Tài liệu Báo cáo khoa học: FOXM1c transactivates the human c-mycpromoter directly via the two TATA boxes P1 and P2 docx

However, PIC assembly willalways require at least two separate steps, namelyTFIID⁄ TFIIA binding and TFIIB ⁄ Pol II binding [46].Here, we describe a new transactivation mechanism by whic

directly via the two TATA boxes P1 and P2

Inken Wierstra1and Ju¨rgen Alves2

1 Institute of Molecular Biology, Medical School Hannover, Germany

2 Institute of Biophysical Chemistry, Medical School Hannover, Germany

c-Myc, a key regulator of proliferation, differentiation

and apoptosis, plays a central role in cell growth

control and can induce quiescent cells to enter into

S-phase [1–7] Because c-Myc potently stimulates

pro-liferation and inhibits differentiation it possesses a high

transformation potential that is supplemented by its

cell growth and angiogenesis-promoting,

cell-adhesion-reducing, immortality and genomic-instability-causing

activities c-myc expression correlates strictly with cell

proliferation c-Myc regulates target genes either by

activation via E-boxes or by repression via initiator

(Inr)-dependent and Inr-independent mechanisms

c-Myc acts as part of the Myc⁄ Max ⁄ Mad network in

which Max is the heterodimerization partner forc-Myc and Mad proteins, the c-Myc antagonists,which repress target genes via E-boxes

The forkhead⁄ winged helix transcription factorFOXM1, expression of which correlates strictly withproliferation, stimulates proliferation by promotingS- and M-phase entry and regulates genes that control

G1⁄ S and G2⁄ M transition [8–27] The activity ofFOXM1 as a conventional transcription factor isincreased by proliferation signals and reduced by anti-proliferative signals Furthermore, FOXM1 is assumed

Abbreviations

BRE, TFIIB recognition element; ChIP, chromatin immunoprecipitation; DBD, DNA-binding domain; DPE, downstream promoter element; EDA, essential domain for activation; EMSA, electrophoretic mobility shift assay; FKH, forkhead domain; GST, glutathione S-transferase; GTF, general transcription factor; Inr, initiator; NE, neutrophile elastase; NLS, nuclear localization signal; NRD, negative regulatory domain; OHT, 4-hydroxy-tamoxifen; PIC, preinitiation complex; RB, retinoblastoma protein; SV40, simian virus 40; TAD, transactivation domain; TAF, TBP-associated factor; TBP, TATA-binding protein; TFIIB, transcription factor IIB; TK, thymidine kinase; TPA, 12-O-tetradecanoylphorbol-13- acetate; TRD, transrepression domain.

transcription factor the splice variant FOXM1c (MPP2)

binds to FOXM1-specific DNA sequences via its

fork-head domain and transactivates via its strong acidic

transactivation domain (TAD) [29–31] This strong

TAD can be kept almost inactive by two different

inhibitory domains The N-terminus functions as a

specific negative regulatory domain (NRD), named

NRD-N, which completely inhibits the TAD by directly

binding to it The central domain functions as a

retino-blastoma protein (RB)-independent transrepression

domain (TRD) [29–31] and as RB-recruiting NRD-C

[31]

Core promoters and basal transcription complexes

were initially thought to be interchangeable at will, but

are now viewed as active participants in gene

regula-tion Their diversity makes essential contributions to

the specificity and variability in combinatorial gene

regulation [32–34] Core promoter elements are the

TATA box, the initiator (Inr), the downstream

promo-ter element (DPE), motif ten element (MTE) and the

transcription factor IIB (TFIIB) recognition element

(BRE) None of these elements is obligatory and

sev-eral different combinations are operational Enhancers

can target certain core promoter elements so that their

activating effect is limited to genes with these elements

[32–35] Basal transcription complexes are not uniform

because of TATA-binding protein (TBP)-related

fac-tors and alternative TBP-associated facfac-tors (TAFIIs)

[36,37] It is believed that the basal transcription

com-plex can adopt different conformations on different

core promoters and that different core promoters can

determine different rate-limiting steps in preinitiation

complex (PIC) assembly and transcription initiation, as

well as different reinitiation rates [32–34,38–48]

TBP plays a central role in the recognition of TATA

box promoters The C-terminal⁄ core region of TBP

has a saddle-like structure: its concave underside binds

to DNA; the convex upper surface binds to a large

variety of TAFIIs, general trancription factors (GTFs),

transcription factors, coactivators and general

cofac-tors [38,49,50] TBP binds to the minor groove of the

TATA box, thereby bending the DNA 80 towards the

major groove, unwinding the DNA by 120 and

kink-ing the TATA box at both ends by intercalation of

two phenylalanine residues TFIIA interacts with the

N-terminal TBP stirrup, which is orientated towards

the 3¢-end of the TATA box, and with TBP helices H1

and H2 TFIIB interacts with the C-terminal TBP

stir-rup, which is orientated towards the 5¢-end of the

TATA box, and with TBP helix H1¢ [38,39,51]

The PIC can be assembled in a stepwise fashion in

reconstituted in vitro systems [38,39] In vivo, PIC

assembly may vary among core promoters between

two extremes: (a) the stepwise assembly of individualGTFs, and (b) recruitment of the complete holo-enzyme in one step [45] However, PIC assembly willalways require at least two separate steps, namelyTFIID⁄ TFIIA binding and TFIIB ⁄ Pol II binding [46].Here, we describe a new transactivation mechanism

by which FOXM1c transactivates the c-myc promotervia its P1 and P2 TATA boxes It does so by binding

to the TATA box and directly to TBP, TFIIB andTFIIA The P1 TATA box TATAATGC requires itssequence context to be FOXM1c responsive In con-trast, the P2 TATA box TATAAAAG alone issufficient to confer FOXM1c responsiveness on anyminimal promoter so that each promoter with thisTATA box is postulated to be transactivated byFOXM1c as seen for c-fos, hsp70 and histone H2B⁄ a

In addition to these new FOXM1c target genes, adatabase search revealed nearly 300 genes with such aTATA box sequence, many of which also play a role

in proliferation and tumorigenesis Accordingly, inant-negative FOXM1c proteins reduce cell growth byapproximately threefold demonstrating a proliferation-stimulating function for wild-type FOXM1c

dom-Results

FOXM1c transactivates the c-myc promoter,namely the minimal P1 and P2 promotersHuman c-myc promoter was transactivated by wild-type FOXM1c and significantly more so by the mutantFOXM1c(189–762) (Fig 1A), which lacks the negat-ive-regulatory N-terminus (see below) Therefore,FOXM1c(189–762) was used in this study In contrast

to c-myc, FOXM1c(189–762) did not transactivate thepromoters of human c-jun, waf1(p21), ink4a(p16),murine neutrophile elastase (NE) or the simian virus(SV)40 early promoter (Fig 1B; data not shown)

To map the FOXM1c-responsive element, severalc-myc–promoter constructs were analyzed (Fig 1D).FOXM1c(189–762) strongly transactivated the P1and P2 promoters, but not the P0 promoter Becauseall potential FOXM1c-binding sites (C⁄ T-AAA-C ⁄ T)

of the c-myc promoter are positioned in the FOXM1c-responsive segment )2486 ⁄ )259 (Fig 1D;data not shown), common elements of the P1 andP2 promoters were analyzed for FOXM1c responsive-ness The P1 and P2 promoters both possess aTATA box and a GC-box-type Sp1-binding site.The Sp1-binding sites )44 (known; position )44relative to the P1 transcription start site) and )66(potential; position )66 relative to the P2 transcriptionstart site), as well as overlapping binding sites for

non-other transcription factors were not FOXM1c

respon-sive (Fig 1D) Minimal promoters include only the

TATA box and the transcription start (+1) These

minimal c-myc P1 and P2 promoters were bothstrongly transactivated by FOXM1c(189–762) (Fig 1C,D) By contrast, the minimal promoters of human

-66

GCTTGGCGGGAAA

GCGGGAAA E2F gGGAA ETS-Core TTGGCGGGAAA STAT3 GGAAA NFATc1-Consensus GGCTT Smad

GGAAAG METS-Consensus cGT

3x

-95

+49 P2

pmycluc

pmyc(-262/+49)luc

TA by FOXM1c (189-762)

+

-+ + +

pmyc(-2486/-259) mintkluc

C

pTATA-WAF-luc pTATA-jun-luc pTATA-P2-luc pTATA-P1-luc pmintkluc

pwaf1 (p21)luc

pmyc luc pjun

luc

C FOXM1c(189-762)

3x

ATCTCCGCCCACC

Fig 1 FOXM1c transactivates the minimal P1 and P2 promoters of c-myc (A, B) RK13 cells were transiently transfected with expression plasmids for the FOXM1c proteins or as control (c) with the empty vector and with the indicated reporter constructs The relative luciferase activity of each reporter construct in the control (c) was set as 1 (C) RK13 cells were transiently transfected with the indicated amounts of pFOXM1c(189–762) and with the indicated reporter constructs The relative luciferase activity of each reporter construct in the absence of pFOXM1c(189–762) was set as 1 (D) c-myc sequences are shown as black lines, TATA boxes as black boxes, transcription start sites (+1)

as arrows, Sp1-binding sites are shown as dark gray boxes and sequences of the thymidine kinase (TK) promoter of herpes simplex virus (HSV) as a light gray box Numbers give the nucleotides of c-myc relative to the transcription start (+1) of P2 p( )44)mintkluc and p( )66)mintkluc contain three adjacent copies of the indicated nucleotide sequences Sp1-binding sites are marked bold and underlined Bind- ing sites for other transcription factors are indicated below It is indicated whether the reporter constructs are transactivated by FOXM1c(189–762) (¼ +) or not (¼ –) TA, transactivation; P0, P1, P2, c-myc promoters; mintk, minimal TK promoter of HSV.

c-jun, waf1(p21) or herpes simplex virus (HSV)

thymi-dine kinase (TK) were not transactivated by

FOXM1c(189–762) (Fig 1C)

The P1 and P2 TATA boxes are the

FOXM1c-responsive elements

The existence of FOXM1c-responsive and

-nonrespon-sive minimal promoters offered the possibility of

con-structing hybrid minimal promoters (Fig 2C) to map

the responsive element exactly Hybrids exchanging the

TATA box half and the transcription start (+1) half

between c-myc P1 or c-myc P2 and c-jun promoters

showed that the TATA box halves of the P1 and P2

promoters both transfer FOXM1c responsiveness

(Fig 2A) Hybrids exchanging only the TATA boxes

between P1 or P2 and the c-jun or waf1⁄ (p21)

promot-ers, and vice versa, showed that the c-myc P1 and P2

TATA boxes are themselves the FOXM1c-responsive

elements (Fig 2B,C) Both are necessary for FOXM1c

responsiveness because replacing them with the TATA

box of a non-FOXM1c-responsive promoter abolished

transactivation by FOXM1c(189–762) (Fig 2B,C) The

P2–TATA box is sufficient as the FOXM1c-responsive

element because insertion of it into a nonresponsive

minimal promoter resulted in very strong

transactiva-tion by FOXM1c(189–762) (Fig 2B) The P1 TATA

box requires its sequence context to function as the

FOXM1c-responsive element because insertion of it

into the minimal promoters of c-jun and waf1(p21)

did not result in transactivation by FOXM1c(189–762)

(Fig 2C) Figure 2D shows the sequence differences

between the TATA boxes used To our knowledge,

transactivation of a promoter by a transcription factor

via its TATA box has not been described previously

and thus represents a new mechanism

FOXM1c domains required for transactivation

of the c-myc promoter

FOXM1c transactivates by two different mechanisms:

(a) the reporter construct p(MBS)3-mintk-luc via its

FOXM1c-binding sites as a conventional transcription

factor [29–31]; and (b) the P1 and P2 promoters ofc-myc via their TATA boxes by a new mechanism.Several FOXM1c mutants (Fig 3F) the expressionlevels of which have been compared previously [30]were analyzed for transactivation of c-myc promoterconstructs (Fig 1D) Two mutants lacking either part

of the TAD (amino acids 721–762) or part of theforkhead domain (amino acids 235–332), and therebythe complete recognition helix 3 (amino acids 277–290) [53], repressed or did not transactivate the P1and P2 promoters (Fig 3A,B) Therefore, both theintact DNA-binding domain (DBD) and the intactTAD are essential for transactivation of the P1 andP2 promoters (Fig 3E,F) Wild-type FOXM1c trans-activated the P1 and P2 promoters considerably lessthan FOXM1c(189–762) (Fig 3A) The N-terminus(amino acids 1–232) in trans repressed transactivation

of the P1 and P2 promoters by FOXM1c(189–762)(Fig 3D), which can be explained by the direct interac-tion of the N-terminus (amino acids 1–194) with theTAD (amino acids 721–762) [30] Therefore, the N-ter-minus as NRD represses transactivation of the P1 andP2 promoters by directly binding to the TAD In sum-mary, the forkhead domain (i.e the DBD) TAD andN-terminus, have the same functions for transactiva-tion of the c-myc promoter via its TATA boxes and fortransactivation as a conventional transcription factor(Fig 3E,F) [30]

FOXM1c(189–348; 573–762)NLS did not vate the P1 and P2 promoters (Fig 3C) In contrast,FOXM1c(189–425; 568–762) transactivated the P1and P2 promoters as strongly as FOXM1c(189–762)

transacti-if the lower expression level of the former [30] wastaken into account (Fig 3A) Thus, these twomutants with deletions in the central domain (aminoacids 349–572) showed that amino acids 349–425 areessential for transactivation of the P1 and P2 promot-ers Therefore, amino acids 349–425 are referred to asthe essential domain for activation (EDA) The cen-tral domain has opposing functions for transactiva-tion of the c-myc promoter via its TATA boxes,where it functions as the EDA, and for transactiva-tion as a conventional transcription factor, where it

Fig 2 The FOXM1c-responsive elements are the P1 and P2 TATA boxes (A ,B) RK13 cells were transiently transfected with the indicated amounts of pFOXM1c(189–762) and with the indicated reporter constructs The relative luciferase activity of each reporter construct in the absence of pFOXM1c(189–762) was set as 1 (C) TATA boxes and transcription start sites (+1) are bold and underlined Symbols below the nucleotide sequences explain the composition of hybrid promoters It is indicated whether the reporter constructs are transactivated by FOXM1c(189–762) (¼ +) or not (¼ –) TA, transactivation (D) Differences of TATA boxes of non-FOXM1c-responsive (¼ –) promoters to the FOXM1c-responsive (¼ +) TATA boxes c-myc-P1 and c-myc-P2 Nucleotides that deviate from the c-myc TATA box are bold Nucleotides that are identical to the c-myc TATA box are replaced by a dash For c-jun and TK both possible TATA box positions are shown c-myc-P0 and ink4a(p16) are TATA-less (¼ –) non-FOXM1c responsive promoters TA by FOXM1c, transactivation by FOXM1c(189–762); NE, murine neutrophile elastase; TK, thymidine kinase of HSV; SV40early, early promoter of simian virus (SV)40.

functions as an inhibitory domain [29–31] (Fig 3E,F).

Consequently, FOXM1c(189–348; 573–762)NLS can

be used to discriminate between these mechanisms:

(a) if it transactivates considerably more strongly

than FOXM1c(189–762), FOXM1c functions as a

conventional transcription factor; and (b) if it does

not transactivate, FOXM1c functions via the TATA

box

FOXM1c transactivates other genes involved incell proliferation that possess the c-myc P2 TATAbox TATAAAAG

The c-myc P2 TATA box is sufficient to transfer verystrong transactivation by FOXM1c(189–762) to a non-responsive minimal promoter (Fig 2) Consequently, itwas postulated that each promoter with this TATA

Fig 4 FOXM1c transactivates other proliferation-associated genes with the c-myc P2 TATA box TATAAAAG (A, B) RK13 cells were transiently transfected with expression plasmids for the FOXM1c proteins or as control (c) with the empty vector and with the indicated reporter con- structs The relative luciferase activity of each reporter construct in the control (c) was set as 1 phsp70luc contains the hsp70 promoter sequence from )2400 to +150 phsp70-TATA-luc contains the hsp70 promoter sequence from )32 to +150, i.e a ‘minimal’ hsp70 promoter (C) Summary of the flanking nucleotides of the TATA box TATAAAAG (bold and underlined) in the six promoters that are activated (¼ +) by FOXM1c The transcription start site (+1) is bold and underlined Symbols below the sequences explain the composition of hybrid promoters.

Fig 3 FOXM1c domains required for c-myc promoter transactivation (A–C) RK13 cells were transiently transfected with expression mids for the indicated FOXM1c proteins or as control (c) with the empty vector and with the indicated reporter constructs The relative lucif- erase activity of each reporter construct in the control (c) was set as 1 (D) RK13 cells were transiently transfected with the expression plasmid for FOXM1c(189–762) or as control (c) with the empty vector and with the indicated reporter constructs The indicated amounts of pFOXM1c(1–232) were cotransfected (E) Functions of FOXM1c domains for transactivation of the c-myc promoter via the P1 and P2 TATA boxes and for transactivation of p(MBS)3-mintk-luc as a conventional transcription factor [29–31] and whether their functions in these two dif- ferent transactivation mechanisms are equivalent or opposite TA, transactivation; IA, interaction; P1, P2, P1- or P2-promoter of c-myc (E, F) TAD, transactivation domain; DBD, DNA-binding domain; TRD, transrepression domain; EDA, essential domain for activation; NRD, negative regulatory domain (F) FOXM1c(189–348; 573–762)NLS possesses the nuclear localization signal (NLS) of SV40 large T between amino acids

plas-348 and 573 FKH, forkhead domain p(MBS)3-mintk-luc is transactivated very strongly (+ + + + +), strongly (+ + +) or weakly (+) or repressed (–) and the c-myc-promoter is transactivated very strongly (+ + + + +), strongly (+ + +) or repressed (–) or neither transactivated nor repressed () Note that the indicated transactivation for FOXM1(189–425; 568–762) is corrected by expression (see text).

box is transactivated by FOXM1c Therefore, the

pro-moters of human c-fos, hsp70 and histone H2B⁄ a which

all possess the c-myc P2 TATA box TATAAAAG

(Fig 4C) were tested As postulated, these three

pro-moters were transactivated by FOXM1c(189–762), but

not transactivated or considerably less so (Fig 4A,B)

by FOXM1c(189–348; 573–762)NLS This also held

true for a ‘minimal’ hsp70 promoter (Fig 4B) showing

that FOXM1c transactivates the hsp70 promoter via its

TATA box The parental vectors used to construct

the reporter plasmids were not FOXM1c responsive

(Figs 1B,D and 4A,B; data not shown) This

transacti-vation of the c-fos, hsp70 and histone H2B⁄ a promoters

confirmed that each promoter with the c-myc P2 TATA

box is transactivated by FOXM1c Comparison of the

six promoters used showed that, in the sequences

flank-ing the c-myc P2 TATA box, almost every nucleotide

was found at almost every position (Fig 4C) Thus the

c-myc P2 TATA box TATAAAAG alone is sufficient

as the FOXM1c-responsive element A database search

for promoters with this TATA box gave a list of almost

300 potential FOXM1c target genes (Fig S1)

FOXM1c binds directly to components of the

basal transcription complex

To characterize this new mechanism by which

FOXM1c transactivates the c-myc P1 and P2 promoters

we analyzed whether FOXM1c binds to their TATA

boxes (Fig 8) and whether it interacts with components

of the basal transcription complex (Figs 5 and 6)

In pull-down experiments (Fig 5, Fig S2), FOXM1c

bound to TBP, TFIIB, TFIIAa⁄ b, TFIIAc and

TAFII250 (TAF1) [52], but not to TFIIEa These

inter-actions are direct for TBP, TFIIB and TFIIAa⁄ b

because they could be verified using in vitro-translated

proteins (Fig 5) The respective interaction domains of

FOXM1c were each mapped to its central domain (see

below; Fig 5, Fig S2) Therefore, the interactions of

TAFII250 and⁄ or TFIIAc with FOXM1c may be

indi-rect via TBP or TFIIAa⁄ b, respectively The

inter-actions of FOXM1c with TBP, TFIIAa⁄ b, TFIIAc and

TAFII250 are also found in vivo because these proteins

could be coimmunoprecipitated with FOXM1c (Fig 6)

TBP bound strongly to FOXM1c ( 28% of the

input TBP was pulled down) (Fig 5B) Deletion

mutants of TBP showed that FOXM1c binds

predom-inantly to the C-terminal half of the conserved TBP

saddle (Fig 5B,C), which is orientated towards the

5¢-end of the TATA box [38,49,50]

More detailed mapping (Fig 5, Fig S2) showed that

TBP and TFIIB both bound to amino acids 380–425 of

FOXM1c, i.e to the EDA (amino acids 349–425)

(Fig 3F), but not to amino acids 1–379 or 574–762.TAFII250 interacted with amino acids 380–477 ofFOXM1c, but not with amino acids 1–379 TFIIAa⁄ band TFIIAc both probably interacted with amino acids359–477 of FOXM1c

In summary, FOXM1c binds directly, via its tially required EDA (amino acids 349–425) (Fig 3F),

essen-to the components TBP, TFIIAa⁄ b and TFIIB of thebasal transcription complex, which are positioned at ornear the TATA box, respectively FOXM1c(189–762)and FOXM1c(189–425; 568–762), which bound toTBP and TFIIB, transactivated the c-myc P1 and P2promoters, whereas FOXM1c(189–348; 573 762)NLS,which did not bind to TBP or TFIIB, failed to transac-tivate both promoters (Figs 3A,C,F, 5A, Fig S2A,F,G;data not shown) Consequently, these interactionsshould be important for the new mechanism by whichFOXM1c transactivates via the c-myc P1 and P2TATA boxes

Binding of TBP and FOXM1c to the P1 and P2TATA boxes

Because TBP binds to all TATA boxes the questionarose: what is the difference between the FOXM1c-responsive TATA boxes of c-myc P1 and c-myc P2versus the non-FOXM1c-responsive TATA boxes ofc-jun, waf1(p21) and HSV TK? The TBP⁄ TFIIA com-plex bound to the c-myc P2 TATA box (P2) with thesame very high affinity as to the identical TATA box

of the adenovirus 2 major late promoter (AdML)(Fig 7A), which is bound very strongly by TBP [50].Its binding affinity for the c-myc P1 TATA box (P1)was lower, although still high (Fig 7A) Its bindingaffinity for the FOXM1c-responsive TATA boxes ofc-myc P1 and c-myc P2 was higher than for the non-responsive TATA boxes of c-jun (jun), waf1(p21)(WAF) and HSV TK (mintk) (Fig 7B,C)

GST–FOXM1c(233–334), which comprised theforkhead domain (amino acids 235–332), and GST–FOXM1c(195–596) bound to the c-myc P1 andc-myc P2 TATA boxes (Fig 8C,D) These protein–DNAcomplexes were supershifted with an antibody [a-GST,a-FOXM1c(1B1)] that recognized the two GST–FOXM1c fusion proteins, but not with a control anti-body [a-FOXM1c(7E4)] (Fig 8C,D; data not shown).These protein–DNA complexes were competed by anexcess of unlabeled c-myc P1 TATA box or c-myc P2TATA box, respectively, but not by an excess ofunlabeled control oligonucleotides (Fig 8A,B,D) ThusFOXM1c binds in a sequence-specific manner and withhigh affinity to the c-myc P1 TATA box and thec-myc P2 TATA box, and the forkhead domain

Fig 5 Direct binding of FOXM1c to TBP, TFIIA and TFIIB (A, B) Pull-down assays were performed in the presence of ethidium bromide [87] with purified GST or the indicated GST–fusion proteins and the indicated in vitro-translated proteins Bound in vitro-translated proteins were detected following SDS⁄ PAGE by autoradiography The input control represents 1 ⁄ 10 of the volume used in the pull-down assays (B) Amount (%) of the input bound to GST–FOXM1c(1–477) wt, wild-type (C) (Upper) RASMOL drawing of the cocrystal structure of the C-ter- minal⁄ core region of human TBP complexed with the TATA element of the adenovirus major late promoter [49] TBP segments are colored

as indicated in the table DNA is shown in gray (Lower) Quantification of the pull-down assay in (B) Contribution (%) made by the TBP ments to total GST–FOXM1c(1–477) binding and which elements of the TBP saddle they included H, a helix; S, b strand.

seg-(amino acids 235–332) is sufficient for this DNA ing The order of binding affinities for the differentTATA boxes was similar for GST–FOXM1c(195–596)

bind-as for the TBP⁄ TFIIA complex (Fig 7C; data notshown) For comparison, the best conventionalFOXM1c-binding site HFH-11 [30] was bound byGST–FOXM1c(195–596) with lower affinity than thec-myc P1 and P2 TATA boxes (Fig 8B)

To examine in vivo binding of FOXM1c to the enous c-myc promoter chromatin immunoprecipitation(ChIP) assays were performed Figure 8E shows thatthe c-myc P1⁄ P2 TATA box region was enriched mark-edly more with a FOXM1c-specific antibody than with

endog-a control endog-antibody (endog-a-b-Gendog-al), indicendog-ating thendog-at in vivoFOXM1c binds to the c-myc promoter As a negativecontrol, the NE promoter (TATA box region) was lessimmunoprecipitated with the FOXM1c-specific anti-body than with the control antibody (Fig 8E), indicat-ing that in vivo this promoter is not bound by FOXM1c

Dominant-negative FOXM1c reduces cell growthc-Myc, a key factor for cell-growth control, potentlystimulates cell proliferation, promotes apoptosis andrepresses differentiation and entry into quiescence.c-Fos also stimulates proliferation, HSP70 and histoneH2B are required for its execution Consequently,transactivation of the four respective genes by FOXM1cshould increase proliferation By contrast, repression ofthese genes by dominant-negative FOXM1c shouldreduce proliferation FOXM1c(189–743)–Engr andFOXM1c(189–566)–Engr were constructed by replacingthe TAD (amino acids 721–762) or its C-terminal halfwith the repressor domain of Drosophila Engrailed(Figs 9A and S3C) These two dominant-negative forms

of FOXM1c repressed p(MBS)3-mintk-luc, the c-mycP1 promoter and the c-myc P2 promoter (Fig.S3A,B; data not shown) Thus they functioned asrepressors for all FOXM1c target genes regardless whe-ther activation is via TATA box binding or binding tothe conventional target sequences

In colony-formation assays, both FOXM1c(189–743)–Engr and FOXM1c(189–566)–Engr reduced the

HA-TBP

FOXM1c (189-762)

FOXM1c (189-762)

WB:α-HA

WB:α-FOXM1c

HA-TBP

FOXM1c (189-762)

+ +

FOXM1c (189-762)

WB:α-FOXM1c

HA-TFIIAγ

+ +

FOXM1c (189-762)

WB:α-FOXM1c

FOXM1c (189-762)

+ +

of the volume used in the coimmunoprecipitations a-FOXM1c, a-FOXM1c(C-20) (B) The control antibody a-C was a-cytochrome c.

number of colonies about threefold, whereas the

con-trols Gal–Engr and Engr–ER–myc had no significant

effect (Fig 9A–C) Therefore, this strong negative effect

on cell growth of the two former proteins depended

spe-cifically on their FOXM1c parts, which recruited them

to FOXM1c target genes Thus, they exerted their

growth-inhibitory effect by repression of FOXM1c

tar-get genes that are normally activated by FOXM1c

Con-sequently, FOXM1c should have a positive effect on cell

proliferation This proliferation-stimulating function ofFOXM1c confirms previous results for FOXM1 [11–21,23–27] Because no increase in apoptosis was found

in FoxM1-deficient mice livers [18,20] or pancreas [27],compared with control organs, and because RNAi ofFoxM1 did not induce apoptosis in breast cancer celllines [23] it is unlikely that the strong negative effect oncell growth of the two dominant-negative forms ofFOXM1c is based on an increased rate of apoptosis

C

P1 5'-ACCGGCCCTT T A T AA TGC GAGGGTCTG-3'

P2 5'-TCGCGCTGAG T A T AAAAG CCGGTTTTCG-3'

AdML 5'-GTTCCTGAAGGGGGGC T A T AAAAG GGGGTGGGGGCGCGTT-3'

jun 5'-GACTGGTAG CAGA T AAGTG TTGAGCTCGGG-3'

WAF 5'-G GGGCGGTTG T A T A TCAG GGCCGCGCTGAG-3'

mintk 5'-GATCCTTCG CA T A TT AAGG TGACGCGTGTG-3'

-66 5'-TCAGA GGCTTGGCGGGAAA AAGAACG-3' SV40 5'-GGAACT GGGCGGAGTTAGGGG-3' CMD 5'-TCAGAC CACGTGGTCGGG-3' HFH-11 5’-TCGACGAAAAAA ACAAA T AACAACGTACTCGA-3’

CATATTAA

TK CAGATAAG

c-jun TATATCAG

waf1(p21) TATAATGC

c-myc -P1 TATAAAAG

c-myc -P2

DNA binding affinity

Fig 7 TBP binds to the P1 and P2 TATA boxes (A, B) EMSAs were performed with radioactively labeled oligonucleotides P1 or P2 and with purified TBP and TFIIA or as control (c) without TBP and TFIIA For supershifts (A), the antibodies a-HA and a-TBP were used For competi- tions (A, B), unlabeled oligonucleotides were used in excess (A) P2, P1 and AdML, 5-, 20- or 100-fold; SV40, 100-fold (B) mintk, WAF, P1, P2, and jun, 5-, 20- or 100-fold; CMD, SV40 and AdML, 100-fold, S, supershift; F, free probe; T, gel slot (C) Binding affinities of the TBP⁄ TFIIA complex and GST–FOXM1c(195–596) for the different TATA boxes For c-jun and TK both possible TATA box positions are shown The TATA box definitions of Patikoglou et al [50] and Bucher [90] and the general TATA box consensus sequence are indicated (D)

In the oligonucleotides TATA boxes (bold and underlined), E-boxes (CMD) and binding sites for Sp1 (SV40, )66, WAF), FOXM1c (HFH-11), E2F, STAT3, ETS, NFATc1, Smad and METS ( )66) (underlined) are marked For transcription factor binding sites in the oligonucleotide )66 see p( )66)mintkluc in Fig 1D.