Báo cáo khoa học: LIN54 is an essential core subunit of the DREAM / LINC complex that binds to the cdc2 promoter in a sequence-specific manner ppt

We identified two binding sites for LIN54 in the cdc2 promoter, one of which overlaps with the cell cycle homology region at the transcrip-tional start site.. Structured digital abstract

complex that binds to the cdc2 promoter in a

sequence-specific manner

Fabienne Schmit, Sarah Cremer and Stefan Gaubatz

Department of Physiological Chemistry I, Biocenter, University of Wuerzburg, Germany

Introduction

The retinoblastoma tumor suppressor protein (pRB)

and the related pocket proteins (p107 and p130)

func-tion together with E2F transcripfunc-tion factors to regulate

the cell cycle, differentiation and development [1] The

activity of the pocket proteins is regulated by

cyclin-dependent kinases which phosphorylate the pocket

proteins and thereby release E2F, which then further

transcriptionally activates cyclins and other cell

cycle-dependent genes [2]

We and others have recently identified a new E2F– pocket protein complex in mammalian cells that plays

an important role in gene repression in quiescent cells and in the activation of mitotic genes [3–5] This com-plex, called LINC or human DREAM, consists of a five-protein core module that dynamically associates with p130 and the E2F4 and B-MYB transcription fac-tors in a cell cycle-dependent manner LINC has been highly conserved throughout evolution A similar

Keywords

cell cycle; CXC; DNA binding; LIN54;

LINC/DREAM

Correspondence

S Gaubatz, Department of Physiological

Chemistry I, Biocenter, University of

Wuerzburg, 97074 Wuerzburg, Germany

Fax: +49 931 3184150

Tel: +49 931 3184138

E-mail: stefan.gaubatz@biozentrum.

uni-wuerzburg.de

(Received 4 June 2009, revised 23 July

2009, accepted 5 August 2009)

doi:10.1111/j.1742-4658.2009.07261.x

Recently, the conserved human LINC⁄ DREAM complex has been described as an important regulator of cell cycle genes LINC consists of a core module that dynamically associates with E2F transcription factors, p130 and the B-MYB transcription factor in a cell cycle-dependent manner

In this study, we analyzed the evolutionary conserved LIN54 subunit of LINC We found that LIN54 is required for cell cycle progression Protein interaction studies demonstrated that a predicted helix–coil–helix motif is required for the interaction of LIN54 with p130 and B-MYB In addition,

we found that the cysteine-rich CXC domain of LIN54 is a novel DNA-binding domain that binds to the cdc2 promoter in a sequence-specific manner We identified two binding sites for LIN54 in the cdc2 promoter, one of which overlaps with the cell cycle homology region at the transcrip-tional start site Gel shift assays suggested that, in quiescent cells, the bind-ing of LIN54 at the cell cycle homology region is stabilized by the bindbind-ing

of E2F4 to the adjacent cell cycle-dependent element Our data demon-strate that LIN54 is an important and integral subunit of LINC

Structured digital abstract

l MINT-7239362 : LIN54 (uniprotkb: Q6MZP7 ) physically interacts ( MI:0915 ) with p130 (uniprotkb: Q08999 ) by anti tag coimmunoprecipitation ( MI:0007 )

l MINT-7239376 : LIN54 (uniprotkb: Q6MZP7 ) physically interacts ( MI:0915 ) with B-Myb (uniprotkb: P10244 ) by anti tag coimmunoprecipitation ( MI:0007 )

Abbreviations

BrdU, bromodeoxyuridine; CDE, cell cycle-dependent element; CHR, cell cycle homology region; pRB, retinoblastoma tumor-suppressor protein; shRNA, small hairpin RNA.

complex, called DREAM or MMB, was first identified

in Drosophila [6,7] The Caenorhabditis elegans

homo-logs of DREAM and LINC form the highly related

DRM complex that mainly acts in the repression of

developmental genes [8]

Unlike the Drosophila and C elegans complexes, the

composition of human LINC is regulated during the

cell cycle Specifically, in quiescent cells, LINC

inter-acts with p130 and E2F4 and contributes to the

repres-sion of E2F target genes [3,5,9] The binding of LINC

with E2F4⁄ p130 is disrupted in the S phase At this

time, B-MYB, a member of the vertebrate MYB

fam-ily, is incorporated into the complex LINC–B-MYB

binds to and activates the promoters of G2⁄ M genes

[4,10,11] One particular well-studied subunit of LINC

is the LIN9 protein LIN9 depletion in human and

mouse cells leads to reduced activation of G2⁄ M target

genes and cell cycle inhibition [9] LIN9 also plays a

role in zebrafish development and in transcriptional

regulation in response to DNA damage [12,13]

Another conserved core subunit of LINC is LIN54

LIN54 homologs are present in many species,

includ-ing plants (Arabidopsis thaliana), invertebrates (C

ele-gans and Drosophila melanogaster) and humans, who,

in addition to LIN54, express a testis-specific homolog,

tesmin [14–18] LIN54 homologs contain dual

cysteine-rich domains, termed CXC, with the consensus

sequence CXCX4CX3YCXCX6CX3CXCX2C The two

CXC domains are separated by a short spacer It is

possible, yet unproven, that the dual CXC domain of

LIN54 functions as a DNA-binding domain

In this study, we investigated the function of LIN54

We found that LIN54 is required for the proliferation

of human cells Using LIN54 deletion mutant

con-structs, we demonstrated that a conserved helix–coil–

helix (HCH) region is essential for the binding of

LIN54 to p130 and B-MYB We also found that the

dual CXC domain of LIN54 is a DNA-binding

domain that binds to the cdc2 promoter in a

sequence-specific manner We identified two binding sites for

LIN54 in the cdc2 promoter, one of which overlaps

with the cell cycle homology region (CHR) at the

tran-scriptional start site Our data suggest that, in

quies-cent cells, the binding of LIN54 at CHR is stabilized

by E2F4 bound to the adjacent cell cycle-dependent

element (CDE) Taken together, these data establish

LIN54 as an essential member of the LINC⁄ DREAM

complex

Results

We have shown previously that the depletion of LIN9

in human cells inhibits proliferation and results in

delayed entry into mitosis [9] To address whether this

is an isolated function of LIN9 or whether it is medi-ated by LINC, we examined another subunit of LINC, LIN54 Immortalized BJ cells containing the

ecotroph-ic receptor (BJ-ET) were infected with a retrovirus encoding a small hairpin RNA (shRNA) against LIN54, or with a control retrovirus The efficiency of LIN54 depletion was tested at the mRNA and protein level LIN54 mRNA was reduced by 70% and a signif-icant reduction in the protein level was detected in immunoblots in cells infected with the LIN54-specific shRNA (Fig 1A, B)

Next, the proliferation of LIN54-depleted cells com-pared with control cells was analyzed When LIN54-depleted cells were monitored for 12 days in culture,

we found that they grew significantly more slowly than control cells, indicating that LIN54 is required for the proliferation of human cells (Fig 1C) To better char-acterize the cell cycle defects, flow cytometry profiles

of control cells and LIN54-depleted BJ-ET cells were compared Cells were labeled with bromodeoxyuridine (BrdU) to identify the population of cells in S phase

As shown in Fig 2A, LIN54 depletion results in an accumulation of cells in G2⁄ M and in a reduction in the fraction of cells in S phase Staining with an anti-body against phosphorylated histone H3 revealed a decrease in the percentage of mitotic cells in LIN54-depleted cells compared with control cells (Fig 2B) Together with the fluorescence-activated cell sorting data, this indicates that cell cycle inhibition occurs in G2 before entry into mitosis To address whether LIN54-depleted cells are completely blocked in G2, the kinetics of cell cycle progression were analyzed in more detail LIN54-depleted cells were labeled with BrdU for 1 h BrdU was washed away and, 6 h later, the cells were analyzed by flow cytometry; 11.6% of the control cells that were in S phase during labeling had progressed to the next G1 phase at 6 h after labeling (Fig 2C); in contrast, only 4.5% of the LIN54-depleted S-phase cells had re-entered into the next G1 phase at 6 h after labeling This shows that the G2 phase of LIN54-depleted cells is prolonged, but not blocked completely Taken together, these results indi-cate that LIN54 is required for entry into mitosis of human cells These results are consistent with the recent identification of LIN54 in an RNA interference (RNAi)-based screen for cell division defects in human cells [19]

LIN54 is an evolutionarily conserved protein that contains a conserved sequence in the C-terminus that is predicted to form an HCH secondary structure (Fig 3A) The corresponding region of tombola, a testis-specific D melanogaster homolog of LIN54, is

required for binding to the interaction partner Aly [14].

It is possible that the HCH region of human LIN54 is

also involved in protein–protein interaction To address

this possibility, we generated a set of LIN54 deletion mutants (Fig 3B) Flag-tagged LIN54 wild-type or mutants were transfected into 293 cells HA-tagged

0 0.2 0.4 0.6 0.8

1

Ctrl LIN54

(normalized to GAPDH) shRNA

C

LIN54

β-tubulin

Ctrl LIN54 shRNA

Ctrl

LIN54 kd

0

5

10

15

20

25

0 4 8 12

Days

Fig 1 Depletion of LIN54 leads to growth

defects hTert immortalized BJ fibroblasts

(BJ-ET) were infected with retroviral shRNA

against LIN54 LIN54 mRNA (A) and protein

(B) levels after shRNA depletion compared

with the levels in control infected cells (C)

Infected and selected BJ-ET cells were

counted and reseeded in triplicate in each

experiment Cell numbers were analyzed

and the cumulative growth was plotted

against time The experiment was repeated

three times One representative experiment

is shown.

DNA content (PI intensity)

59.4 % 17.8 %

66.3 % 2.7 % G1:

S:

G2/M: 10.3 %

G1:

S:

G2/M: 22.0 %

DNA content (PI intensity)

BrdU positive cells

in G1: 11.6 %

BrdU positive cells

in G1: 4.5 %

C

A

Phospho-H3 positive cells

B

Fig 2 Depletion of LIN54 leads to defects in the G2 ⁄ M transition (A) LIN54-depleted and control-depleted BJ-ET cells were labeled with BrdU and analyzed by flow cytometry to determine the percentage of cells in different phases of the cell cycle (B) Control and LIN54-depleted cells were stained with an antibody against phosphorylated histone H3 (PH3) Nuclei were counterstained with Hoechst 33258 At least 300 cells were counted in triplicate in each experiment and the percentage of PH3-positive cells was determined Representative regions of one of three independent experiments are shown (C) To determine the number of cells re-entering the next cell cycle, control-depleted and LIN54-depleted BJ-ET cells were pulse labeled with BrdU for 1 h and grown for an additional 6 h without BrdU The percentage of BrdU-positive cells in G1 was determined by flow cytometry.

p130 was coexpressed The binding of LIN54 mutants

to p130 was analyzed by immunoprecipitation of

LIN54 from lysates with anti-flag IgG and the detection

of bound p130 with anti-HA IgG Compared with the

other deletion mutants, we found that it was very

diffi-cult to overexpress full-length LIN54, DHCH and DN

mutants in a panel of different cell lines Although

diffi-cult to overexpress, distinct bands corresponding to

LIN54 and the DHCH and DN mutants could be

detected (marked by asterisks) Fluorescence-activated

cell sorting studies did not suggest that the expression

of LIN54, DHCH and DN was toxic to the cells (data not shown) Therefore, it is possible that full-length LIN54, DHCH and DN are unstable Although some LIN54 constructs were difficult to express, coimmuno-precipitation experiments showed unequivocally that p130 bound to full-length LIN54 and to the DN, DCXC and HCH mutants, all of which contained an intact HCH region (Fig 3C) In contrast, mutants that lacked the HCH region (DHCH and CXC) showed no binding

A

B

pCDNA LIN54 LIN54

HA-B-MYB

* * *

*

*

*

*

Bound HA-B-MYB (IP: flag > WB: HA)

Input HA-B-MYB

10% gel Input flag-LIN54

15% gel Input flag-LIN54

pCDNA

HA-p130

Input HA-p130

LIN54 LIN54

*

*

*

*

*

*

*

(IP: flag > WB: HA) Bound HA-p130

10% gel Input flag-LIN54

15% gel Input flag-LIN54

LIN54 LIN54 ΔN LIN54 CXC

LIN54 ΔCXC

LIN54 HCH LIN54 ΔHCH

H.s LIN54 H.s Tesmin

C.e LIN-54

D.m mip120 D.m tombola

G.m CPP1 A.t TSO1

1

1

1

1

1

1

1

749

695

896

243

950

435

509

CXC Helix-Coil-Helix

Fig 3 The putative helix–coil–helix (HCH) region of LIN54 is required for binding to p130 and B-MYB (A) Schematic alignment of LIN54 pro-teins A.t., Arabidopsis thaliana; C.e., Caenorhabditis elegans; D.m., Drosophila melanogaster; G.m., Glycine max; H.s., Homo sapiens (B) Scheme of the flag-tagged LIN54 deletion mutants to analyze the function of the conserved CXC and HCH domains (C, D) 293T cells were co-transfected with the indicated flag-tagged LIN54 constructs and HA-p130 (C) or HA-B-MYB (D) Whole-cell lysates were immunoprecipitated with anti-flag agarose Input levels of full-length LIN54 and mutants were detected by immunoblotting with flag antibodies (indicated by aster-isks) Because of their small size, 15% SDS gels were used to detect LIN54 CXC and HCH mutants Bound HA-p130 and HA-B-MYB were detected by immunoblotting.

to p130 Importantly, these two mutants were both

strongly overexpressed, indicating that a lack of

bind-ing of these mutants to p130 was not a result of weak

expression

In similar experiments, binding of B-MYB could

also be confined to HCH of the LIN54 region

(Fig 3D) These data indicate that the LIN54 HCH

domain is required for the interaction of LIN54 with

p130 and B-MYB

Chromatin immunoprecipitation assays have shown

that LINC subunits, including LIN54, bind to the

promoters of mitotic genes, such as cdc2 [4] As

described above, LIN54 proteins contain two

cysteine-rich CXC domains that could function as a

DNA-binding domain (Fig 3A) In this study, therefore, we

wanted to address whether purified LIN54 can

inter-act directly with DNA in a sequence-specific manner

Because it was not possible to express full-length

LIN54 in heterologous bacterial expression systems,

we decided to focus on the CXC domains The dual

CXC domain of LIN54 was fused to GST

(GST-CXC), expressed in bacteria and affinity purified

(Fig 4A) Next, we performed gel shift experiments

with GST-CXC and with a 400 bp fragment of the

human cdc2 promoter, a known LINC target gene

[4,9] As shown in Fig 4B, binding of GST-CXC to

the cdc2 promoter was readily detected Binding was

competed with the unlabeled cdc2 promoter,

indicat-ing that the bindindicat-ing was specific Next, to test

whether the conserved cysteines in the dual CXC

domain were required for binding to the cdc2

pro-moter, we mutated two conserved cysteine residues in the first CXC domain (C525 and C527) to tyrosine Mutation of the residue equivalent to C525 of Ara-bidopsis TSO1 disrupts flower development and cell division [20] We found that mutation of C525 and C527 abolished the binding of GST-CXC to the cdc2 promoter (Fig 4B) Thus, the conserved cysteine resi-dues are essential for the binding of LIN54 to DNA Previous studies have indicated that multiple posi-tively and negaposi-tively acting elements are involved in cell cycle-dependent transcription of the human cdc2 promoter [21,22] For example, a CDE–CHR element overlapping the transcription start site mediates repres-sion in quiescent cells and an E2F site is responsible for activation by E2F1–3 [22,23] In addition, a MYB binding site upstream of the E2F site and CCAAT elements between the E2F sites are involved in cdc2 promoter regulation [23] Examination of the cdc2 promoter sequence identified four additional putative MYB elements and a sequence related to a CHR ele-ment adjacent to MYB binding site 1 (Fig 5A) To address which elements are involved in the binding of LIN54 to the cdc2 promoter, we generated mutated cdc2 promoter fragments and used them in competi-tion experiments (Fig 5)

In the first set of experiments, we analyzed the CDE-CHR region at the transcription start site, the potential MYB binding sites and the upstream region with CHR homology (CHR-up) (Fig 5C) As expected, the binding of GST-CXC to the cdc2 pro-moter was competed with the wild-type cdc2 propro-moter

GST- CXC

GST- CXC- C525/

527Y GST

*

100

72

55

40

24

33

M

GST-CXC

GST-CXC- C525/527Y

Free probe

GST- CXC

Competitor

Fig 4 The dual CXC domain of LIN54 binds

to the cdc2 promoter (A) Recombinant

GST, GST-CXC (comprising the two CXC

domains of human LIN54) and

GST-CXC-C525 ⁄ 527Y (with two C to Y point

mutations in the first CXC domain) were

expressed in bacteria and purified A

Coo-massie-stained gel of the purified proteins is

shown The asterisks indicate the position

of GST and the GST-CXC fusion proteins.

(B) Recombinant GST-CXC and

GST-CXC-C525 ⁄ 527Y were incubated with a [ 32

P]-labeled fragment of the cdc2 promoter and

separated on a non-denaturing acrylamide

gel; 250 ng of unlabeled cdc2 promoter

frag-ment was used to compete for the binding

of the GST protein to the labeled probe (+).

Labeled DNA was detected by

autoradiogra-phy The positions of the free probe and the

shifted GST-CXC protein are indicated.

cdc2 promoter fragments in which the CDE region,

the upstream CHR region or potential MYB binding

sites 1, 4 or 5 were mutated (MYB1-mut, MYB4-mut,

MYB5-mut) also specifically competed for binding

This indicates that these regions are not essential for

the binding of LIN54 to the cdc2 promoter In

con-trast, the same amount of a promoter fragment with a

mutated CHR region (CHR-mut) did not compete

effi-ciently, indicating that the CHR region is important

for the binding of LIN54 to the cdc2 promoter (Fig 5C)

We next asked whether the same element was involved in the binding of a LIN54-containing com-plex from cellular lysates To address this possibility,

we performed gel shift assays with nuclear extracts from T98G cells and an oligonucleotide encompassing the CDE-CHR element of the cdc2 promoter as a probe Because the CDE-CHR element is occupied by

CDE

CHR

CHR-up

MYB1

MYB4

MYB5

Wt sequence

TAGCGCGGT AGTTTGAAAC ATTTGAA GAACTGTG AGAAACAGT CAGTTGGCG

Mut sequence

TAGCGCtGT AGTagctAAC ATccGAA GAAtcGTG AGAggaAGT CAGcctGCG

A

B

wt MYB1 MYB2 MYB3 MYB4 MYB5 E2F CAAT CAAT CDE CHR

MYB4 mut MYB5 mut

CDE mut CHR mut CHR-up mut MYB1 mut

5 ′ del CHRup

GST-CXC

Competitor:

GST-

CXC

CDE mut CHR mut CHRup mut MYB1 mut MYB4 mut MYB5 mut

Free probe

C

Fig 5 Binding of LIN54 to the CHR element of the cdc2 promoter Scheme (A) and sequences (B) of the mutated cdc2 promoter mutants used for competition in electrophoretic motility shift assay experiments (C) Purified GST-CXC was incubated with a [ 32 P]-labeled fragment

of the cdc2 promoter Competition was performed with 100 ng of the unlabeled wild-type cdc2 promoter or the indicated promoter mutants.

repressor complexes in G0, we prepared nuclear

extracts from serum-starved cells In gel shift

reac-tions, a specific band shift was observed that was

competed with a 100-fold excess of unlabeled

oligo-nucleotide, suggesting that the binding is specific (Fig 6A) To identify which part of the CDE-CHR is required for this binding activity, we performed com-petition experiments with oligonucleotides in which

C

D

G0-compl

*

*

Free probe

NE – –

0.25 0.25 0.5 0.5

NE

Comp.:

α-LIN54 (µg):

Peptide (µg):

– –

–

G0-compl

G0-compl

– wt E2F –

NE

p130 E2F4 LIN54

– + + + + + +

– E2F1

Ab:

– LIN9 Ab:

WT CDE CHR

– – Comp.:

G0-compl

E

0

10

20

15

5

25

Fig 6 Binding of LINC from nuclear

extracts of serum-starved T98G cells to the

cdc2 CDE-CHR element (A) Gel shift

analysis with nuclear extracts from

serum-starved T98G cells and the CDE-CHR

ele-ment of the cdc2 promoter Competition

was performed with a 100-fold excess of

the unlabeled oligonucleotide (B) Gel shift

assay as described in (A) Competition with

a 100-fold excess of wild-type CDE-CHR or

with oligonucleotides with mutation in the

CDE or CHR region (C) A gel shift assay

was performed as described in (A)

Compet-itor oligonucleotides or polyclonal antibodies

were added as indicated E2F, competition

with the canonical E2F-binding site from the

DHFR promoter is an additional

demonstra-tion that the G0-specific complex contains

E2F (D) The LIN54 antibody was

preincu-bated with the indicated amount of peptide

against which it was raised A gel shift

assay was performed as described in (A).

NE, nuclear extract; *, nonspecific band (E)

The activities of the indicated cdc2 promoter

constructs (wild-type, CDE mutant or CHR

mutant) were compared in a luciferase

reporter assay.

either the CDE or CHR element was mutated As

shown in Fig 6B, the mutated oligonucleotides were

unable to compete for binding, indicating that both

elements are necessary for the binding of the complex

in serum-starved cells Next, we asked whether the

G0 binding activity contains members of the LINC

complex To address this possibility, we performed gel

shift experiments with antibodies directed at LINC

subunits As a control, we used an antibody directed

at E2F1, which is not a component of LINC and

which did not interfere with the formation of the G0

phase-specific band shift (Fig 6C, middle panel) In

contrast, polyclonal antibodies against p130, E2F4,

LIN9 and LIN54 abolished binding (Fig 6C) To

ver-ify that the inhibition of binding by the LIN54

anti-body is a specific effect, we preincubated LIN54

antiserum with the peptide against which it was

raised After blocking with the peptide, the antibody

failed to affect complex formation, verifying the

speci-ficity of the LIN54 antibody (Fig 6D) Together,

these data demonstrate that a complex containing

p130, E2F4, LIN9 and LIN54 binds to the

CDE-CHR element of the cdc2 promoter in serum-starved

cells, and that both parts of the CDE-CHR element

are required for binding of the complex These

find-ings are consistent with previous studies that reported

the binding of E2F4 and p130 to the CDE element in

G0 [22] Binding of LIN54 to the adjacent CHR

element could therefore stabilize the binding of

E2F4⁄ p130 at CDE To address whether the

CDE-CHR element is required for repression, we performed

luciferase reporter assays using the same CDE and

CHR mutations as were employed in the gel

retarda-tion experiments Consistent with previous studies, the

activity of the cdc2 reporter constructs that contained

mutations in either the CDE or CHR element was

strongly increased when compared with the wild-type

promoter (Fig 6E) Thus, the loss of DNA binding

by the LINC complex correlates with increased

activ-ity of the cdc2 promoter

Because chromatin immunoprecipitation experiments

indicate that LINC also binds to the cdc2 promoter in

the S phase [9], we next asked whether LINC also

binds to the CDE-CHR element during this phase of

the cell cycle In S-phase extracts, however, the

G0-specific band was no longer present, indicating that

LINC dissociates from CDE-CHR after cell cycle

re-entry (data not shown) Because LINC is no longer

present at CDE-CHR in the S phase, it should bind to

other elements in the cdc2 promoter at this time in the

cell cycle This binding could be mediated by

addi-tional LIN54-binding sites in the cdc2 promoter, or

may be independent of LIN54

To address the possibility of additional LIN54-bind-ing sites in the cdc2 promoter, we performed addi-tional gel shift experiments with purified GST-CXC and the cdc2 promoter In competition experiments in which the CHR region was mutated, little competition was observed with small amounts of this competitor,

as shown previously However, partial competition was observed with larger amounts This indicates that the binding of GST-CXC is reduced, but not completely abolished, when the proximal CHR element is mutated (Fig 7A) To further confirm that LIN54 can still bind

to the cdc2 promoter with a mutated CHR element, the reverse experiment was performed and CHR-mut was used as a probe in gel shift experiments (Fig 7A) GST-CXC was able to shift this mutated promoter construct, confirming that it can indeed still bind to a promoter in which the CHR element is mutated Thus,

it appears that there are additional binding sites for LIN54 in the cdc2 promoter

Further evidence for the presence of additional bind-ing sites came from the observation that although a 5¢ cdc2 promoter deletion (5¢ del) competed for binding

it required higher concentrations than the longer cdc2 promoter to fully compete for binding (Fig 7B,C) Thus, a second binding site for LIN54 appears to be located in or close to the upstream region that is miss-ing in this deletion

Further support for this concept came from the observation that a cdc2 promoter fragment that con-tained a deletion at the 5¢ region and a mutated down-stream CHR site (CHRmut + del) did not compete for binding (Fig 8B, lane 4) To define more precisely the upstream binding site for LIN54, we generated a set of point mutants Each of the constructs contained

a three-base-pair mutation within the upstream LIN54-binding region, in addition to a mutation in the downstream CHR element (mutants A–I, Fig 8A) The downstream CHR region was mutated in these constructs, because we wanted to analyze binding to the upstream region independently from binding to the downstream CHR element Mutated cdc2 promoter fragments were used as competitors in gel shift experi-ments As shown in Fig 8B, the ability of mutant E (lane 10) to compete was reduced slightly, whereas the two mutants G and I (lanes 12 and 14) completely lost their ability to compete for LIN54 binding Thus, in addition to the CHR region, the cdc2 promoter regions corresponding to mutants G and I are required for the binding of LIN54 Taken together, these data indicate that LIN54 interacts with the cdc2 promoter

at two different sites Although not addressed in this study, it is possible that the in vivo binding of LIN54

to the upstream element is stabilized by the binding of

B-MYB to adjacent low-affinity sites Our results

provide the basis for a further investigation of the

regulation of the cdc2 promoter by LINC in different

phases of the cell cycle

Discussion

The primary goal of this study was to investigate

LIN54, a conserved subunit of the human DREAM⁄

LINC complex We found that RNAi-mediated

deple-tion of LIN54 in primary human cells results in cell

cycle arrest and delayed entry into mitosis A similar

phenotype has been found previously on depletion of

human LIN9 and B-MYB [9–11] This suggests that

the ability to promote cell cycle progression and

mito-tic entry is not an isolated function of LIN9 and

B-MYB, but that it is mediated by the LINC complex Our study demonstrates that LIN54 is an integral and essential subunit of this complex

To analyze the function of LIN54, we created a set

of deletion mutants Using these mutants, we found that a region of LIN54 that is predicted to form an HCH secondary structure is required and sufficient for binding to p130 and B-MYB Because the correspond-ing region of tombola, a testis-specific Drosophila homolog of LIN54, is involved in binding to the LIN9 homolog Aly [14], it is possible that the LIN54 HCH domain does not interact directly with p130 and B-MYB, but is required for the formation of the LINC complex, with which p130 and B-MYB interact LIN54 is an evolutionarily conserved protein of the tesmin⁄ TSO1 family [14,24] Members of this family

A

50 100 200 – 50 100 200

GST-

CXC

wt

– Comp (ng)

CHRmut Probe:

Free

probe

C

B

5′ del wt – Comp

GST-CXC

GST- CXC

Free probe

50 100 150 200 –

GST-CXC 5′ del (ng)

GST-

CXC

Fig 7 Evidence for additional binding sites for LIN54 in the cdc2 promoter (A) Purified GST-CXC was incubated with a [32P]-labeled frag-ment of the cdc2 promoter or a cdc2 promoter carrying a mutation in the CHR region Competition was performed with the indicated amounts of wild-type cdc2 promoter or CHR mutant promoter fragment (B) GST-CXC was incubated with a [ 32 P]-labeled fragment of the cdc2 promoter Bound labeled probe was competed with 150 ng of wild-type or deleted cdc2 promoter (C) Gel shift assay with GST-CXC and the cdc2 promoter Competition experiments were performed with increasing amounts of 5¢ del.

contain one or two related cysteine-rich domains,

termed the CXC domain CXC proteins have been

found in plants and animals, but not in yeast The

founding family member TSO1 is essential for flower

development [17,18] Mutations in TSO1 cause defects

in mitosis and cytokinesis Interestingly, all tso1

mutants that have been described harbor point

muta-tions in the CXC domain, indicating that this domain

is critical for the function of this protein A function

for the CXC domain is suggested by the observation

that soybean CPP1, a member of the family, binds to

the promoter of the leghemoglobin gene Gmlbc3

through the cysteine-rich domains [25] Moreover,

Mip120, the Drosophila LIN54 protein of the DREAM

complex, and C elegans Lin-54 also bind to DNA

[15,16] Therefore, it has been suggested that the CXC

domain functions as a sequence-specific DNA-binding

domain To test this concept directly, we performed

gel shift experiments with the dual CXC domain of

human LIN54 We found that the CXC domain binds

to the cdc2 promoter, an in vivo target of LINC

Bind-ing was critically dependent on the conserved cysteines,

as mutation of two cysteine residues in the first CXC

domain completely abolished binding to the cdc2

pro-moter Interestingly, our data indicate that there are

two LIN54-binding sites in the cdc2 promoter The

first binding site overlaps with the CHR at the tran-scriptional start site of the cdc2 promoter The second binding site is found further upstream in the cdc2 pro-moter, and is located between two potential MYB-binding sites

The CHR element, which is required for the binding

of LINC subunits in quiescent cells, is typically found

in cell cycle-regulated promoters adjacent to a second element, termed the ‘cell cycle-dependent element’ or CDE [26] Mutation in either the CDE or CHR ele-ment abolishes repression of cell cycle genes in G0⁄ G1 [27–30] Previous studies have reported the binding of E2F4 and p130 to the CDE part of the composite element [22,31] Because LIN54 binds to CHR, inter-action of E2F4⁄ p130 with CDE could be stabilized by binding of LIN54 to the adjacent CHR element Two different unidentified in vitro binding activities that interact with CHR have been described [32–34] The first, CDF-1, interacts with CHR elements of multiple promoters in quiescent cells [32,33] However, unlike LINC, CDF-1 does not interact with E2Fs or pocket proteins, suggesting that CDF-1 is unrelated to LIN54 Because the second known CHR-interacting activity, termed CHF, more selectively interacts with the CHR element found in the cyclin A promoter, it is also unli-kely that it corresponds to LIN54 [34] Although

A

B

MYB2 MYB3 MYB4 MYB5 E2F CAAT CAAT CDE CHR

MYB1

CHR A B C D E F G H I

A T T T G A A C T G T G C C A A T G C T G G G A G A A A A A A T T T A A G A T C T

CHRup MYB1

A T T T G A A C T A G A C C A A T G C T G G G A G A A A A A A T T T A A G A T C T M u t A

A T T T G A A C T G T G A A G A T G C T G G G A G A A A A A A T T T A A G A T C T M u t B

M u t C

M u t D

A T T T G A A C T G T G C C A G G A C T G G G A G A A A A A A T T T A A G A T C T

A T T T G A A C T G T G C C A A T G A G A G G A G A A A A A A T T T A A G A T C T

A T T T G A A C T G T G C C A A T G C T G A A G G A A A A A A T T T A A G A T C T

A T T T G A A C T G T G C C A A T G C T G G G A A G G A A A A T T T A A G A T C T

A T T T G A A C T G T G C C A A T G C T G G G A G A A G G G A T T T A A G A T C T

A T T T G A A C T G T G C C A A T G C T G G G A G A A A A A G G G T A A G A T C T

A T T T G A A C T G T G C C A A T G C T G G G A G A A A A A A T T G G G G A T C T

M u t E

M u t F

M u t G

M u t H

M u t I

A B C D E F G H I – wt del wt

CHR mut + GST-CXC

Free

probe

Comp.

GST-CXC

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Fig 8 Identification of the 5¢-binding region for LIN54 in the cdc2 promoter (A) Scheme

of the mutants in the 5¢ region of the cdc2 promoter used for competition in gel shift experiments (B) GST-CXC was incubated with a [ 32 P]-labeled fragment of the cdc2 promoter Bound labeled probe was competed with the indicated fragments containing a mutated triplet in the 5¢ region and a mutated CHR region Del, construct contains deletion of the 5¢ region in addition

to a mutated CHR region.