Biochem. J. (2011) 437, 323–333 (Printed in Great Britain) doi:10.1042/BJ20101388 323

Oestrogen causes degradation of KLF5 by inducing the E3 ubiquitin ligaseEFP in ER-positive breast cancer cellsKe-Wen ZHAO*†1, Deepa SIKRIWAL*1, Xueyuan DONG*, Peng GUO*, Xiaodong SUN* and Jin-Tang DONG*2

*Department of Hematology and Medical Oncology, School of Medicine and Winship Cancer Institute, Emory University, 1365 Clifton Road, Atlanta, GA 30322, U.S.A., and†Key Laboratory of Cell Differentiation and Apoptosis of Chinese Ministry of Education, Department of Pathophysiology, Shanghai Jiaotong University School of Medicine,Shanghai, China 200025

KLF5 (Kruppel-like factor 5) is a multifunctional transcriptionfactor involved in cell proliferation, differentiation andcarcinogenesis. In addition to frequent inactivation in differenttypes of human cancers, including breast cancer, KLF5 has beenidentified as an essential co-factor for the TGF-β (transforminggrowth factor β) tumour suppressor. In our previous studydemonstrating a negative regulation of ER (oestrogen receptorα) function by KLF5 in breast cancer cells [Guo, Dong, Zhao,Sun, Li and Dong (2010) Int. J. Cancer 126, 81–89], wenoticed that oestrogen reduced the protein level of KLF5. Inthe present study, we have tested whether and how oestrogen/ER signalling regulates KLF5 protein. We found that oestrogencaused the degradation of KLF5 protein, and the degradation wassensitive to proteasome inhibitors, but not other inhibitors. Theoestrogen-inducible E3 ligase EFP (oestrogen-responsive fingerprotein) was identified as a key player in oestrogen-mediateddegradation of KLF5, as knockdown and overexpression of EFP

increased and decreased KLF5 protein levels respectively, andthe decrease continued even when protein synthesis was blocked.EFP-mediated degradation impaired the function of KLF5 ingene transcription. Although only unubiquitinated EFP interactedwith KLF5, overexpression of EFP appeared to prevent theubiquitination of KLF5, while resulting in heavy ubiquitinationof the E3 itself. Furthermore, ubiquitination of EFP interruptedits interaction with KLF5. Although the mechanism for how EFPdegrades KLF5 remains to be determined, the results of the presentstudy suggest that oestrogen causes the degradation of KLF5protein by inducing the expression of EFP in ER-positive breastcancer cells.

Key words: breast cancer, Kruppel-like factor 5 (KLF5),oestrogen, oestrogen receptor (ER), oestrogen-responsive fingerprotein (EFP).

INTRODUCTION

Kruppel is a segmentation gene in Drosophila melanogasterwhich encodes a protein composed of several zinc-fingermotifs [1]. KLFs (Kruppel-like factors) are mammalian proteinshomologous with the zinc-finger part of Kruppel. KLF5 (alsonamed IKLF or BTEB2) is a transcription factor with a proline-rich N-terminal region and a C-terminus that contains threeconsecutive zinc-finger motifs [2]. Preceding the zinc-fingermotifs is a short basic region that may contribute to theability of KLF5 to bind to a GC-rich promoter. By regulatingtarget genes such as PDGFα (platelet-derived growth factorα), PPARδ (peroxisome-proliferator-activated receptor δ), NF-κB (nuclear factor κB), cyclin D1, p15 and Myc [3–9], KLF5modulates a variety of cellular processes including proliferation,differentiation and apoptosis [4,6,8,10–13]. Consequently, KLF5has been implicated in various physiological and pathologicalprocesses, including carcinogenesis [13].

In epithelial proliferation, KLF5 has dual functions, beingstimulatory in basal-like cells but inhibitory in epithelial cellstreated with TGF-β (transforming growth factor β) [7–9,13–21]. One mechanism for the reversal of KLF5 function is thatTGF-β recruits p300 to acetylate KLF5, which alters the KLF5transcriptional complex and reverses KLF5 function in generegulation [8,9,14].

KLF5 appears to play a role in breast cancer, but its precisefunction remains to be determined. On one hand, the KLF5locus at chromosome 13 is frequently deleted in human breastcancer and its protein is degraded by the WWP1 (WW-domain-containing E3 ubiquitin protein ligase 1) oncogenic ubiquitin E3ligase [19,22,23], which suggests a tumour suppressor functionof KLF5. On the other hand, increased expression of KLF5 isassociated with the expression of the HER2 (human epidermalgrowth factor receptor 2) oncoprotein and shorter survival inpatients [24], which suggests an oncogenic function of KLF5in breast cancer. In our effort to understand the role of KLF5 inbreast cancer, we examined whether and how KLF5 affectsthe function of ER (oestrogen receptor α) in breast cancercells, and found that KLF5 inhibits oestrogen-promoted cellproliferation and gene regulation in ER-positive breast cancercells, which involves the interaction of KLF5 with the ER [25].We noticed during that study that treatment of ER-positive cellswith oestrogen (17β-oestradiol, E2) decreased protein levels ofKLF5.

EFP (oestrogen-responsive finger protein; also named ZNF147and TRIM25) is an oestrogen-responsive RING finger protein thathas an E3 ligase activity in both ubiquitination and ISGylation ofproteins [26,27]. Although present in the cytoplasm of normalmammary epithelial cells, EFP expression in breast cancer ispositively associated with lymph node metastasis and decreased

Abbreviations used: CHX, cycloheximide; CMV, cytomegalovirus; EFP, oestrogen-responsive finger protein; ER, oestrogen receptor α; FBS, fetal bovineserum; FGF-BP, fibroblast growth factor-binding protein; GFP, green fluorescent protein; HA, haemagglutinin; IP, immunoprecipitation; KLF, Kruppel-likefactor; mAb, monoclonal antibody; Ni-NTA, Ni2 + -nitrilotriacetate; RNAi, RNA interference; RT, reverse transcription; siRNA, small interfering RNA; TGF-β,transforming growth factor β; WWP1, WW-domain-containing E3 ubiquitin protein ligase 1.

1 These two authors contributed equally to this work.2 To whom correspondence should be addressed (email j.dong@emory.edu).

c© The Authors Journal compilation c© 2011 Biochemical Society

324 K.-W. Zhao and others

patient survival [28], and functional studies have demonstratedthat EFP plays an oncogenic role in breast cancer, at leastby targeting proteolysis of 14-3-3σ , a tumour suppressor thatsuppresses cell-cycle progression [29]. In the antiviral responseof human cells, interferon up-regulates EFP and induces itsconjugation with the ubiquitin-like protein ISG15 [30], and EFPis indeed an ISG15 E3 ligase for 14-3-3σ and its ISGylationenzyme activity is RING-domain-dependent [27]. Interestingly,EFP is also autoISGylated, and the autoISGylation negativelyregulates its ISG15 E3 ligase activity [31].

In the present study we characterized whether and how ERsignalling regulates KLF5. We found that ER signalling regulatedKLF5 at both the RNA and protein levels, mediating a dualregulation. However, the regulation of the RNA transcript wasobserved to be a late event, whereas KLF5 protein regulation wasobserved to be a prominent and early event. We found that ERsignalling down-regulated KLF5 protein through the proteasomemachinery in ER-positive breast cancer cells, and the oestrogen-induced EFP E3 ubiquitin ligase played a crucial role in ER-induced proteolysis of KLF5. EFP interacted with KLF5, butdid not cause obvious ubiquitination of KLF5. Rather, EFP itselfwas ubiquitinated, and its ubiquitination appeared to prevent theubiquitination of KLF5. These results suggest the dual regulationof KLF5 at both the RNA and protein levels by ER signallingand that the protein regulation of KLF5 precedes the RNA levelregulation of KLF5. KLF5 protein is down-regulated by ERsignalling through the EFP ubiquitin E3 ligase.

EXPERIMENTAL

Plasmid constructions and transient transfection

Expression plasmids pcDNA3-KLF5, pcDNA3-FLAG-KLF5(FLAG tag added to the N-terminus), pcDNA3-KLF5-FLAG(FLAG tag added to the C-terminus), HA (haemagglutinin)–ubiquitin (HA-tagged ubiquitin) and Myc–WWP1 (Myc-taggedWWP1) have been described in our previous study [32]. ThepcDNA3-FLAG-EFP plasmid (FLAG–EFP) was constructedusing a PCR-based approach with the following primers(restriction enzyme sites are underlined): 5′-GGGGGTACCATGGCAGAGCTGTGCCCCCTG-3′ (forward) and 5′-CCC-GAATTCCTACTTGGGGGAGCAGATGGAGAGTG-3′ (rev-erse). For the construction of the KLF5 plasmid pcDNA3-HA-KLF5-His, in which an HA tag was added to the N-terminusand a histidine tag was added to the C-terminus of KLF5, wefirst made the pcDNA3-HA-His vector by inserting a fragmentof DNA containing an HA tag, EcoRI, XhoI, HindIII, BamHI,NotI, PstI and XbaI cloning sites and a histidine tag into thepcDNA3.1 vector (Invitrogen) between restriction enzyme sitesNcoI and ApaI. The KLF5 cDNA was then cloned into thispcDNA3-HA-His vector between the HA and histidine tags atEcoRI and NotI sites using a PCR-based approach with thefollowing primers (restriction enzyme sites are underlined):5′-TTTGAATTCATGGCTACAAGGGTGC-3′ (forward) and5′-TTTGCGGCCGCGTTCTGGTGCCTCTTCATA-3′ (reverse).All of the plasmid transfections were carried out usingthe LipofectamineTM 2000 transfection reagent (Invitrogen)according to the manufacturer’s instructions.

Construction of EFP mutants

Deletion mutants of EFP were created by PCR-basedapproaches with FLAG–EFP as the template and FLAG-pcDNA3 as the vector. Primer sequences for creating deletion

mutants FLAG–EFP-M1 and FLAG–EFP-M2 are summarizedin Supplementary Table S1 (at http://www.BiochemJ.org/bj/437/bj4370323add.htm). To construct FLAG–EFP-cS andFLAG–EFP-kR substitution mutants, a PCR-directed muta-genesis method was used using FLAG–EFP as the template.PCR primers used for these mutants have been summarizedin Supplementary Table S2 (at http://www.BiochemJ.org/bj/437/bj4370323add.htm).

Cell culture and treatment

An ER-positive MCF-7 breast cancer cell line was purchasedfrom the A.T.C.C. and maintained in Eagle’s minimumessential medium supplemented with 10% FBS (fetal bovineserum), 0.01 M Hepes, 1 mM sodium pyruvate, 0.15 % sodiumbicarbonate, 0.45% glucose, and 1% pencillin/streptomycinsolution (10000 units of pencillin and 10000 μg/ml streptomycin;Invitrogen). ER-positive T-47D cells were also purchasedfrom the A.T.C.C. and maintained in RPMI 1640 mediumsupplemented with 10% FBS, 0.01 M Hepes, 1 mM sodiumpyruvate, 0.15% sodium bicarbonate, 0.45% glucose and1% penicillin/streptomycin solution. The COS-1 African greenmonkey kidney cell line, which is ER-negative, was alsopurchased from the A.T.C.C. and propagated following theprocedures described by the A.T.C.C. Oestrogen (E2), tamoxifen,the proteasome inhibitors MG132, MG115 and PS341, and thelysosome inhibitor ammonium chloride were purchased fromSigma. These reagents were dissolved and used following themanufacturer’s recommendations and our previously publishedpapers [22,33].

To investigate the direct role of oestrogen responsiveness of EFPand KLF5 degradation, MCF-7 and T-47D cells were maintainedin Phenol Red-free medium supplemented with 10% charcoal-dextran-stripped FBS for at least 3 days. Subsequently, the cellswere treated with 1 μM oestrogen (E2) and were harvested at theindicated times to isolate total RNA and protein for analysis. In asimilar way, MCF-7 cells cultured in hormone-free medium weretreated with 1 μM tamoxifen for 4 h, and subsequently total RNAand protein were isolated for analysis.

Western blot analysis and IP (immunoprecipitation)

A total of 5×105 cells were seeded on to each well ofa six-well plate, to which plasmids were transfected usingthe LipofectamineTM 2000 reagent (Invitrogen). At 48 h aftertransfection, cells were treated with 20 μM MG132 for 4 h,and collected with RIPA buffer [50 mM Tris/HCl (pH 8.0),150 mM NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate,0.1% SDS and 1% proteinase inhibitor I cocktail (Sigma)]for Western blot analysis. Anti-FLAG, anti-β-actin and anti-HArabbit polyclonal antibodies were purchased from Sigma. Anti-EFP mouse mAb (monoclonal antibody) was purchased from BDBiosciences. Anti-ubiquitin mAb and all secondary antibodieswere purchased from Cell Signaling Technology. For IP, celllysates were prepared from 60-mm dishes with 0.5 ml of RIPAbuffer, and incubated with 30 μl of FLAG-M2 beads with rotationat 4 ◦C overnight. Beads were then washed three times withwashing buffer [20 mM Tris/HCl (pH 8.0), 100 mM KCl, 5 mMMgCl2, 0.2 mM EDTA, 10% glycerol and 0.1% Tween 20, with10 mM 2-mercaptoethanol and 1% proteinase inhibitor cocktailadded just prior to use]. Protein precipitates were boiled with2×SDS loading buffer (Bio-Rad) and subjected to Western blotanalysis.

c© The Authors Journal compilation c© 2011 Biochemical Society

EFP mediates oestrogen-caused KLF5 degradation 325

siRNA (small interfering RNA) transfection

An siRNA with the sequence 5′-GCACUGGAUGAUG-UGAGAA-3′ was used to target EFP. An siRNA targetingthe luciferase gene with the sequence 5′-CUUACGCU-GAGUACUUCGAUU-3′ was used as a negative control. Bothof the siRNAs were chemically synthesized (Dharmacon). MCF-7 cells were transfected with 100 nM of each siRNA andtransfection was carried out using the siPORT Amine Reagent(Ambion) in six-well plates. At 48 h after transfection, totalprotein was collected for analysis using Western blotting.

CHX (cycloheximide) chase assay

COS-1 cells were seeded on to six-well plates at a density of5×105 cells per well. After incubating at 37 ◦C overnight, cellswere transfected with untagged KLF5 construct in the presence orabsence of EFP plasmid with LipofectamineTM 2000 (Invitrogen).At 24 h after transfection, cells were treated with 50 μg/ml CHXfor different time periods. Total proteins were collected andanalysed by Western blotting for KLF5, EFP and β-actin. Theintensity of protein bands on a film was measured using theImageJ program (http://rsb.info.nih.gov/ij), and the degradationcurves were plotted as described in the legend to Figure 2.

RNA extraction and real-time quantitative RT (reversetranscription)–PCR

An RNeasy kit (Qiagen) was used to isolate total RNA,and the iScript cDNA synthesis kit (Bio-Rad) was used toperform RT. SYBR-based real-time PCR was performed toexamine the expression of KLF5, EFP and β-actin. The primersused for the study are summarized in Supplementary TableS3 (at http://www.BiochemJ.org/bj/437/bj4370323add.htm). TheABI PRISM 7000 sequence detection system (AB AppliedBiosystems) was used to perform SYBR real-time PCR intriplicate. The �Ct method, in which β-actin was co-amplifiedto normalize the amount of RNA added to a reaction, was usedto determine threshold values according to previously publishedprocedures [34].

Ni-NTA (Ni2 + -nitrilotriacetate)–agarose purification

COS-1 cells were co-transfected with HA- and histidine-tagged KLF5 (pcDNA3-HA-KLF5-His), HA-tagged ubiquitin,and FLAG-tagged EFP or Myc-tagged WWP1 for 48 h. Aftertreatment with MG132 at 20 μM for 4 h, cells were washed withPBS and lysed in RIPA buffer containing 10 mM imidazole. A20 μl aliquot of Ni-NTA–agarose beads (Qiagen) was added tothe cell lysate and rotated at 4 ◦C for 6 h. Precipitates were washedthree times with washing buffer containing 20 mM imidazole,boiled with 2×SDS loading buffer and subjected to Western blotanalysis.

Promoter-luciferase reporter assay

The FGF-BP (fibroblast growth factor-binding protein) proximalpromoter ( − 128 to + 61) cloned into the pGL3-BASIC plasmidwas provided by Dr Ceshi Chen (Albany Medical College, Albany,NY, U.S.A.) [35]. MCF-7 or T-47D cells were seeded on to a 12-well plate at a density of 1×105 cells per well. The following daythese cells were transfected with the FGF-BP promoter-reporterconstruct (0.5 μg per well) and siEFP (75 nM final concentration)using LipofectamineTM 2000 reagent following the manufacturer’s

protocol. The siRNA for KLF5, siKLF5, has been describedpreviously [9], and the siRNA for luciferase (siLuc) was usedas a control. At 48 h after transfection, cells were lysed and aluciferase assay was carried out using the luciferase assay kitfrom Promega as described previously [36]. Experiments werecarried out in triplicate.

RESULTS

Oestrogen immediately down-regulates the protein, but not theRNA, of KLF5 in ER-positive MCF-7 breast cancer cells

In our previous study describing the role of KLF5 in regulatingER function, we noticed that oestrogen treatment decreased theprotein level of transfected KLF5 in the MCF-7 ER-positivebreast cancer cell line, whereas depletion of hormone in culturemedium slightly increased the protein level of KLF5 [25]. Thisresult led us to hypothesize that ER signalling might cause thedegradation of KLF5 protein. To test this possibility, we culturedMCF-7 cells in hormone-free medium, and then added oestrogen(E2) into the medium for different time periods. As shown inFigure 1(A), oestrogen (E2) treatment mediated a time-dependentdecrease in KLF5 protein level. Oestrogen (E2) decreased theamount of KLF5 protein as early as 2 h after treatment, whichcorresponded to the induction of EFP protein by oestrogen (E2)(Figure 1A). Similarly, in another ER-positive cell line T-47D,oestrogen (E2) treatment also mediated the time-dependentdecrease in KLF5 protein with a corresponding increase in EFPprotein (Figure 1B). KLF5 protein was observed to decrease asearly as 6 h after treatment with oestrogen (E2) with an increasein the EFP level. In another ER-positive cell line, ZR-75-1, thesame treatment also decreased KLF5 protein, but the effect wasnot detectable until 24 h after treatment (Supplementary FigureS1A at http://www.BiochemJ.org/bj/437/bj4370323add.htm).Consistently, induction of EFP protein was not noticeableuntil 24 h after oestrogen (E2) treatment in ZR-75-1 cells(Supplementary Figure S1A). Induction of EFP correlated with adecrease in KLF5 at different time points, suggesting a role forEFP in the down-regulation of KLF5 protein.

We also examined the effect of oestrogen (E2) treatment on theRNA levels of KLF5 and EFP at different time points. In MCF-7cells, oestrogen (E2)-induced EFP transcription was detectableas early as 2 h after treatment (Figure 1C), which is consistentwith the increase in its protein levels; in T-47D cells, oestrogen(E2) induced EFP expression as early as 4 h (Figure 1D),with the increase in its protein level at 6 h (Figure 1B); butin ZR-75-1 cells the induction was not detectable until 8 hafter treatment (Supplementary Figure S1B). In all of the celllines, the transcription of KLF5 showed no changes within24 h, but decreased at 48 or 72 h after oestrogen (E2) treatment(Figures 1C and 1D, and Supplementary Figure S1B), indicatingthat oestrogen (E2) also down-regulates KLF5 transcription, butthe decrease in RNA is not as early as the decrease in protein.

We also treated MCF-7 cells maintained in hormone-freemedium with the non-steroidal anti-oestrogen tamoxifen andoestrogen (E2) individually and in combination, and measured theprotein levels of KLF5. Treatment of cells with oestrogen (E2)for 4 h resulted in a decreased level of KLF5 protein (Figure 1E).The decrease in the KLF5 protein level was not seen when thecells were treated with either tamoxifen alone or a combination oftamoxifen with oestrogen (E2) (Figure 1E). The RNA expressionof KLF5 was not affected by either tamoxifen or oestrogen (E2)alone, or in combination (Figure 1F). We also treated MCF-7cells maintained in regular medium with the non-steroidal anti-oestrogen tamoxifen, and measured the KLF5 protein level.

c© The Authors Journal compilation c© 2011 Biochemical Society

326 K.-W. Zhao and others

Figure 1 Oetrogen treatment causes the degradation of KLF5 in ER-positive breast cancer cells, and the ER-induced EFP E3 ligase is involved

MCF-7 (A) or T-47D (B) cells grown in hormone-free medium and treated with 1 μM oestrogen (E2) for the indicated times were subjected to Western blotting to detect protein expression of KLF5,EFP and β-actin. MCF-7 (C) or T-47D (D) cells in (A) and (B) were subjected to real-time PCR for RNA expression of EFP and KLF5. (E and F) MCF-7 cells grown in hormone-free medium andtreated with 1 μM of the ER inhibitor tamoxifen (Tam) and/or oestrogen (E2) for 4 h were subjected to Western blotting for protein expression of KLF5, EFP and β-actin (E) and real-time PCR forRNA expression of EFP and KLF5 (F).

Although weak, treatment of cells with tamoxifen for 4 h or longerresulted in a detectable level of KLF5 protein (SupplementaryFigure S1C), and the RNA expression of KLF5 was not affected bytamoxifen treatment (Supplementary Figure S1D). These resultssuggest that inhibition of oestrogen/ER signalling leads to anincreased protein level of KLF5, whereas the RNA level is notaffected.

The oestrogen-responsive protein EFP, an ubiquitin E3 ligase, playsa role in oestrogen-mediated down-regulation of KLF5 protein

Previous studies have indicated that KLF5 is regulated by theUPP (ubiquitin–proteasome pathway). WWP1 [22,33] and Fbw7[37,38] have been identified as E3 ubiquitin ligases mediating theubiquitination and degradation of KLF5. However, neither WWP1nor Fbw7 are oestrogen- or ER-responsive (Supplementary FigureS1E). It is thus possible that one or more other E3 ligases mediate

oestrogen-induced degradation of KLF5. There are at least sevenubiquitin E3 ligases that are ER-responsive, including SKP2 (S-phase kinase-associated protein 2), Cul-4A, E6-AP, EFP, BCA2(breast-cancer-associated gene 2), MDM2 (murine double minute2) and RNF11 (RING finger protein 11). In our search for ER-responsive E3 ligases that are inducible by oestrogen and ERin ER-negative cells after transfection with ER and treatmentwith oestrogen (E2), we found that EFP was the only ER-responsive E3 ligase among these seven in ER-negative cells(X. Y. Dong, X. Fu, S. Fan, P. Guo, D. Su, X. Sun and J. T. Dong,unpublished work). We therefore tested whether EFP could be theE3 ligase involved in the degradation of KLF5. As expected,the expression of EFP was induced by oestrogen (E2) treatmentat both the RNA and protein levels in MCF-7, T-47D and ZR-75-1cells, although the induction in ZR-75-1 cells was slower than thatin MCF-7 and T-47D cells (Figures 1A–1D, and SupplementaryFigures S1A and S1B). The increase in EFP protein correlatedwith the decrease in KLF5 protein at different time points in

c© The Authors Journal compilation c© 2011 Biochemical Society

EFP mediates oestrogen-caused KLF5 degradation 327

Figure 2 Oestrogen-induced expression of EFP degrades KLF5 protein

(A and B) Transfection of siRNA against EFP (siEFP) into MCF-7 cells cultured in regular medium (A) or in hormone-free medium but treated with 1 μM oestrogen (E2) for 18 h (B) increases proteinlevels of KLF5, as determined by Western blot analysis. (C) MCF-7 cells grown in regular medium were treated with the proteasome inhibitors MG132, MG115 and PS341, or the lysosome inhibitorammonium chloride (NH4Cl), and protein expression of KLF5, EFP and β-actin was evaluated by Western blotting. (D) Different amounts of EFP plasmid (from 0 to 0.8 μg) were co-transfected with0.8 μg of KLF5 plasmid into COS-1 cells, and Western blotting was performed to detect protein levels of KLF5 and EFP in the presence ( + ) or absence ( − ) of MG132. (E) MCF-7 cells in regularmedium (left-hand panel) and COS-1 cells (right-hand panel) were transfected with expression plasmids for FLAG-tagged EFP and untagged KLF5, KLF5 with a FLAG tag at the N-terminus (F-KLF5),or KLF5 with a FLAG tag at C-terminus (KLF5-F), and cell lysates were analysed by Western blot analysis for protein expression of KLF5 and EFP. The presence or absence of EFP is indicated by ‘ + ’or ‘ − ’. (F and G) CHX chase assay of COS-1 cells transfected with KLF5 and vector control or EFP. Western blotting was performed with cells treated with CHX for the indicated times to detect proteinlevels of KLF5 and EFP (F). Signal intensities for bands of KLF5 and β-actin were quantified using the ImageJ program, and relative levels of KLF5 were plotted against times of CHX treatment (G).

these cell lines (Figures 1A and 1B, and Supplementary FigureS1A), suggesting a role for EFP in the degradation of KLF5.Consistently, treatment for 4 h with tamoxifen, an anti-oestrogenagent, interfered with the induction of EFP by oestrogen (E2) atboth the RNA and protein levels and the decrease in KLF5 proteinlevel (Figures 1E and 1F). Again, the RNA level of KLF5 was notaffected by tamoxifen with 4 h of treatment (Figure 1F).

To further test whether EFP functions in oestrogen-mediateddegradation of KLF5, we knocked down the expression ofEFP by RNAi (RNA interference) in MCF-7 cells grown ineither regular medium (Figure 2A) or hormone-free mediumsupplemented with oestrogen (E2) (Figure 2B), and analysedthe expression of KLF5 protein. Although it is more difficult

to detect KLF5 protein due to degradation in cells in normalmedium, we could still detect an increase in the protein level ofKLF5 upon the knockdown of EFP (Figure 2A). In MCF-7 cellscultured in hormone-free medium and treated with oestrogen (E2),knockdown of EFP also increased the protein level of KLF5 incells treated with or without oestrogen (E2) (Figure 2B). Similarexperiments were done in MCF-7 cells using different siRNAsagainst EFP, and it was observed that the knockdown of EFPdid increase the KLF5 protein level (Supplementary Figure S2at http://www.BiochemJ.org/bj/437/bj4370323add.htm). Theseresults indicate that the ER-responsive EFP E3 ubiquitin ligaseplays a role in oestrogen-mediated down-regulation of KLF5protein.

c© The Authors Journal compilation c© 2011 Biochemical Society

328 K.-W. Zhao and others

Figure 3 Detection of protein interaction between KLF5 and EFP by IP combined with Western blotting in COS-1 and MCF-7 cells

(A and B) COS-1 cells were co-transfected with KLF5 and FLAG-tagged EFP (FLAG–EFP) (A) or Myc-tagged WWP1 (B) for 48 h and treated with 20 μM MG132 for 4 h. Cell lysates were subjectedto Western blotting (Input) or to IP with an anti-FLAG antibody (for EFP) or an anti-Myc antibody (for WWP1) with subsequent Western blotting to detect proteins of KLF5 and EFP (EFP IP in A) orKLF5 and WWP1 (WWP1 IP in B). The WWP1 panel serves as a positive control. (C) MCF-7 cells grown in hormone-free medium for 2 days were lysed and subjected to IP with an anti-EFP antibodyor mouse IgG. Western blotting was performed to detect endogenous EFP and KLF5 in the precipitates (IP) or in cell lysates not subjected to IP (Input).

Protein degradation is responsible for EFP-mediateddown-regulation of KLF5

To determine whether the proteasome pathway is responsible forthe oestrogen-mediated decrease in KLF5 protein, we treatedMCF-7 cells grown in regular medium with the proteasomeinhibitors MG132, MG115 and PS-341. The lysosome inhibitorammonium chloride was also used. Although KLF5 protein washardly detectable in MCF-7 cells grown in regular medium,each of the three proteasome inhibitors enriched KLF5 proteinto a detectable level (Figure 2C). Cells treated with ammoniumchloride still had no detectable KLF5 (Figure 2C).

To further test whether protein degradation is responsible forEFP-mediated down-regulation of KLF5, we transfected KLF5plasmid and different amounts of EFP plasmid into COS-1 cellsin the presence or absence of the MG132 proteasome inhibitor.Ectopic expression of EFP caused a dose-dependent decreasein the KLF5 protein level, and treatment with MG132 enrichedprotein levels of KLF5, but did not completely inhibit the effectof EFP (Figure 2D). We also transfected expression plasmidsfor EFP and different forms of KLF5 (wild-type, N-terminallytagged and C-terminally tagged) into MCF-7 cells in regularmedium and found that ectopic expression of EFP dramaticallydecreased protein expression for each form of KLF5 (Figure 2E,left-hand panel). In a similar way, we transfected the same sets ofplasmids into the COS-1 monkey kidney cells, and again founda dramatic decrease in the KLF5 protein level in EFP-expressingcells (Figure 2E, right-hand panel). Because KLF5 expressionwas driven by the CMV (cytomegalovirus) promoter in theseexperiments, and the CMV promoter is a viral promoter notusually affected by the regulatory mechanisms in mammaliancells, these results suggest that protein degradation rather thantranscriptional regulation is responsible for the EFP-mediatedreduction in the KLF5 protein level. To examine whether KLF5degradation caused by EFP is a specific event, we transfectedexpression plasmids for GFP (green fluorescent protein) andKLF5 into COS-1 cells along with EFP, and found that ectopicexpression of EFP dramatically decreased the protein level ofKLF5, but had no effect on GFP protein (Supplementary FigureS3A at http://www.BiochemJ.org/bj/437/bj4370323add.htm). Wealso transfected a fusion expression plasmid in which GFP iscloned at the N-terminus of KLF5, and observed that ectopicexpression of EFP could no longer degrade the GFP–KLF5 fusionprotein (Supplementary Figure S3B).

To more definitely determine whether EFP decreases the KLF5protein level by protein degradation, we performed a CHX chaseassay by transfecting EFP and KLF5 into COS-1 cells, treatingcells with the protein synthesis inhibitor CHX, and measuring the

protein expression of KLF5 by Western blot analysis (Figure 2F)and signal quantification (Figure 2G). Consistent with previousfindings showing protein degradation of KLF5 in different cells[22,32,33], KLF5 protein expression decreased in EFP-expressingcells treated with CHX for 2, 4 and 6 h (Figures 2F and 2G),whereas the level of KLF5 protein did not change in cellswithout EFP expression (Figures 2F and 2G). The EFP-induceddecrease in KLF5 protein levels was 20%, 38% and 84% ofcontrol levels for CHX treatments of 2, 4 and 6 h respectively(Figure 2G). We also performed a CHX chase assay in MCF-7cells where we depleted hormone for 3 days and then treatedcells with oestrogen (E2) and CHX for the indicated timesand performed Western blotting to examine protein expression.We found that hormone depletion increased the level of KLF5protein, and treatment with CHX did not decrease the level ofKLF5 in the absence of oestrogen (Supplementary Figure S4 athttp://www.BiochemJ.org/bj/437/bj4370323add.htm). Similarly,CHX treatment did not alter the EFP level in the absenceof oestrogen (E2) (Supplementary Figure S4). In the presence ofoestrogen (E2), whereas the KLF5 level decreased considerablyand the EFP level increased as expected, CHX treatment unex-pectedly increased the KLF5 level and decreased the EFP level(Supplementary Figure S4), and the increase in KLF5 levels withdifferent CHX concentrations was significantly correlatedwith the decrease in EFP levels (P < 0.005, linear regressionand correlation test). These results suggest that EFP itself couldundergo protein degradation, and EFP degradation could then leadto increased KLF5 protein levels. These results further suggestthat EFP down-regulates KLF5 through protein degradation.

EFP interacts with KLF5 at the protein level

It is well established that, in the proteasome pathway, E3 ligasesinteract with substrates to transfer ubiquitin and cause theirdegradation by the proteasome. To test whether EFP interacts withKLF5, we transfected plasmids for KLF5 and FLAG-tagged EFPinto COS-1 cells, and performed an IP with FLAG-M2 beads.Expression of transfected genes were confirmed by Westernblotting (Figure 3A, Input). In the protein complex pulled down byFLAG-M2 beads against FLAG–EFP, KLF5 protein was detected(Figure 3A). We used WWP1 as a positive control, which waspreviously identified as an E3 ligase that interacts with KLF5 toubiquitinate and degrade KLF5 [22]. As expected, KLF5 wasdetected in the protein complex precipitated by an antibodyagainst WWP1 (Figure 3B).

To determine whether the KLF5–EFP interaction occursbetween endogenous KLF5 and EFP, we performed an IP with

c© The Authors Journal compilation c© 2011 Biochemical Society

EFP mediates oestrogen-caused KLF5 degradation 329

Figure 4 Characterization of KLF5 ubiquitination, EFP ubiquitination and protein interaction between EFP and KLF5 in COS-1 cells co-transfected withubiquitin and treated with MG132

COS-1 cells were transfected with different combinations of plasmids, as indicated above the images, for 48 h. MG132 treatment of 20 μM was applied to cells during the last 15 h of transfection.Cell lysates were subjected to Western blotting (Input, all panels), IP combined with Western blotting (IP, A–C), or Ni-NTA purification combined with Western blotting (KLF5 pull down, D) to detectdifferent proteins indicated at the left-hand side of each panel. HA-tagged ubiquitin plasmid and anti-HA antibody were used for ubiquitin (Ub), Myc-tagged WWP1 plasmid and anti-Myc antibodyfor WWP1, and FLAG-tagged EFP plasmid and FLAG-M2 beads for EFP. For KLF5, untagged pcDNA3-KLF5 plasmid was used in (A–C), whereas pcDNA3-HA-KLF5-His plasmid was used in (D).Anti-KLF5 serum was used to detect KLF5 protein in all panels. Arrows indicate the two EFP bands predominantly detected in cells transfected with EFP but not ubiquitin.

an EFP antibody in MCF-7 cells, which express both EFP andKLF5 (Figure 1). Because MCF-7 cells in regular medium havealmost no detectable KLF5 protein, whereas hormone depletioncould significantly enrich KLF5 protein (Figure 1A), we grewMCF-7 cells in hormone-free medium (Phenol Red-free mediumsupplemented with 10% charcoal-dextran-stripped FBS) for2 days before carrying out the IP. As a negative control for themouse anti-EFP mAb, a mouse IgG was used in the IP. As shownin Figure 3(C), KLF5 was detected in the protein complex pulleddown by the EFP antibody, indicating that endogenous EFP andKLF5 proteins interact with each other.

Although EFP itself is ubiquitinated, it does not appear toubiquitinate KLF5

To determine whether EFP ubiquitinates KLF5 as an E3 ligase,we further characterized its interaction with KLF5 in the presenceof ectopically expressed ubiquitin and the MG132 proteasomeinhibitor. We transfected COS-1 cells with KLF5, FLAG-taggedEFP and HA-tagged ubiquitin, treated cells with MG132, andperformed IP and Western blotting. Expression of KLF5, EFPand ubiquitin was confirmed by Western blotting (Figure 4,Input). We first precipitated ubiquitinated proteins using theanti-HA antibody against ubiquitin. As expected, a smear ofproteins was detected with the anti-HA antibody in each groupregardless of EFP and KLF5, and the smear was much darkerin precipitated samples than in input samples (Figure 4A, Ubpanel). For KLF5, its expression was strong, EFP did not makean obvious difference, and no ladder of ubiquitinated KLF5 was

visible in the input samples (Figure 4A, KLF5 panel, lanes 1and 3). In the precipitated ubiquitin complexes, on the otherhand, a ladder of ubiquitinated KLF5 was strong in cells withoutEFP, whereas no KLF5 band was visible in the unubiquitinatedform (Figure 4A, KLF5 panel, lane 4). Unexpectedly, expressionof EFP dramatically decreased the level of ubiquitinated KLF5,leaving only a weak band of KLF5 that appeared to have one unitof ubiquitin (Figure 4A, KLF5 panel, lane 6). For EFP, strongsmears were detected in cells transfected with EFP (Figure 4A,EFP panel, lanes 2 and 3), and the smears were darker in ubiquitinprecipitates (Figure 4A, EFP panel, lanes 5 and 6), indicating thatEFP is heavily ubiquitinated with the co-transfection of ubiquitin.

As a positive control for the ubiquitination of KLF5, we alsotransfected WWP1 into COS-1 cells and performed an IP with ananti-HA antibody, and Western blotting with antibodies againstWWP1, KLF5 and ubiquitin. Again, a smear of proteins wasdetected with the anti-HA antibody for ubiquitin in each groupregardless of WWP1 and KLF5, and the smear was much darkerin precipitated samples (Figure 4B, Ub panel, both Input and IP).For KLF5, it was again strongly expressed in the unubiquitinatedform in cells not subjected to IP, and there were no obvious bandsof ubiquitinated KLF5 (Figure 4B, KLF5 panel, lanes 1 and 3).A band of WWP1 was detected in WWP1-transfected cells notsubjected to IP (Figure 4B, WWP1 panel, lane 3), but no WWP1was detected in ubiquitin precipitate (Figure 4B, WWP1 panel,lane 6). In ubiquitin complexes precipitated with an anti-HAantibody against ubiquitin, a ladder of ubiquitinated KLF5 wasdetected in cells with and without WWP1 transfection (Figure 4B,KLF5 panel, lanes 4 and 6), and expression of WWP1 did notobviously decrease larger bands of ubiquitinated KLF5 while

c© The Authors Journal compilation c© 2011 Biochemical Society

330 K.-W. Zhao and others

dramatically decreasing the smallest band of ubiquitinated KLF5(close to the size of KLF5 with one unit of ubiquitin) (Figure 4B,KLF5 panel, lane 6).

We then precipitated EFP complexes with FLAG-M2 beadsin cells transfected with ubiquitin and different combinations ofKLF5 and EFP, and performed Western blot analysis to detecteach of the molecules. Consistent with the results in Figure 4(A),a smear of EFP was detected in cells transfected with EFP,regardless of IP or KLF5 expression (Figure 4C, EFP panel, bothInput and IP lanes), and similar signals were also detected usingthe anti-HA antibody for ubiquitin in the precipitate (Figure 4C,Ub panel). For KLF5, although strong KLF5 expression wasdetected in cells not subjected to IP (Figure 4C, KLF5 panel, lanes1 and 3), which is consistent with the results in Figures 4(A) and4(B), there was no detectable KLF5 signal in the EFP precipitates(Figure 4C, KLF5 panel, lanes 4 and 6). This result suggests thatubiquitinated EFP no longer interacts with KLF5.

We also examined the presence of EFP and ubiquitin inKLF5 protein complexes in cells transfected with histidine-taggedKLF5, HA-tagged ubiquitin and FLAG-tagged EFP or Myc-tagged WWP1. Cell lysates were incubated with Ni-NTA–agarosebeads to pull down KLF5 protein complexes, which were thensubjected to Western blotting. WWP1 was used as a positivecontrol that interacts with KLF5. Although expression of eitherWWP1 or EFP decreased the protein level of KLF5 in cells notsubjected to pull down, abundant KLF5 was detected in each of thepull-down groups (Figure 4D, KLF5 panel). In the KLF5 proteincomplex, both WWP1 and EFP were clearly detected (Figure 4D,pull down). It is worth noting that EFP bands detected in theKLF5 protein complex were primarily in an unubiquitinated form(Figure 4D, EFP panel, lane 6), which is different from thosedetected in EFP or ubiquitin precipitates.

Role of the RING finger domain of EFP in the degradation of KLF5

EFP comprises three essential domains: a RING finger domain,a B-box coiled-coil domain and a C-terminal SPRY domain [39].The essential E3 ligase activity of EFP is mostly conserved to itsRING finger domain and the cysteine residues in the RING finger.We examined the role of the RING finger domain in degradingKLF5. We generated some deletion mutants of EFP (Figure 5A),and analysed their effect on the degradation of KLF5 by Westernblotting. We also analysed the interaction between KLF5 and EFPby IP. As shown in Figure 5(B), the EFP-M1 mutant, comprisingonly the RING finger, could not interact with and degrade KLF5,suggesting that the RING finger alone is not sufficient for theEFP–KLF5 interaction and the degradation of KLF5. The EFP-M2 mutant, in which the RING finger domain was deleted, alsocould not degrade KLF5, although it still interacted with KLF5(Figure 5B). When the two conserved cysteine residues in theRING finger of EFP were replaced with serine residues, EFPcould no longer interact with KLF5, nor could it degrade KLF5(Figure 5C). The EFP-kR mutant, in which Lys117 was changedto arginine, and thus the ISGylation of EFP was interrupted asdemonstrated in a previous study [31], had no obvious effect on theEFP–KLF5 interaction and the degradation of KLF5 (Figure 5C).Therefore the degradation of KLF5 by EFP is dependent on bothits interaction with KLF5 and its RING finger domain, includingthe conserved cysteine residues.

EFP negatively regulates the function of KLF5

Among the KLF5-regulated genes, FGF-BP has been shown asa direct target of KLF5 that involves the binding of KLF5 tothe GC-rich/Sp1 site of the FGF-BP promoter [35]. Furthermore,

Figure 5 Role of different domains of EFP in the degradation of KLF5

(A) Schematic diagram depicting different mutants of FLAG–EFP in which different domains orresidues were deleted or mutated. aa, amino acids. (B and C) Effect of different mutations on theinteraction of EFP with KLF5, and the degradation of KLF5. FLAG–EFP or its mutants (M1 andM2 in B, or cS and kR in B) were co-transfected with KLF5 into COS-1 cells, and cell lysateswere subjected to IP with FLAG-M2 beads and immunoblotting with the indicated antibodies(right-hand panels). Input panels (left-hand side) were immunoblot results from cell lysates notsubjected to IP. Arrows indicate wild-type and mutant EFP proteins. In (C), the panels in themiddle and right-hand side were from the same cell lysates, whereas the panel at the left-handside was from another batch of cell lysates.

up-regulation of FGF-BP partially mediates the pro-proliferationfunction of KLF5 in breast cells [35]. Therefore we examinedthe effect of EFP on the function of KLF5 in the regulation ofFGF-BP1. Knockdown of EFP in MCF-7 and T-47D cells grownin regular medium, which did not change the RNA level of KLF5but increased the protein level of KLF5, significantly increased theRNA level of FGF-BP1 (Figure 6A). Consistently, oestrogen (E2)treatment in MCF-7 and T-47D cells, which significant decreasedKLF5 protein levels (Figures 1A and 1B), also decreased theRNA levels of FGF-BP in these cells (Figure 6B). With theFGF-BP1 promoter-luciferase activity as a readout, knockdownof EFP increased the promoter activity of FGF-BP1 in both MCF-7 and T-47D cells (Figure 6C). These results suggest that EFP cannegatively regulate the transcriptional function of KLF5.

c© The Authors Journal compilation c© 2011 Biochemical Society

EFP mediates oestrogen-caused KLF5 degradation 331

Figure 6 Negative regulation of KLF5 function by EFP in ER-positive breastcancer cells

(A) Knockdown of EFP by RNAi increased the expression of FGF-BP1 in MCF-7 or T-47D cellscultured in regular medium, as determined by real-time PCR. (B) Expression of FGF-BP1 in thesame MCF-7 or T-47D cells from Figure 1, which were treated with oestrogen (E2) for differenttimes, as measured by real-time PCR analysis. (C) Knockdown of EFP by RNAi increased thepromoter activity of FGF-BP1 in MCF-7 and T-47D cells cultured in regular medium. FGF-BPpromoter-reporter plasmid was co-transfected with siRNAs for 48 h, and the luciferase assay wasperformed. Control, no siRNA transfection; siLuc, siRNA against the luciferase gene (negativecontrol); siKLF5, siRNA against KLF5 (positive control); siEFP, siRNA against EFP.

DISCUSSION

Oestrogen causes KLF5 degradation primarily through theoestrogen-inducible E3 ubiquitin ligase EFP

In eukaryotic cells, the proteasome is the major intracellularproteolytic machinery for protein destruction [40,41]. Previousstudies have established that the KLF5 transcription factoris degraded by the proteasome, and the degradation could

be ubiquitin-dependent or -independent [32,33]. The WWP1E3 ubiquitin ligase has been demonstrated to interact with KLF5,causing its ubiquitination and degradation [22]. In the presentstudy, we found that in the ER-positive MCF-7, T-47D and ZR-75-1 breast cancer cell lines, the activation of oestrogen/ER signallingdown-regulated KLF5 protein while inducing the EFP ubiquitinE3 ligase (Figures 1A and 1B, and Supplementary Figure S1A),and inhibitors for proteasome, but not for lysosome, couldinterrupt ER-mediated down-regulation of KLF5 (Figure 2C).In MCF-7 cells grown in both regular medium and hormone-depleted medium supplemented with oestrogen (E2), knockdownof the oestrogen-inducible E3 ubiquitin ligase EFP increased en-dogenous KLF5 protein (Figures 2A and 2B, and SupplementaryFigures S2A–S2E), whereas ectopic expression of EFP decreasedectopically expressed KLF5 protein (Figure 2E). Furthermore,transfection of EFP into ER-negative COS-1 cells also decreasedthe protein level of transfected KLF5 (Figure 2E), and the decreasewas sensitive to the inhibition of protein synthesis (Figures 2Fand 2G). Functionally, the oestrogen-induced EFP impaired thefunction of KLF5 in the regulation of its target genes such asFGF-BP1, whereas knockdown of EFP enhanced the function ofKLF5 in gene regulation (Figure 6). These results not only indicatethat the oestrogen/ER signalling causes the degradation of KLF5,they also establish EFP as another E3 ligase that mediates thedegradation and modulates the function of KLF5.

Although EFP interacts with KLF5, it does not appear toubiquitinate KLF5, and the mechanism for EFP-mediated KLF5degradation remains to be determined

Proteasomal degradation often involves ubiquitination, in whichthe substrate is attached with a chain of polyubiquitin throughits interaction with an E3 ligase. Ubiquitinated substrates arethen recognized by the 19S regulatory complex of the 26Sproteasome for degradation. For example, the WWP1 HECT-domain-containing E3 ligase mediates the ubiquitination anddegradation of KLF5 through its interaction with KLF5 [22,42].Similarly, the RING-finger-containing EFP E3 ligase has beenshown to ubiquitinate 14-3-3σ to mediate its degradation [29].In the degradation of KLF5 by EFP, EFP interacted with KLF5at the protein level (Figures 3 and 4D), and such an interactionappeared to be necessary for KLF5 degradation, as deletion ofthe B-box coiled-coil domain of EFP interrupted its interactionwith and the degradation of KLF5 (Figures 5A and 5B). Asexpected for a typical E3 ligase in protein degradation, theRING domain and its conserved cysteine residues of EFP areessential for KLF5 degradation (Figure 5). Unexpectedly, EFP didnot cause detectable ubiquitination of KLF5 (Figure 4A, KLF5panel, lane 6). In fact, expression of EFP appeared to abolish theubiquitination of KLF5 (Figure 4A, KLF5 panel, lane 6), whichis different from the WWP1 positive control (Figure 4B, KLF5panel, lane 6). It is likely that in the presence of ubiquitin, EFPhas more preference of being ubiquitinated over KLF5.

We further noticed that EFP itself was ubiquitinated(Figure 4), and co-expression of exogenous ubiquitin enhancedthe ubiquitination of EFP (Figures 4A and 4C), which is consistentwith a previous report in which auto-ubiquitination of the EFPE3 ligase was described [43]. Unexpectedly, ubiquitination ofEFP prevented the interaction of EFP with KLF5, as no KLF5was detected in the protein complexes pulled down by an anti-EFP antibody (Figure 4C, KLF5 panel, lane 6) or by an anti-ubiquitin antibody (Figure 4A, KLF5 panel, lane 6), whereasstrong smears were detected for both ubiquitin and EFP. It isunlikely that the lack of KLF5 ubiquitination is due to excessivedegradation caused by EFP, because the MG132 proteasome

c© The Authors Journal compilation c© 2011 Biochemical Society

332 K.-W. Zhao and others

inhibitor was applied to cells and abundant unubiquitinatedKLF5 was detected in the input (Figures 4A–4C, Input forKLF5). Furthermore, in precipitates pulled down by an anti-KLF5 antibody (Figure 4D), the predominant form of EFP wasunubiquitinated. Therefore, in the complex between EFP andKLF5, it appears that neither EFP nor KLF5 is ubiquitinated.

In addition to its ubiquitin E3 ligase activity, EFP is also anISG15 E3 ligase that mediates the ISGylation of 14-3-3σ inviral responses [27]. Interestingly, EFP is autoISGylated, andits autoISGylation impairs its ISG15 E3 ligase activity [31].Although the ISGylation activity of EFP appears to be dispensablefor its function in KLF5 degradation in the present study where nointerferon treatment or viral infection was applied (the kR mutantin Figure 5), we speculate that ubiquitination of EFP could impairits function as an E3 ligase in the degradation of KLF5 similarto the effect of its autoISGylation and, when too much EFP isavailable, more EFP becomes ubiquitinated to decrease its activityin KLF5 degradation. It is even possible that ubiquitination of EFPleads to its protein degradation because, in the CHX chase assay,the EFP protein level decreased in CHX-treated MCF-7 cellsin the presence of oestrogen (Supplementary Figure S4). Whetherubiquitination of EFP causes its degradation and alters its functionin the degradation of KLF5 remains to be determined, as does howEFP mediates the degradation of KLF5.

Mutation of the two cysteine residues in the RING fingerdomain of EFP, which dramatically impairs the catalytic activityof EFP in its autoISGylation [27], almost abolished the interactionbetween KLF5 and EFP (Figure 5C), although deletion ofthe entire RING finger did not have an obvious effect on theinteraction (Figure 5B). One possibility is that a third protein,protein X, could be ubiquitinated by EFP and interact withboth EFP and KLF5, and ubiquitinated protein X could beresponsible for the transfer of KLF5 to the proteasome fordegradation. EFP with the cysteine mutations could still fold andbind to protein X, but could not efficiently ubiquitinate protein X.Consequently, mutated EFP could prevent other E3 ligases frombinding and ubiquitinating protein X, which could otherwise occuras compensation if the RING domain is deleted.

A negative regulatory relationship between KLF5 and oestrogen/ERsignalling

In our previous study, we found that KLF5 inhibits oestrogen-induced gene transcription and cell proliferation [25]. In thepresent study we found that oestrogen negatively regulates KLF5in two ways. First, oestrogen promptly induces one of the E3ligases, EFP, which causes protein degradation of KLF5, althoughthe mechanism of the degradation is unclear. Secondly, oestrogengradually down-regulates the transcription of KLF5, which takesa longer time to occur. Taken together, from our previous andpresent studies it appears that KLF5 and oestrogen signallingregulate each other through a feedback mechanism (Figure 7). Asthe presence of more KLF5 helps inhibit the function of oestrogenby interacting with the ER, control in the level of KLF5 isachieved by the up-regulation of EFP, which immediately down-regulates the protein level of KLF5 by protein degradation.Oestrogen/ER signalling also gradually down-regulates the RNAlevel of KLF5 for a more extensive control of KLF5 (Figure 7).

EFP is predominantly expressed in oestrogen-responsivetissues including mammary glands, uteri and osteoblasts [44–46].As an E3 ligase, EFP has been shown to ubiquitinate and degradea number of proteins including DDX58/RIGI [47], ER [43] and14-3-3σ [29], thus playing a critical role in the regulation ofvarious pathways. Functional and expression studies suggestan oncogenic role of EFP in human cancer [26,29]. KLF5,

Figure 7 Feedback regulation of KLF5 and oestrogen signalling in ER-positive breast cancer cells

Based on results from the present study and a previous study [25], we propose that, whereas KLF5inhibits oestrogen-induced cell proliferation and gene regulation through a protein interactionwith ER and thus disrupts the functional effect of oestrogen [25], oestrogen-ER signalling inhibitsKLF5 function by immediate protein degradation through the induction of the EFP E3 ligaseand gradual down-regulation of KLF5 transcription to accomplish tight control in the overallexpression of KLF5.

on the other hand, has been suggested as a tumour suppressorbecause it has frequent genomic deletion in breast cancer, itsre-expression inhibits the proliferation of breast cancer cells[19,48] and it participates in the inhibitory function of the TGF-βtumour suppressor in cell proliferation [8,9,14]. It also appearsto interact with oestrogen receptor β to suppress tumour growth[49]. Although the KLF5–EFP interaction could play a role in thestructure and function of normal mammary tissue, interruption ofthis interaction could contribute to mammary tumorigenesis.

In summary, we found that the oestrogen/ER signal regulatesKLF5 at both the RNA and protein level. This dual regulation isaccomplished at an early stage by increasing the EFP E3 ligaselevel, which mediates the degradation of KLF5 protein, and ata late stage by decreasing the RNA transcript level of KLF5.We found that the oestrogen/ER signal caused the degradation ofKLF5 in ER-positive MCF-7 breast cancer cells through the pro-teasome machinery. These results suggest that the oestrogen/ERsignal negatively regulates the KLF5 transcription factor, andEFP-mediated proteasome degradation is one of the mechanismsfor the negative regulation of KLF5 in breast epithelial cells.

AUTHOR CONTRIBUTION

Ke-Wen Zhao, Deepa Sikriwal, Xueyuan Dong, Peng Guo and Xiaodong Sun performedthe experiments, and analysed and interpreted the data. Ke-Wen Zhao, Deepa Sikriwal andJin-Tang Dong designed the research project, analysed the data, and drafted and finalizedthe paper and Figures.

FUNDING

This work was supported by the National Institutes of Health/National Cancer Institute[grant number R01 CA087921], and the Shanghai Municipal Education Commission[grant number 11ZZ106].

REFERENCES

1 Wieschaus, E., Nusslein-Volhard, C. and Kluding, H. (1984) Kruppel, a gene whoseactivity is required early in the zygotic genome for normal embryonic segmentation. Dev.Biol. 104, 172–186

2 Sogawa, K., Imataka, H., Yamasaki, Y., Kusume, H., Abe, H. and Fujii-Kuriyama, Y. (1993)cDNA cloning and transcriptional properties of a novel GC box-binding protein, BTEB2.Nucleic Acids Res. 21, 1527–1532

3 Aizawa, K., Suzuki, T., Kada, N., Ishihara, A., Kawai-Kowase, K., Matsumura, T., Sasaki,K., Munemasa, Y., Manabe, I., Kurabayashi, M. et al. (2004) Regulation of platelet-derivedgrowth factor-A chain by Kruppel-like factor 5: new pathway of cooperative activation withnuclear factor-κB. J. Biol. Chem. 279, 70–76

4 Oishi, Y., Manabe, I., Tobe, K., Tsushima, K., Shindo, T. and Fujiu, K. (2005) Kruppel-liketranscription factor KLF5 is a key regulator of adipocyte differentiation. Cell Metab. 1,27–39

c© The Authors Journal compilation c© 2011 Biochemical Society

EFP mediates oestrogen-caused KLF5 degradation 333

5 Chanchevalap, S., Nandan, M. O., McConnell, B. B., Charrier, L., Merlin, D., Katz, J. P.and Yang, V. W. (2006) Kruppel-like factor 5 is an important mediator forlipopolysaccharide-induced proinflammatory response in intestinal epithelial cells.Nucleic Acids Res. 34, 1216–1223

6 Chen, C., Benjamin, M. S., Sun, X., Otto, K. B., Guo, P., Dong, X. Y., Bao, Y., Zhou, Z.,Cheng, X., Simons, J. W. and Dong, J. T. (2006) KLF5 promotes cell proliferation andtumorigenesis through gene regulation in the TSU-Pr1 human bladder cancer cell line.Int. J. Cancer 118, 1346–1355

7 Bateman, N. W., Tan, D., Pestell, R. G., Black, J. D. and Black, A. R. (2004) Intestinaltumor progression is associated with altered function of KLF5. J. Biol. Chem. 279,12093–12101

8 Guo, P., Zhao, K. W., Dong, X. Y., Sun, X. and Dong, J. T. (2009) Acetylation of KLF5 altersthe assembly of P15 transcription factors in TGFβ-mediated induction in epithelial cells.J. Biol. Chem. 284, 18184–18193

9 Guo, P., Dong, X. Y., Zhao, K. W., Sun, X., Li, Q. and Dong, J. T. (2009) Opposing effectsof KLF5 on the transcription of Myc in epithelial proliferation in the context of TGFβ .J. Biol. Chem. 284, 28243–28252

10 Sun, R., Chen, X. and Yang, V. W. (2001) Intestinal-enriched kruppel-like factor(kruppel-like factor 5) is a positive regulator of cellular proliferation. J. Biol. Chem. 276,6897–6900

11 Adam, P. J., Regan, C. P., Hautmann, M. B. and Owens, G. K. (2000) Positive- andnegative-acting Kruppel-like transcription factors bind a transforming growth factor β

control element required for expression of the smooth muscle cell differentiation markerSM22α in vivo. J. Biol. Chem. 275, 37798–37806

12 Yang, Y., Goldstein, B. G., Chao, H. H. and Katz, J. P. (2005) KLF4 and KLF5 regulateproliferation, apoptosis and invasion in esophageal cancer cells. Cancer Biol. Ther. 4,1216–1221

13 Dong, J. T. and Chen, C. (2009) Essential role of KLF5 transcription factor in cellproliferation and differentiation and its implications for human diseases. Cell. Mol. LifeSci. 66, 2691–2706

14 Guo, P., Dong, X. Y., Zhang, X., Zhao, K. W., Sun, X., Li, Q. and Dong, J. T. (2009)Pro-proliferative factor KLF5 becomes anti-proliferative in epithelial homeostasis uponsignaling-mediated modification. J. Biol. Chem. 284, 6071–6078

15 Sur, I., Rozell, B., Jaks, V., Bergstrom, A. and Toftgard, R. (2006) Epidermal andcraniofacial defects in mice overexpressing Klf5 in the basal layer of the epidermis. J. CellSci. 119, 3593–3601

16 Nandan, M. O., McConnell, B. B., Ghaleb, A. M., Bialkowska, A. B., Sheng, H., Shao, J.,Babbin, B. A., Robine, S. and Yang, V. W. (2008) Kruppel-like factor 5 mediates cellulartransformation during oncogenic KRAS-induced intestinal tumorigenesis.Gastroenterology 134, 120–130

17 McConnell, B. B., Klapproth, J. M., Sasaki, M., Nandan, M. O. and Yang, V. W. (2008)Kruppel-like factor 5 mediates transmissible murine colonic hyperplasia caused byCitrobacter rodentium infection. Gastroenterology 134, 1007–1016

18 Goldstein, B. G., Chao, H. H., Yang, Y., Yermolina, Y. A., Tobias, J. W. and Katz, J. P.(2007) Overexpression of Kruppel-like factor 5 in esophageal epithelia in vivo leads toincreased proliferation in basal but not suprabasal cells. Am. J. Physiol. Gastrointest.Liver Physiol. 292, G1784–G1792

19 Chen, C., Bhalala, H. V., Qiao, H. and Dong, J. T. (2002) A possible tumor suppressor roleof the KLF5 transcription factor in human breast cancer. Oncogene 21, 6567–6572

20 Chen, C., Bhalala, H. V., Vessella, R. L. and Dong, J. T. (2003) KLF5 is frequently deletedand down-regulated but rarely mutated in prostate cancer. Prostate 55, 81–88

21 Chanchevalap, S., Nandan, M. O., Merlin, D. and Yang, V. W. (2004) All-trans retinoicacid inhibits proliferation of intestinal epithelial cells by inhibiting expression of the geneencoding Kruppel-like factor 5. FEBS Lett. 578, 99–105

22 Chen, C., Sun, X., Guo, P., Dong, X. Y., Sethi, P., Cheng, X., Zhou, J., Ling, J., Simons,J. W., Lingrel, J. B. and Dong, J. T. (2005) Human Kruppel-like factor 5 is a target of theE3 ubiquitin ligase WWP1 for proteolysis in epithelial cells. J. Biol. Chem. 280,41553–41561

23 Chen, C., Sun, X., Guo, P., Dong, X. Y., Sethi, P., Zhou, W., Zhou, Z., Petros, J., Frierson,H. F., Vessella, R. L. et al. (2007) Ubiquitin E3 ligase WWP1 as an oncogenic factor inhuman prostate cancer. Oncogene 26, 2386–2394

24 Tong, D., Czerwenka, K., Heinze, G., Ryffel, M., Schuster, E., Witt, A., Leodolter, S. andZeillinger, R. (2006) Expression of KLF5 is a prognostic factor for disease-free survivaland overall survival in patients with breast cancer. Clin. Cancer Res. 12, 2442–2448

25 Guo, P., Dong, X. Y., Zhao, K. W., Sun, X., Li, Q. and Dong, J. T. (2010) Estrogen-inducedinteraction between KLF5 and estrogen receptor (ER) suppresses the function of ER inER-positive breast cancer cells. Int. J. Cancer. 126, 81–89

26 Ikeda, K., Orimo, A., Higashi, Y., Muramatsu, M. and Inoue, S. (2000) Efp as a primaryestrogen-responsive gene in human breast cancer. FEBS Lett. 472, 9–13

27 Zou, W. and Zhang, D. E. (2006) The interferon-inducible ubiquitin-protein isopeptideligase (E3) EFP also functions as an ISG15 E3 ligase. J. Biol. Chem. 281, 3989–3994

28 Suzuki, T., Urano, T., Tsukui, T., Horie-Inoue, K., Moriya, T., Ishida, T., Muramatsu, M.,Ouchi, Y., Sasano, H. and Inoue, S. (2005) Estrogen-responsive finger protein as a newpotential biomarker for breast cancer. Clin. Cancer Res. 11, 6148–6154

29 Urano, T., Saito, T., Tsukui, T., Fujita, M., Hosoi, T., Muramatsu, M., Ouchi, Y. and Inoue,S. (2002) Efp targets 14-3-3σ for proteolysis and promotes breast tumour growth. Nature417, 871–875

30 Takeuchi, T., Inoue, S. and Yokosawa, H. (2006) Identification and Herc5-mediatedISGylation of novel target proteins. Biochem. Biophys. Res. Commun. 348, 473–477

31 Zou, W., Wang, J. and Zhang, D. E. (2007) Negative regulation of ISG15 E3 ligase EFPthrough its autoISGylation. Biochem. Biophys. Res. Commun. 354, 321–327

32 Chen, C., Zhou, Z., Guo, P. and Dong, J. T. (2007) Proteasomal degradation of the KLF5transcription factor through a ubiquitin-independent pathway. FEBS Lett. 581, 1124–1130

33 Chen, C., Sun, X., Ran, Q., Wilkinson, K. D., Murphy, T. J., Simons, J. W. and Dong, J. T.(2005) Ubiquitin-proteasome degradation of KLF5 transcription factor in cancer anduntransformed epithelial cells. Oncogene 24, 3319–3327

34 Pfaffl, M. W. (2001) A new mathematical model for relative quantification in real-timeRT-PCR. Nucleic Acids Res. 29, e45

35 Zheng, H. Q., Zhou, Z., Huang, J., Chaudhury, L., Dong, J. T. and Chen, C. (2009)Kruppel-like factor 5 promotes breast cell proliferation partially through upregulating thetranscription of fibroblast growth factor binding protein 1. Oncogene 28, 3702–3713

36 Chen, C., Zhou, Y., Zhou, Z., Sun, X., Otto, K. B., Uht, R. M. and Dong, J. T. (2004)Regulation of KLF5 involves the Sp1 transcription factor in human epithelial cells. Gene330, 133–142

37 Zhao, D., Zheng, H.Q., Zhou, Z. and Chen, C. (2010) The Fbw7 tumor suppressor targetsKLF5 for ubiquitin-mediated degradation and suppresses breast cell proliferation. CancerRes. 70, 4728–4738

38 Liu, N., Li, H., Li, S., Shen, M., Xiao, N., Chen, Y., Wang, Y., Wang, W., Wang, R., Wang,Q. et al. (2010) The Fbw7/human CDC4 tumor suppressor targets proproliferative factorKLF5 for ubiquitination and degradation through multiple phosphodegron motifs. J. Biol.Chem. 285, 18858–18867

39 Inoue, S., Orimo, A., Hosoi, T., Kondo, S., Toyoshima, H., Kondo, T., Ikegami, A., Ouchi,Y., Orimo, H. and Muramatsu, M. (1993) Genomic binding-site cloning reveals anestrogen-responsive gene that encodes a RING finger protein. Proc. Natl. Acad. Sci.U.S.A. 90, 11117–11121

40 Coux, O., Tanaka, K. and Goldberg, A. L. (1996) Structure and functions of the 20S and26S proteasomes. Annu. Rev. Biochem. 65, 801–847

41 Rechsteiner, M. and Hill, C. P. (2005) Mobilizing the proteolytic machine: cell biologicalroles of proteasome activators and inhibitors. Trends Cell Biol. 15, 27–33

42 Jariel-Encontre, I., Bossis, G. and Piechaczyk, M. (2008) Ubiquitin-independentdegradation of proteins by the proteasome. Biochim. Biophys. Acta 1786, 153–177

43 Nakajima, A., Maruyama, S., Bohgaki, M., Miyajima, N., Tsukiyama, T., Sakuragi, N. andHatakeyama, S. (2007) Ligand-dependent transcription of estrogen receptor α ismediated by the ubiquitin ligase EFP. Biochem. Biophys. Res. Commun. 357, 245–251

44 Orimo, A., Inoue, S., Ikeda, K., Noji, S. and Muramatsu, M. (1995) Molecular cloning,structure, and expression of mouse estrogen-responsive finger protein Efp.Co-localization with estrogen receptor mRNA in target organs. J. Biol. Chem. 270,24406–24413

45 Inoue, S., Urano, T., Ogawa, S., Saito, T., Orimo, A., Hosoi, T., Ouchi, Y. and Muramatsu,M. (1999) Molecular cloning of rat efp: expression and regulation in primary osteoblasts.Biochem. Biophys. Res. Commun. 261, 412–418

46 Das, N., Wang, J. and Dey, S. K. (1997) Uterine preparation for implantation in the mouseis associated with coordinate expression of estrogen-responsive finger protein andestrogen receptor. Mol. Reprod. Dev. 46, 499–506

47 Gack, M. U., Shin, Y. C., Joo, C. H., Urano, T., Liang, C., Sun, L., Takeuchi, O., Akira, S.,Chen, Z., Inoue, S. and Jung, J. U. (2007) TRIM25 RING-finger E3 ubiquitin ligase isessential for RIG-I-mediated antiviral activity. Nature 446, 916–920

48 Knuutila, S., Aalto, Y., Autio, K., Bjorkqvist, A. M., El-Rifai, W., Hemmer, S., Huhta, T.,Kettunen, E., Kiuru-Kuhlefelt, S., Larramendy, M. L. et al. (1999) DNA copy numberlosses in human neoplasms. Am. J. Pathol. 155, 683–694

49 Nakajima, Y., Akaogi, K., Suzuki, T., Osakabe, A., Yamaguchi, C., Sunahara, N., Ishida, J.,Kako, K., Ogawa, S., Fujimura, T. et al. (2011) Estrogen regulates tumor growth through anonclassical pathway that includes the transcription factors ERβ and KLF5. Sci. Signaling4, ra22

Received 27 August 2010/21 April 2011; accepted 4 May 2011Published as BJ Immediate Publication 4 May 2011, doi:10.1042/BJ20101388

c© The Authors Journal compilation c© 2011 Biochemical Society

Biochem. J. (2011) 437, 323–333 (Printed in Great Britain) doi:10.1042/BJ20101388

SUPPLEMENTARY ONLINE DATAOestrogen causes degradation of KLF5 by inducing the E3 ubiquitin ligaseEFP in ER-positive breast cancer cellsKe-Wen ZHAO*†1, Deepa SIKRIWAL*1, Xueyuan DONG*, Peng GUO*, Xiaodong SUN* and Jin-Tang DONG*2

*Department of Hematology and Medical Oncology, School of Medicine and Winship Cancer Institute, Emory University, 1365 Clifton Road, Atlanta, GA 30322, U.S.A., and†Key Laboratory of Cell Differentiation and Apoptosis of Chinese Ministry of Education, Department of Pathophysiology, Shanghai Jiaotong University School of Medicine, Shanghai,China 200025

EXPERIMENTAL

Cell culture and treatment

The ZR-75-1 ER-positive breast cancer cell line was purchasedfrom the A.T.C.C.) and maintained in RPMI 1640 mediumsupplemented with 10% FBS, 0.01 M Hepes, 1 mM sodiumpyruvate, 0.15% sodium bicarbonate, 0.45% glucose and 1%penicillin/streptomycin solution. To investigate the direct role ofoestrogen responsiveness of EFP and KLF5 degradation, ZR-75-1cells were maintained in Phenol Red-free medium supplementedwith 10% charcoal-dextran-stripped FBS for at least 3 days.Subsequently, the cells were treated with 1 μM 17β-oestradiol(E2) and were harvested at the times indicated to isolate totalRNA and protein for analysis.

To investigate the direct role for oestrogen responsiveness ofEFP and KLF5 degradation, MCF-7 cells cultured in regularmedium were treated with 1 μM tamoxifen (Tam) for the timesindicated, and subsequently total RNA and protein were isolatedfor analysis.

Western blot analysis

Cell lysates for Western blotting were prepared as described in theExperimental section of the main text. Briefly, following treatmentthe cells were lysed with RIPA buffer [50 mM Tris/HCl (pH 8.0),150 mM NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate,

0.1% SDS and 1% proteinase inhibitor I cocktail (Sigma)]. Thelysates were then processed for Western blotting using standardprotocols and later probed with appropriate antibodies.

siRNA transfection

siRNAs were obtained from Santa Cruz Biotechnology(catalogue numbers sc-37825 and sc-37825b) and wereused to target EFP. Another siRNA against EFP waschemically synthesized (Dharmacon) with the sequence 5′-GGUGGAGCAGCUACAACAA-3′, and was also used to targetEFP. MCF-7 cells were transfected with 100 nM of each siRNA,and transfection was carried out using the siPORT Amine Reagent(Ambion) in six-well plates. At 48 h after transfection, totalprotein was collected for analysis using Western blot analysis.

CHX chase assay of MCF-7 cells

MCF-7 cells were maintained in Phenol Red-free mediumsupplemented with 10% charcoal-dextran-stripped FBS for3 days. Subsequently, the cells were treated with 1 μM 17β-oestradiol (E2) for 4 h. Following oestrogen treatment, cells weretreated with 50 μg/ml CHX for different time periods. Totalproteins were collected and analysed by Western blotting forKLF5, EFP and β-actin.

1 These two authors contributed equally to this work2 To whom correspondence should be addressed (email j.dong@emory.edu).

c© The Authors Journal compilation c© 2011 Biochemical Society

K.-W. Zhao and others

Figure S1 Oestrogen treatment causes the degradation of KLF5 in ER-positive breast cancer cells

(A) ZR-75-1 cells grown in hormone-free medium and treated with 1 μM oestrogen (E2) for the indicated times were subjected to Western blotting to detect protein expression of KLF5, EFP andβ-actin. (B) ZR-75-1 cells in (A) were subjected to real-time PCR for RNA expression of EFP and KLF5. (C) MCF-7 cells grown in regular medium were treated with the ER signalling inhibitortamoxifen (Tam) for the indicated times, and protein expression of KLF5, EFP and β-actin was evaluated by Western blotting. (D) MCF-7 cells grown in regular medium and treated with 1 μMtamoxifen for the indicated times were analysed by real-time PCR for RNA expression of EFP and KLF5. (E) RNA samples of MCF-7 cells from (C) were subjected to real-time PCR to detect RNAexpression of WWP1 and FBW7.

c© The Authors Journal compilation c© 2011 Biochemical Society

EFP mediates oestrogen-caused KLF5 degradation

Figure S2 Knockdown of EFP increases the protein level of KLF5

Transfection of a mixture of EFP siRNA (siEFP) purchased from Santa Cruz Biotechnology (A)or a chemically synthesized siRNA (C), or a single siEFP from Santa Cruz Biotechnology (E)into MCF-7 cells cultured in regular medium increases the protein levels of KLF5. MCF-7 cellscultured in hormone-free medium were transfected with either the Santa Cruz BiotechnologysiRNA (B), synthesized siRNA (D) or control siRNA for 24 h, treated with 1 μM oestrogen (E2)for 18 h, and subjected to Western blotting for protein expression of KLF5, EFP and β-actin.

c© The Authors Journal compilation c© 2011 Biochemical Society

K.-W. Zhao and others

Figure S3 EFP down-regulates KLF5 by protein degradation

COS-1 cells were transfected with expression plasmids for EFP, KLF5 and GFP (A), or EFP and GFP–KLF5 fusion protein (B) for 48 h, and cell lysates were analysed by Western blotting for proteinexpression of KLF5, EFP, GFP and β-actin. The presence or absence of a construct is indicated by ‘ + ’ or ‘ − ’. The molecular mass in kDa is indicated on the right-hand side.

Figure S4 CHX chase assay of MCF-7 cells in the presence and absence ofoestrogen (E2)

(A) MCF-7 cells cultured in hormone-free medium for 3 days were incubated with 1 μMoestrogen (E2) or solvent control for 4 h, followed by CHX treatment for the indicated times.Western blotting was performed to detect the protein expression of KLF5, EFP and β-actin. (B)Band intensities for all proteins were determined using the ImageJ program, and the intensityfor KLF5 or EFP in each lane was divided by that for β-actin to indicate the relative level ofKLF5 or EFP. The ratio for either KLF5 or EFP without CHX treatment, with or without oestrogentreatment, was defined as 1 and the ratios for other lanes were normalized accordingly andplotted against CHX treatments.

c© The Authors Journal compilation c© 2011 Biochemical Society

EFP mediates oestrogen-caused KLF5 degradation

Table S1 Primer sequences (5′→3′) for creating EFP-deletion mutants

The underlined sequences represent the restriction sites

Mutant Forward primer Reverse primer

EFP-M1 GGGGAATTCTGATGGCAGAGCTGTGCCCCCTG CCCTCTAGATGCTGGGTGCAGAGGCGCGGEFP-M2 GGGGGTACCCCGAATGCCCAGGTGGCCTGC CCCGAATTCCTACTTGGGGGAGCAGATGGAGAGTG

Table S2 Primer sequences (5′→3′) for creating EFP-substitution mutants

The underlined sequences represent the amino acid mutated

Mutant Primer-1 Primer-2

EFP-cS CCATACCTGAGCCCGCAGAGCCGCGCCGTC GACGGCGCGGCTCTGCGGGCTCAGGTATGGEFP-kR GAGGCCGCCGTGAGGACGTGCTTGGTG CACCAAGCACGTCGTCACGGCGGCCTC

Table S3 Primer sequences (5′→3′) for real-time PCR analysis

Name Forward primer Reverse primer

KLF5 CAGAGGACCTGGTCCA GACAAGATG GAGGCCAGTTCTCAGGTGAGTGATGEFP GTGACCACGGCTTTGTCATCT TAAAGTCCACCC TGAACTTATACATCAGβ-Actin GAAACTACCTTCAACTCCATC CTAGAAGCATTTGCGGTGGACGATGGAGGGGCCFGF-BP GCAGATGGGCTGCTACTGAG CGGGCAACTTGTTTCCAATAWWP1 GTATGGATCCTGTACGGCAGCA GTTGTGGTCTCTCCCATGTGGTFbw-7 CCCGGAGCTGTGCAGCAA CAGATGTAATTCGGCGTCGTT

Received 27 August 2010/21 April 2011; accepted 4 May 2011Published as BJ Immediate Publication 4 May 2011, doi:10.1042/BJ20101388

c© The Authors Journal compilation c© 2011 Biochemical Society