, 120 mM NaCl, 0.5% Nonidet P-40, and 10% (vol/vol) glycerol] supplemented with a protease inhibitor mixture (11873580001; Roche) and 1 mM β-glycerophosphate, 2.5 mM sodium pyrophosphate, 1 mM Na₃VO₄, and 10 mM NaF. WCEs were resolved by SDS/polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. Primary antibodies were mouse anti-GFP (sc-9996; Santa Cruz), mouse anti-vinculin (V9131; Sigma), mouse anti-tubulin (T5168; Sigma), a mouse monoclonal antibody against human EID1 that was raised against purified GST-EID1 as described previously (5), mouse anti-Flag M2 (F1804; Sigma), HRP-conjugated mouse anti-T7 (69048-3; Millipore), mouse anti-T7 (69522-3; Millipore), mouse anti-FBXO21 (ab119762; Abcam), mouse anti-HA (2999S; Cell Signaling Technology), mouse anti-Myc (2040s; Cell Signaling Technology), mouse anti-Cul1 (322400; Invitrogen), and rabbit anti-Skp1 (040p409E; Neomarkers). Bound antibodies were detected with horseradish peroxidase-conjugated secondary antibodies (Pierce) and by Immobilon Western chemiluminescent horseradish peroxidase substrate (Millipore) or SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific) according to the manufacturer’s instructions.

For immunoprecipitations, cell lysates were clarified by centrifugation at 18,000 × g in an Eppendorf centrifuge for 10 min at 4 °C. We then rocked 1 mg of clarified extract with 15 μL anti-Flag M2 beads (A2220; Sigma) or anti–c-Myc beads (A7470; Sigma) for 30 min at 4 °C. After five washes with NETN buffer [20 mM Tris∙HCl (pH 8.0), 100 mM NaCl, 1 mM EDTA, and 0.5% Nonidet P-40], bound proteins were eluted by boiling in SDS loading buffer and detected by immunoblot analysis.

For TAP tag immunoprecipitation, cells were washed thoroughly with cold PBS and resuspended in hypotonic buffer [10 mM Tris pH 7.3, 10 mM KCl, 1.5 mM MgCl₂, 10 mM β-mercaptoethanol, 1 mM β-glycerophosphate, 2.5 mM sodium pyrophosphate, 1 mM Na₃VO₄, and 10 mM NaF and 1× protease inhibitor (Roche)]. After 20 min of incubation on ice, the cells were disrupted by dounce homogenization. The homogenate was aliquoted to 1.5 mL Eppendorf tubes and centrifuged at 1,000 × g for 10 min at 4 °C. The supernatant was transferred to new 1.5 mL Eppendorf tubes and recentrifuged. The resulting crude cytoplasmic fraction was transferred to fresh Eppendorf tubes and the solute concentrations adjusted to that of Flag-IP buffer (50 mM Tris pH 7.5, 150 mM NaCl, 0.5% Nonidet P-40) and incubated with 20 μL anti-Flag M2 agarose (Sigma) for 30 min at 4 °C followed by three washes with Flag-IP buffer and two washes with HA–IP buffer (50 mM Tris pH 7.5, 150 mM NaCl, 0.05% Nonidet P-40). Bound proteins were eluted by three sequential incubations with 3×FLAG peptide (F3290, Sigma) (0.5 mg/mL, dissolved in HA–IP buffer) for 30 min at 4 °C. Next, the Flag eluates were pooled and incubated with anti-HA agarose (sc-7392AC; Santa Cruz) overnight at 4 °C with gentle rocking. The HA agarose was then washed with HA–IP buffer five times and finally eluted sequentially three times with HA peptide (PEP-101P; Covance) (0.4 mg/mL, dissolved in HA–IP buffer) for 30 min at 4 °C. We took 0.2% Flag eluates and 2% (vol/vol) HA eluates for Western blot analysis, whereas 2.5% (vol/vol) Flag eluates and 10% (vol/vol) HA eluates were used for silver staining analysis. The remainders of the HA eluates were analyzed by MS.

Mass Spectrometry.

Purified protein complexes were analyzed by MS (LC/MS-MS) as recently described (33) with minor modifications. Proteins from the tandem affinity purification (TAP) samples were directly processed in solution. Cysteine residues were reduced with 10 mM DTT for 30 min at 56 °C in the presence of 0.1% RapiGest SF (Waters). Proteins were digested overnight at 37 °C using 5 μg of trypsin after adjusting the pH to 8.0 with Tris. HA peptide was removed from the TAP samples by incubating the digests with 20 μL of a 50% (vol/vol) slurry of Activated Thiol Sepharose 4B (GE healthcare) for 30 min at room temperature. The HA peptide-depleted solutions were acidified by adding TFA to a final concentration of 1% and desalted by C18 solid phase extraction followed by Strong Cation Exchange, both performed in batch-mode format. Eluted peptides were concentrated in a vacuum concentrator and reconstituted with 20 μL of 0.1% TFA.

Purified tryptic peptides were analyzed by LC-MS/MS on an LTQ-Orbitrap-XL mass spectrometer (Thermo) equipped with a Digital PicoView electrospray source platform (New Objective) (34).

The spectrometer was operated in data-dependent mode where the top 8 most abundant ions in each MS scan were subjected to collisionally activated dissociation (CAD) (35% normalized collision energy) and subjected to MS2 scans (isolation width, 2.8 Da; threshold, 20,000). Dynamic exclusion was enabled with a repeat count of 1 and exclusion duration of 30 s. Electrospray ionization (ESI) voltage was set to 2.2 kV.

MS spectra were recalibrated using the background ion [Si(CH3)2O]6 at m/z 445.12 ± 0.03 and converted into a Mascot generic file format (.mgf) using multiplierz scripts (35, 36). Spectra were searched using Mascot (version 2.4) against three appended databases consisting of (i) human protein sequences (downloaded from RefSeq on 07/11/2011), (ii) common laboratory contaminants, and (iii) a decoy database generated by reversing the sequences from these two databases. For Mascot searches, precursor tolerance was set to 15 ppm and product ion tolerance to 0.6 Da. Search parameters included trypsin specificity, up to two missed cleavages, and fixed carbamidomethylation (C, +57 Da) and variable oxidation (M, +16 Da). Spectra matching to peptides from the reverse database were used to calculate a global false discovery rate and were discarded. Data were further processed to remove peptide spectral matches to the forward database with an FDR greater than 1.0%. Peptides shared by two or more genes were excluded from consideration when constructing the final protein list. Any protein identified in more than 1% of 108 negative TAP controls (37) was removed from the sets of interactors. Proteins identified in the control TAP experiments were also removed.

Peptide Pull-Down.

Biotinylated peptides were synthesized by the Keck Biotechnology Resource Center of Yale University and dissolved in 0.2 M Tris pH 8.0. Peptides (1 μg) were incubated with 1 mg of 293FT WCEs (prepared with buffer A) and 20 μL of High Capacity Streptavidin Agarose (20357; Pierce) for 2 h at 4 °C. The agarose was then washed five times with NETN buffer. Bound proteins were eluted by boiling in SDS loading buffer, resolved by SDS/PAGE, and detected by immunoblot analysis.

GST Pull-Down.

GST pull-down assays were performed basically as described previously (38). Escherichia coli extracts containing GST-EID1, GST-EID1(164–166AAA), or GST-EID1(1–157) fusion proteins were immobilized by incubation with glutathione-Sepharose (17-0756-01; Pharmacia) (prewashed with NETN buffer 3 times) for 30 min at 4 °C. The Sepharose was then washed three times with NETN buffer before incubation with 10 μL of Flag-FBXO21 in vitro translate or unprogrammed translate in 1 mL of NETN buffer for 1 h at 4 °C with gentle rocking. The Sepharose was washed five times with NETN, and bound proteins were eluted by boiling in SDS sample buffer and resolved by SDS/polyacrylamide gel electrophoresis. Comparable loading of GST fusion proteins was confirmed by Coomassie blue staining, and recovery of in vitro translated proteins was detected by immunoblotting.

Mammalian Two-Hybrid Assay.

The 293FT cells were transiently transfected in 12-well plates, in duplicate, with pG5luc (5×Gal E1b-luc) (39), TK-Renilla (Clontech), and Gal4dbd-FBXO21, or empty vector pM (Clontech) along with pVP16-EID1, pVP16-EID1(164–166AAA), or pVP16-EID1(1–157). The cells were lysed 24 h later, and the firefly and renilla luciferase activity were measured using Dual-Glo Luciferase Assay system (E2940; Promega). The ratio of FLuc/Rluc was calculated for each well.

In Vitro Ubiquitination.

The ³⁵S-labeled HA-EID1, HA-EID1(164–166AAA), or HA-EID1(1–157) in vitro translates were produced using the TNT T7 Coupled Reticulocyte Lysate System (L1170; Promega) according to the manufacturer’s instruction. The FBXO21 E3 ligase complex was immunopreciptated using anti-Myc Sepharose (A7470; Sigma) from 293FT cells that were transiently transfected with plasmids encoding Myc-tagged wild-type FBXO21 (myc-FBXO21), F-box–deleted FBXO21 (Myc-FBXO21ΔF), or empty vector. The Sepharose was washed three times with NETN buffer, twice with 1× ubiquitylation reaction buffer (URB) buffer (50 mM Tris pH 7.4, 5 mM KCl, 5 mM NaF, 5 mM MgCl₂, and 0.5 mM DTT), and then resuspended in 20 μL of 1× URB buffer containing 100 ng recombinant E1 (E-305; Boston Biochem), 250 ng UbcH5a (E2-616; Boston Biochem), 0.4 μM Ubiquitin aldehyde (U-201; Boston Biochem), 10 μg Ubiquitin (U-100H; Sigma), 1× energy regeneration system (ERS) (20 mM creatine phosphate, 0.2 μg/μL creatine phosphokinase, 5 mM Mg-ATP), 1× protease/phosphatase inhibitor (Roche), 10 nM N-acetyl-L-leucyl-L-leucyl-L-norleucinal (LLNL) (I-160; Boston Biochem), 0.5 mM ATP, and 3 μL ³⁵S-labeled EID1 (wild type or mutant) and incubated for 40 min at 30 °C.

In Vivo Ubiquitination.

The 293FT cells were transfected with plasmids encoding T7-EID1, T7-EID1(164–166AAA), or T7-EID1(1–157) together with plasmids encoding Flag-FBXO21 (or empty vector) and HA-ubiquitin. The cells, 21 h later, were treated with MG132 (10 μM) for 3 h before being scraped off the plates and washed twice with ice-cold PBS. The cells were transferred to 1.5 mL Eppendorf tubes, pelleted, and frozen at –80 °C after aspirating the residual liquid. For the ubiquitination assays, 100 μL of TSD buffer (50 mM Tris pH 7.5, 1% SDS, 5 mM DTT) was added to the cell pellets and boiled for 10 min. After centrifugation at 18,000 × g for 5 min at room temperature, the supernatants were transferred to fresh tubes and mixed with 1.2 mL of TNN buffer (50 mM Tris pH 7.5, 250 mM NaCl, 5 mM EDTA, 0.5% Nonidet P-40, 1× protease inhibitor) and precleared by incubation with 50 μL of protein G agarose (11243233001; Roche) for 1 h at 4 °C with gentle rocking. The cleared lysates were then immunoprecipitated with mouse anti-T7 antibody (Millipore). Bound proteins were then eluted by boiling in 1× SDS loading buffer, resolved by SDS/polyacrylamide gel electrophoresis, and detected by immunoblot analysis.

Real-Time RT-PCR.

Total RNA was extracted using RNeasy Mini Kit (74106; Qiagen) and reverse-transcribed into first-strand cDNA using AffinityScript QPCR cDNA Synthesis Kit (600559; Agilent) with a mixture of oligo (dT) and random primers. Real-time PCR was performed in duplicate using RT² SYBR Green ROX qPCR Mastermix (330523; Qiagen) and the Mx3000P QPCR system (Stratagene). Values were averaged to calculate the expression level and normalized against the level of 18S rRNA. PCR primers were as follows:

EID1	F	ACGAGGGCGAGGAATTTGATGACT	R	ATCCAGGGCTGGTTCTCTTGTTCT
FBXO21	F	TAAGGCCTTTCTTCAGCAGCCAGA	R	GATGCTGTCAATTTGGGCCTGGAT
FBXO1	F	CCAAAGCCTGTGCAAATGCAAACC	R	TGACAGCCTGGGAAGCCTGAAATA
FBXO2	F	TTTGAGTGGTGTCGCAAAGCACA	R	TTAACGGTGAGCTCGTAGAGGCA
FBXO5	F	TGCTATTTACAACGGGCAACCTG	R	AGGAGCTTGCCATCTGAACAGTC
FBXO7	F	AGCTGGTGTATCCTCTTCTGGCT	R	AACCTCCACAGGAGTGGGTCATT
FBXO11	F	CACAACACCACAATGCACACCACT	R	TTGATGGTGGGACATGCTCCTTGA
FBXO31	F	TCAAGCCTGGCCTCTTCAAAGGTA	R	GCTCATTGAAGTTGCGCTGGTTCT
FBXO32	F	AGGGCAGCTGGATTGGAAGAAGAT	R	AACGGAGCAGCTCTCTGGGTTATT
Cul1	F	TTCGCTTGCATTGGTGACTTGGAG	R	GTAGGGCCCTTTGCAAATGCATCA
Cul2	F	TGCTGTGTCCACTGGTTTACCTCA	R	TCCTGAGTAAGGTTGCTGGTTGCT
Cul3	F	TCAGCCACACCAAAGTGCAACATC	R	CTTGTGCACCTCCAACACCAACTT
Cul4a	F	TCTGCTTCCAGAAGAATGAGCGGT	R	GTGGCTTCTTTGTTGCCTGCTCTT
Cul4b	F	AAGCGCCTGTTAGTCGGAAAGAGT	R	GTGAAAGCAGCTCCGCATTCATGT
Cul5	F	TCGGATGTGCATGCAGTCTGTCTT	R	AGTACTCGTGCCTGTGCTTGCTTA
Cul7	F	ACCTCTACCGCAAACTCATCACCA	R	AGCACTTGTCCTCCTTCACTTCCA
MDM2	F	GGCAGGGGAGAGTGATACAGA	R	GAAGCCAATTCTCACGAAGGG
18S rRNA	F	AAGACGATCAGATACCGTCGTAG	R	GTTTCAGCTTTGCAACCATACTC

F, forward primer; R, reverse primer.

Protein Half-Life Analysis.

For CHX assays, 293FT cells were transfected with siRNAs. We added, 70 h later, 100 μg/mL CHX (C1988; Sigma) into the DMEM growth medium. At various time points thereafter, cell extracts were prepared using buffer A and subjected to immunoblot analysis.

RNA Interference.

Cells grown in 24-well plates were transfected with 80 nM siRNAs using Lipofectamine 2000 (11668019; Invitrogen). siRNAs were obtained from Thermo Scientific and listed as follows: siFBXO21 (M-012917-01-0005), siFBXO1 (M-003215-02-0005), siFBXO2 (M-012429-00-0005), siFBXO5 (M-012434-01-0005), siFBXO7 (M-013606-01-0005), siFBXO11 (M-012428-00-0005), siFBXO31 (M-016541-01-0005), siFBXO32 (M-013005-02-0005), siCul1 (M-004086-01-0005), siCul2 (M-007277-00-0005), siCul3 (M-010224-02-0005), siCul4a (M-012610-01-0005), siCul4b (M-017965-01-0005), siCul5 (M-019553-01-0005), siCul7 (M-017673-00-0005), siMDM2 (M-003279-04-0005), and siCONTROL (M-010230-00-0005).

Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) Genome Editing.

Duplex oligonucleotides encoding sgRNAs were cloned into the lentiCRISPR v2 vector (52961; Addgene) cut with BsmBI. The targeted FBXO21 sequences were as follow: (i) TCGCTGACGGCCGCCGACAT and (ii) GGAGACACGGCCGATGTCGG. The two MDM2-targeted sequences were as follows: GGTGTCACCTTGAAGGT and GTAGCAGTGAATCTACA. The nontargeting control sequences were CTTCCGCGGCCCGTTCAA and GGAGGCTAAGCGTCGCAA (a gift of Dr. Feng Zhang Broad Institute of MIT and Harvard, Cambridge, MA). Virus was generated by cotransfecting 293T cells with the lentiCRISPR vector together with psPAX2 and pMD2.G. Tissue culture supernatant was harvested 1 and 2 d later, pooled, and passed through a 0.45-μM filter. The filtrate was used to infect recipient cells. Successfully infected cells were selected by growth in the presence of 1 μg/mL puromycin for 3 d. The surviving cells were maintained as polyclonal pools or, for 293FT, seeded to obtain single clones by serial dilution. Target gene expression was monitored by immunoblot analysis. For A549 cells, maximal induction of p53 was observed 5 d after infection with the viruses encoding the MDM2 sgRNAs.

T-Cell Stimulation.

Human T cells were isolated from fresh peripheral blood mononuclear cells of normal healthy donors with the EasySep Human T Cell Enrichment Kit (19051; Stemcell Technologies) by negative selection according to the manufacturer’s instructions. The purified T cells were activated by Dynabeads Human T-activator CD3/CD28 (111.31D; ThermoFisher Scientific) with a bead-to-cell ratio of 2:1. The cells were pelleted at 600 × g and washed once with PBS before being stored at –80 °C before analysis.