Proteasome-dependent Degradation of Transcription Factor Activating Enhancer-binding Protein 4 (TFAP4) Controls Mitotic Division*

Background: TFAP4 is a transcription factor that controls cell proliferation, stemness and epithelial-mesenchymal transition and is up-regulated in colorectal cancer. Results: TFAP4 is targeted for degradation by the SCFβTrCP ubiquitin ligase. Failure to degrade TFAP4 leads to aberrant mitosis. Conclusion: TFAP4 degradation is required for the fidelity of mitosis. Significance: Misregulation of TFAP4 might contribute to genomic instability and tumorigenesis. TFAP4, a basic helix-loop-helix transcription factor that regulates the expression of a multitude of genes involved in the regulation of cellular proliferation, stemness, and epithelial-mesenchymal transition, is up-regulated in colorectal cancer and a number of other human malignancies. We have found that, during the G2 phase of the cell division cycle, TFAP4 is targeted for proteasome-dependent degradation by the SCFβTrCP ubiquitin ligase. This event requires phosphorylation of TFAP4 on a conserved degron. Expression of a stable TFAP4 mutant unable to interact with βTrCP results in a number of mitotic defects, including chromosome missegregation and multipolar spindles, which eventually lead to the activation of the DNA damage response. Our findings reveal that βTrCP-dependent degradation of TFAP4 is required for the fidelity of mitotic division.

TFAP4, a basic helix-loop-helix transcription factor that regulates the expression of a multitude of genes involved in the regulation of cellular proliferation, stemness, and epithelialmesenchymal transition, is up-regulated in colorectal cancer and a number of other human malignancies. We have found that, during the G 2 phase of the cell division cycle, TFAP4 is targeted for proteasome-dependent degradation by the SCF ␤TrCP ubiquitin ligase. This event requires phosphorylation of TFAP4 on a conserved degron. Expression of a stable TFAP4 mutant unable to interact with ␤TrCP results in a number of mitotic defects, including chromosome missegregation and multipolar spindles, which eventually lead to the activation of the DNA damage response. Our findings reveal that ␤TrCP-dependent degradation of TFAP4 is required for the fidelity of mitotic division.
Transcription factor activating enhancer-binding protein 4 (TFAP4) 3 is a ubiquitously expressed basic helix-loop-helix leucine-zipper transcription factor that binds to the consensus E-box sequence 5Ј-CAGCTG-3Ј (1). It has been reported that TFAP4 controls the expression of numerous genes regulating cell proliferation (2,3). TFAP4 target genes also include stem cell markers such as LGR5 and regulators of epithelial-mesenchymal transition such as SNAIL and E-cadherin (2,3). Importantly, TFAP4 protein levels have been found elevated in human colorectal, hepatocellular, and gastric carcinoma (3)(4)(5). Despite the described role of TFAP4 in cancer, the molecular mechanisms controlling TFAP4 protein levels and the posttranslational modifications regulating TFAP4 stability are currently unknown.
In the present study, we identify TFAP4 as a new interactor of the F-box protein ␤TrCP (␤-transducin repeat-containing protein) in a mass spectrometry-based screen aimed at the identification of substrates of SCF ␤TrCP , a multisubunit Cullin-RING ubiquitin ligase composed of a cullin scaffold, Cul1, the adaptor protein Skp1, and the RING subunit Rbx1. It is well established that, through its WD40 repeats, ␤TrCP recognizes a diphosphorylated motif with the consensus DpSG⌽X(X)pS in which the serine residues are phosphorylated to allow binding to ␤TrCP (6). Notably, work by several groups has demonstrated that SCF ␤TrCP plays a fundamental role in the regulation of many cellular processes that are related to cancer through targeted destruction of its specific substrates (7,8).
We also show that during the G 2 phase of the cell cycle, SCF ␤TrCP targets TFAP4 for proteasomal degradation and demonstrate that defective degradation of TFAP4 results in aberrant mitotic divisions and chromosome missegregation, which lead to the activation of the DNA damage checkpoint.

EXPERIMENTAL PROCEDURES
Cell Culture, Synchronization, and Drug Treatment-HeLa, HEK293T, 293GP2, U2OS, hTERT-RPE1, DLD1, and HCT116 cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% fetal calf serum. HeLa and hTERT-RPE1 cells were synchronized at G 1 /S by thymidine treatment as described (9). DLD1 cells were synchronized at the G 1 /S transition by using FBS-depleted medium for 32 h followed by treatment with aphidicolin (2 g/ml) for another 24 h. The following drugs were used: nocodazole (0.1 g/ml) to arrest cells in prometaphase, MG132 (10 M) to inhibit the proteasome, cycloheximide (100 g/ml) to block protein synthesis and doxycycline (1 g/ml) to induce the expression of TFAP4 (wild type or mutants).
Biochemical Methods-Immunoprecipitation and immunoblotting were performed by standard methods and have been described previously (10 -12).
Purification of ␤TrCP2 Interactors-HEK293T cells were transfected with pcDNA3-FLAG-HA-␤TrCP2 or pcDNA3-FLAG-HA-␤TrCP2(R447A) and treated with 10 M MG132 for 6 h. Cells were harvested and subsequently lysed in lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, plus protease and phosphatase inhibitors). ␤TrCP2 was immunopurified with anti-FLAG agarose resin (Sigma-Aldrich). After washing, proteins were eluted by competition with FLAG peptide (Sigma-Aldrich). The eluate was then subjected to a second immunopurification with anti-HA resin (12CA5 monoclonal antibody cross-linked to protein G-Sepharose; Invitrogen) prior to elution in Laemmli sample buffer. The final eluate was separated by SDS-PAGE, and proteins were visualized by Coomassie colloidal blue. Bands were sliced out from the gels and subjected to in-gel digestion. Gel pieces were then reduced, alkylated, and digested according to a published protocol (13). For mass spectrometric analysis, peptides recovered from in-gel digestion were separated with a C18 column and introduced by nanoelectrospray into the LTQ Orbitrap XL (Thermo Fisher) with a configuration as described (14). Peak lists were generated from the MS/MS spectra using MaxQuant build (version 1.0.13.13) (15) and then searched against the IPI Human database (version 3.37, 69164 entries) using Mascot search engine (Matrix Science). Carbaminomethylation (ϩ57 Da) was set as fixed modification, and protein N-terminal acetylation and methionine oxidation were set as variable modifications. Peptide tolerance was set to 7 ppm, and fragment ion tolerance was set to 0.5 Da, allowing two missed cleavages with trypsin enzyme. Finally, Scaffold (version Scaffold_3.6.1, Proteome Software, Inc.) was used to validate MS/MS-based peptide and protein identifications. Peptide identifications were accepted if their Mascot scores exceeded 20.
Transient Transfections and Retrovirus-and Lentivirus-mediated Gene Transfer-HEK293T cells were transfected using the calcium phosphate method as described (11). HCT116 cells were transfected with Lipofectamine according to the manufacturer's instructions. Retrovirus-mediated gene transfer was described previously (11,12). For lentivirus production, HEK293T cells were co-transfected with pHAGE2-EF1␣ and packaging vectors by using polyethylenimine. Virus-containing medium was collected 48 h after transfection and supplemented with 8 g/ml polybrene. Cells were incubated with virus-containing medium for 6 h for 2 consecutive days.
Gene Silencing by Small Interfering RNA-The sequences and validation of the oligonucleotides corresponding to ␤TrCP1 and ␤TrCP2 were described previously (9,16,17). Cells were transfected with the oligonucleotides twice (24 and 48 h after plating) using Oligofectamine (Invitrogen) according to the manufacturer's recommendations. Forty-eight hours after the last transfection, lysates were prepared and analyzed by SDS-PAGE and immunoblotting.
Phosphorylation Analysis by Mass Spectrometry-Samples were reduced with 10 mM DTT for 30 min at 60°C, followed by addition of iodoacetamide to 20 mM followed by a 30-min incubation in the dark at room temperature. The first digestion was performed using Lys-C for 4 h at 37°C. Subsequently, the digest was diluted 5-fold using 50 mM ammonium bicarbonate to a final urea concentration of Ͻ2 M, and a second digestion with trypsin was performed overnight at 37°C. Finally, the digestion was quenched by addition of formic acid to a final concentration of 0.1% (v/v). The resulting solution was desalted using 200 mg Sep-Pak C18 cartridges (Waters Corp.), lyophilized and reconstituted in 10% formic acid. LC-MS/MS was performed with both collision-induced dissociation and electron transfer dissociation in the form of data-dependent decision tree (19,20). MS spectra to peptide sequence assignment is performed with Proteome Discoverer (version 1.3), with MASCOT (version 2.3) as search engine and the localization of phosphorylated sites was evaluated with PhosphoRS (version 2) (21).
Immunofluorescence-Cells were plated and cultured on coverslips coated with polylysine (Sigma-Aldrich) with complete medium. Cells were fixed and permeabilized in 100% methanol at Ϫ20°C for 10 min and then incubated with the primary FITC-conjugated antibody for 1 h at room temperature in 0.5% Tween 20 in PBS (TBST). Cells were washed three times in 0.5% TBST. Hoechst 33342 (Sigma-Aldrich) was included to detect the nuclei. Slides were mounted by using Vectashield mounting media (Vector Labs).
Live Cell Imaging-Cells were plated and cultured in 24-well plates in complete medium. Microscope was equipped with an  incubating chamber at 37°C and a CO 2 supply. Images were taken at 10-min intervals over 24 h.

RESULTS
Structural studies have shown that arginine 474 in the WD40 ␤-propeller of ␤TrCP1 contacts the DpSG⌽X(X)pS destruction motif of ␤TrCP substrates (22). It has been also reported that the ␤TrCP1(R474A) mutant is not able to promote polyubiquitylation of its substrate IB␣ in an in vitro ubiquitylation assay (22). We confirmed that the equivalent ␤TrCP2 mutant (␤TrCP2(R447A)) does not bind to previously established ␤TrCP substrates such as ␤-catenin, REST (repressor element-1-silencing transcription factor), and PDCD4 (programmed cell death-4), while retaining the ability to interact with the SCF subunits Cul1 and Skp1 (Fig. 1A). We took advantage of these properties of ␤TrCP and used a differential immunopurification strategy followed by mass spectrometry analysis to identify novel substrates of SCF ␤TrCP . We expressed either wild type ␤TrCP2 or the ␤TrCP2(R447A) mutant (both FLAG-HA epitope-tagged at the N terminus) in HEK293T cells and purified ␤TrCP2 immunoprecipitates, which were then analyzed by mass spectrometry. Six unique peptides corresponding to the basic helix-loop-helix leucine zipper transcription factor TFAP4 were recovered in wild type ␤TrCP2 immunoprecipitates, but not in ␤TrCP2(R447A) immunopurifications (Fig. 1B).
To confirm the binding between ␤TrCP and TFAP4, we immunoprecipitated FLAG-tagged wild type ␤TrCP2 and assessed its binding to endogenous TFAP4 by immunoblotting. As shown in Fig. 1C, wild type ␤TrCP2, but not the ␤TrCP2(R447A) mutant, coimmunoprecipitated with endogenous TFAP4. Similar results were obtained with ␤TrCP1 (Fig. 1D).
To assess the specificity of the ␤TrCP-TFAP4 interaction, we immunoprecipitated a number of FLAG epitope-tagged F-box proteins as well as the related proteins Cdh1 and Cdc20 from HEK293T cells and analyzed their ability to pull down endogenous TFAP4. ␤TrCP1 and ␤TrCP2 coimmunoprecipitated with endogenous TFAP4 (Fig. 1E), whereas other members of the FBXW family of F-box proteins, FBXW2, FBXW4, FBXW5, FBXW7, FBXW8, or the APC/C activators Cdh1 and Cdc20 (also containing WD40 repeats) did not.
When we immunopurified FLAG-HA epitope-tagged TFAP4, we recovered unique peptides corresponding to subunits of T  r  C  P  b  i  n  d  i  n  g  s  i  t  e  s  I  B  (  H  s  )  2  8  -D  R  H  D  S  G  L  D  S  M  K  D  -3  9  E  M  I  1  (  H  s  )  1  4  1  -L  Y  E  D  S  G  Y  S  S  F  S  L  -1  5  2  C  a  t  e  n  i  n  (  H  s  )  2  9  -S  Y  L  D  S  G  I  H  S  G  A  T  -4  0   W  e  e  1  (  H  s  )  To test whether SCF ␤TrCP is directly responsible for the ubiquitylation of TFAP4, we reconstituted TFAP4 ubiquitylation in vitro. Immunopurified wild type ␤TrCP1, but not ␤TrCP1(⌬Fbox), an inactive mutant that lacks the F-box domain, triggered efficient ubiquitylation of TFAP4 in vitro (Fig. 1G).
Substrates of SCF ␤TrCP contain a conserved DpSG⌽X(X)pS degron that mediates the interaction with ␤TrCP (6, 7, 22). TFAP4 has a modified motif in which the first serine residue is replaced by glutamic acid as other established substrates of SCF ␤TrCP ( Fig. 2A) (11,23,24). We generated a mutant TFAP4 protein in which glutamic acid 135 and serine 139 were substituted with alanine ( Fig. 2A) and examined its ability to interact with ␤TrCP. The TFAP4(E135A/S139A) mutant did not bind endogenous ␤TrCP in immunoprecipitation experiments (Fig.  2B) and was markedly stabilized as demonstrated by half-life experiments carried out in various cell lines (Fig. 2, C-E).
To test whether Ser-139 is phosphorylated in cultured cells, we purified ␤TrCP2 from HEK293T cells treated with the proteasome inhibitor MG132 and analyzed the phosphorylation of co-immunoprecipitating TFAP4 by tryptic digestion followed by phosphopeptide enrichment and LC-MS/MS. Analysis of the recovered TFAP4 phosphopeptides demonstrated that Ser-139 is phosphorylated in cells (Fig. 2F).
As it has been reported that TFAP4 controls the expression of many cell cycle regulators (2, 3), we hypothesized that the protein levels of TFAP4 might oscillate during cell cycle progression. To test this hypothesis, we analyzed by immunoblotting the abundance of TFAP4 throughout the cell cycle. HeLa cells were synchronized at the G 1 /S transition by double thymidine block before releasing them into fresh medium (to analyze TFAP4 protein levels in the following S, G 2 , and M phases) and in mitosis by the nocodazole block-and-release procedure (Fig.  3, A and B). Synchronization was monitored by immunodetection of cyclin A (an S/G 2 phase protein) and phospho-histone H3 (a mitotic marker) as well as by flow cytometry (data not shown). We found that the abundance of TFAP4 oscillates during the cell cycle. The levels of TFAP4 decreased in early G 2 (at a time when the levels of cyclin A, which is degraded in early mitosis, were still elevated), remained low in mitosis and in the following G 1 , and increased at G 1 /S preceding cyclin A up-regulation. Similar kinetics of TFAP4 decrease in early G 2 was observed in other cell types such as non-transformed hTERTimmortalized retinal pigment epithelial (hTERT-RPE1) cells (Fig. 3C) and DLD1 cells (Fig. 3D). The proteasome inhibitor MG132 prevented the disappearance of TFAP4 in HeLa cells arrested in prometaphase, indicating that the degradation of TFAP4 in G 2 is mediated by the proteasome (Fig. 3E).
To test whether the degradation of TFAP4 is mediated by ␤TrCP, we silenced the expression of both ␤TrCP1 and ␤TrCP2 by RNAi (18,25) in HeLa cells. Fig. 3F shows that the knockdown of ␤TrCP caused accumulation of TFAP4 in prometaphase, whereas it had no effect in cells arrested at G 1 /S and a moderate effect in asynchronous cells. . C, hTERT-RPE1 cells were synchronized at G 1 /S by double thymidine block and released into nocodazole containing medium. D, to synchronize DLD1 cells at G 1 /S, cells were first serum starved (32 h) and then incubated in serum-containing medium in the presence of aphidicolin for 24 h. Cells were then released from the G 1 /S block in nocodazole-containing medium. Cells were collected at the indicated times, lysed, and analyzed by immunoblotting with the indicated antibodies. Actin or Cul1 were used as loading controls. E, asynchronously growing or nocodazole-treated HeLa cells were collected and analyzed by immunoblotting with antibodies for the indicated proteins. When indicated, cells were treated with the proteasome inhibitor MG132 for 6 h. F, HeLa cells were transfected with siRNA corresponding to a non-relevant mRNA (control siRNA) or to ␤TrCP mRNA. Twelve hours after transfection, cells were either left untreated of treated with thymidine (to synchronize them at G 1 /S) or nocodazole (to synchronize them in prometaphase). Cells were collected, lysed, and analyzed by immunoblotting. G, HeLa cells, transfected as in F, were synchronized as in A. Cells were then collected at the indicated time points and analyzed by immunoblotting with antibodies for the indicated proteins. Actin levels are shown as loading control. The asterisk indicates a nonspecific band.  To pinpoint when ␤TrCP targets TFAP4 for proteasome-dependent degradation, ␤TrCP-silenced cells were released from a thymidine block and collected at different time points. As shown in Fig. 3G, the knockdown of ␤TrCP prevented the destruction of TFAP4 occurring ϳ4 -6 h after cells were released from the thymidine block.
To study the function of the ␤TrCP-mediated degradation of TFAP4 during cell cycle, we transduced cells with lentiviruses expressing wild type TFAP4 and the degradation-resistant TFAP4(E135A/S139A) mutant. Cells approaching mitosis were then analyzed by immunoblotting (Fig. 4A). As expected, although levels of wild type TFAP4 decreased as cells moved toward mitosis, levels of the TFAP4(E135A/S139A) mutant remained unchanged. Moreover, compared with cells expressing wild type TFAP4, cells expressing the non-degradable TFAP4 mutant displayed a slower kinetics of mitotic entry as indicated by the phosphorylation of histone H3 on Ser-10.
To examine whether the late mitotic entry of cells expressing the non-degradable TFAP4 mutant was caused, at least in part, by a delay in G 2 progression, we employed FUCCI (26) live cell imaging. This system is based on the expression of both a red (RFP) and green (GFP) fluorescent protein fused to Cdt1 and geminin, respectively. Based on the fact that Cdt1-RFP is targeted for degradation in G2 by the SCF Skp2 ubiquitin ligase and geminin-GFP is degraded in late M and early G 1 by APC/C Cdh1 , cells are red in G 1 , become yellow upon entry in S phase (where both fusion proteins are present), are green throughout S and G 2 , and are colorless at the end of mitosis. We generated stable transfectants of HCT116-FUCCI cells in which the expression of TFAP4 (wild type or TFAP4(E135A/S139A)) could be pulsed-induced by doxycycline. These cells were then analyzed by live cell imaging. As shown in Fig. 4, B and C, cells expressing wild type TFAP4 and the TFAP4(E135A/S139A) mutant displayed a similar Cdt1-positive G 1 phase; however, cells expressing the non-degradable TFAP4(E135A/S139A) mutant showed a prolonged geminin-positive S/G 2 (9.9 h) when compared with cells expressing wild type TFAP4 (6.8 h).
In the course of our experiments, we observed that the expression of the non-degradable TFAP4(E135A/S139A) mutant in HCT116 cells, which have a relatively stable karyotype, resulted in notable nuclear atypia, including lobulated and enlarged nuclei (Fig. 4, D and E). This phenotype becomes increasingly penetrant with longer time of doxycycline treatment (which triggers the expression of wild type TFAP4 or TFAP4(E135A/S139A)), eventually occurring in virtually all cells. Notably, although less pronounced, nuclear atypia was also present in cells overexpressing wild type TFAP4.
To test whether these cellular abnormalities were caused by defective mitotic divisions, we expressed wild type TFAP4 and the non-degradable TFAP4(E135A/S139A) mutant in HCT116 cells expressing enhanced green fluorescent protein-labeled histone H2B, which were then analyzed by time-lapse microscopy. We found that cells expressing TFAP4(E135A/S139A) displayed chromosome missegregation (Fig. 4F). The percentage of mitotic cells with chromosome missegregation was 28% (n ϭ 100) for the non-degradable TFAP4 and 14% (n ϭ 100) for wild type TFAP4. In addition, HCT116 cells expressing TFAP4(E135A/S139A) displayed a number of mitotic spindle abnormalities, including the formation of multipolar spindles, as judged by ␣-tubulin and Hoechst staining. The percentage of mitotic cells with multipolar spindles was ϳ30% (n ϭ 300) for the degradation-resistant TFAP4 mutant and close to 10% (n ϭ 300) for wild type TFAP4. Examples of mitoses with multipolar spindles in HCT116 and hTERT-RPE1 cells are shown in Figs. 4G.
Notably, the expression of the degradation-resistant TFAP4(E135A/S139A) mutant intensified the phenotypes observed in cells expressing wild type TFAP4, likely as a result of the persistence of TFAP4 expression in G 2 rather than TFAP4 overexpression. In fact, the same mitotic aberrations and nuclear defects were also observed when TFAP4(E135A/ S139A) was expressed at moderate levels in a retroviral vector (Fig. 4H).
It has been shown that missegregating chromosomes are often damaged during cytokinesis, triggering a DNA doublestrand break response (27). The observed chromosome segregation errors in cells expressing the degradation-resistant TFAP4 mutant (Fig. 4, D-G), as well as their slower kinetics of mitotic entry and prolonged progression through G2 (Fig. 4, A-C) prompted us to test whether failure to degrade TFAP4 results in the activation of the G 2 DNA damage checkpoint. As shown in Fig. 4I, expression of the TFAP4(E135A/S139A) mutant leads to accumulation of p53 and phosphorylation of Chk2 on threonine 68, two markers of damaged DNA, indicating that stabilizing mutations of TFAP4 induce the activation of the DNA damage response.

DISCUSSION
In sum, we have shown that during the G 2 phase of the cell division cycle, TFAP4 is targeted for degradation by the SCF ␤TrCP ubiquitin ligase. Notably, a genome-wide characterization of TFAP4-controlled genes by mRNA profiling and DNA binding analysis has been recently reported (2) and hundreds of TFAP4 target genes, which are either induced or repressed by TFAP4, have been identified. Gene ontology analysis revealed that genes encoding cell cycle regulators were highly enriched among the TFAP4-regulated genes suggesting that ␤TrCP-dependent degradation TFAP4 in G 2 is required to modulate the expression of a multitude of genes controlling cell cycle progression.
Evidence for overexpression of TFAP4 in colorectal, hepatocellular, and gastric carcinoma has been shown previously (3)(4)(5). Our study, demonstrating that overexpression of wild type TFAP4 and, more dramatically, the non-degradable TFAP4 mutant, leads to aberrant mitotic division, suggests a mechanism by which the misregulation of TFAP4 observed in cancer may contribute to genomic instability and tumor progression. As mutations in c-MYC, a direct activator of TFAP4, are frequent in cancer, an analysis of the occurrence of stabilizing and/or activating mutations in the TFAP4 gene in human tumors should be warranted.