The HIC1 central region recruits the corepressor CtBP C-ter-minal binding protein through a conserved GLDLSKK motif, a variant of the consensus C-terminal binding protein interaction dom
Trang 1abolishes its interaction with the corepressor CtBP
Nicolas Stankovic-Valentin1,*, Alexis Verger2, Sophie Deltour-Balerdi1,†, Kate G R Quinlan2,
Merlin Crossley2and Dominique Leprince1,*
1 CNRS UMR 8526, Institut de Biologie de Lille, Institut Pasteur de Lille, France
2 School of Molecular and Microbial Biosciences, University of Sydney, New South Wales, Australia
HIC1 (hypermethylated in cancer 1) encodes a
tran-scriptional repressor and is located in 17p13.3 in a
region frequently hypermethylated or deleted in many
types of prevalent human tumour [1] HIC1 is a
tumour suppressor gene, since heterozygous HIC1+⁄ –
mice develop, after 70 weeks, a gender-dependent
spec-trum of spontaneous malignant tumours [2] HIC1 is a direct target gene of P53 [1,3,4] Moreover, elegant animal models using Hic1 and p53 double heterozy-gous knockout mice have shown that the epigenetically silenced gene, Hic1, cooperates with the mutated tumour suppressor gene Trp53 in determining cancer
Keywords
17p13.3; CtBP; HIC1; Miller–Dieker
syndrome; transcriptional repression
Correspondence
D Leprince, CNRS UMR 8161, Institut de
Biologie de Lille, Institut Pasteur de Lille,
1 Rue Calmette, 59017 Lille Cedex, France
Fax: +33 3 20 87 1111
Tel: +33 3 20 87 1119
E-mail: dominique.leprince@ibl.fr
Present address
*CNRS UMR 8161, Institut de Biologie de
Lille, Institut Pasteur de Lille, France
†Wellcome Trust, University of Cambridge,
UK
(Received 3 March 2006, revised 27 April
2006, accepted 2 May 2006)
doi:10.1111/j.1742-4658.2006.05301.x
HIC1 (hypermethylated in cancer) is a tumour suppressor gene located in 17p13.3, a region frequently hypermethylated or deleted in many types of prevalent human tumour HIC1 is also a candidate for a contiguous-gene syndrome, the Miller–Dieker syndrome, a severe form of lissencephaly accompanied by developmental anomalies HIC1 encodes a BTB⁄ POZ-zinc finger transcriptional repressor HIC1 represses transcription via two auton-omous repression domains, an N-terminal BTB⁄ POZ and a central region,
by trichostatin A-insensitive and trichostatin A-sensitive mechanisms, respectively The HIC1 central region recruits the corepressor CtBP (C-ter-minal binding protein) through a conserved GLDLSKK motif, a variant of the consensus C-terminal binding protein interaction domain PxDLSxK⁄ R Here, we show that HIC1 interacts with both CtBP1 and CtBP2 and that this interaction is stimulated by agents increasing NADH levels Further-more, point mutation of two CtBP2 residues forming part of the structure
of the recognition cleft for a PxDLS motif also ablates the interaction with
a GxDLS motif Conversely, in perfect agreement with the structural data and the universal conservation of this residue in all C-terminal binding pro-tein-interacting motifs, mutation of the central leucine residue (leucine 225
in HIC1) abolishes the interaction between HIC1 and CtBP1 or CtBP2 As expected from the corepressor activity of CtBP, this mutation also impairs the HIC1-mediated transcriptional repression These results thus demon-strate a strong conservation in the binding of C-terminal binding protein-interacting domains despite great variability in their amino acid sequences Finally, this L225A point mutation could also provide useful knock-in ani-mal models to study the role of the HIC1–CtBP interaction in tumorigenesis and in development
Abbreviations
AD, activation domain; CID, CtBP-interacting domain; CR, central region; CtBP, C-terminal binding protein; DBD, DNA-binding domain; EMT, epithelial-to-mesenchymal transition; HDAC, histone deacetylase; HPE, holoprosencephaly; KI, knock-in; MDS, Miller–Dieker syndrome; TSA, trichostatin A.
Trang 2progression and spectrum [5] Finally, a circular
regu-latory loop has been proposed for HIC1, SIRT1 and
p53, since HIC1 represses the transcription of SIRT1,
SIRT1 deacetylates P53, thus negatively modulating
its DNA-binding properties, and P53 transactivates
HIC1[6]
Besides its role in tumorigenesis, HIC1 is essential
during development and is also a candidate for a
conti-guous-gene syndrome, the Miller–Dieker syndrome
(MDS), a severe form of lissencephaly accompanied by
developmental anomalies [7] HIC1 is located within the
critical 350 kbp region deleted in most patients [8], and
together with perinatal death and a reduction in overall
size, Hic1– ⁄ –mouse embryos have many developmental
defects resembling those found in MDS patients [9] In
addition, parts of Hic1 expression territories in mice
embryos and in zebrafish strikingly overlap with regions
that exhibit abnormalities in MDS patients, e.g
cranio-facial and limb mesenchymes [10,11]
The HIC1 protein is a sequence-specific
transcrip-tional repressor containing three main functranscrip-tional
domains: a conserved protein–protein interaction
domain called BTB⁄ POZ at the N-terminus, five
Kru¨ppel-like C2H2zinc fingers near its C-terminus, and
a central region, which is not well-conserved among
the HIC1 and the paralogous HRG22 proteins from
various species (Fig 1) [12–14] HIC1 binds specifically
to DNA through its zinc finger domain, which
recogni-zes the recently defined consensus sequence
5¢-C⁄GNGC⁄GGGGCAC⁄ACC-3¢ [15] The BTB ⁄ POZ
domain is a conserved structural motif found mainly
in transcription factors, actin-binding proteins and
substrate-specific adapters of CUL-3-based ubiquitin ligases [16] The BTB⁄ POZ domain is essential for the function of transcriptional repressors by directly recruiting nuclear corepressor (SMRT, N-CoR or B-CoR)–histone deacetylase complexes (HDACs), as shown for the human PLZF and BCL6 proteins [17,18] Previously, we have shown that the HIC1 and HRG22 BTB⁄ POZ domains are autonomous trans-criptional repression domains, unable to recruit class I
or II HDACs, since they are insensitive to the specific inhibitor trichostatin A (TSA) [12,13] The HIC1 central region is also an autonomous transcriptional repression domain, which recruits the corepressor C-terminal binding protein 1 (CtBP1) [19,20] and represses transcription in a TSA-sensitive manner [14] Notably, HIC1 recruits CtBP1 through a short phylo-genetically conserved sequence, GLDLSKK, slightly divergent from the canonical CtBP-interacting domain (CID) containing a PxDLSxK⁄ R motif found in virtu-ally all CtBP-interacting proteins [11,14,20]
The vertebrate genomes contain two different genes, CtBP1 and CtBP2, widely expressed in normal human tissues and cancer cell lines as well as throughout development in mice [21] These genes encode at least four protein isoforms, including the corepressors CtBP1 and CtBP2 These two proteins have partially redundant functions For example, during murine development, Ctbp1 is poorly detectable in the extra-embryonic structures that express Ctbp2 Furthermore, mouse Ctbp2–⁄ – embryos die between E9 and E10.5 due to defects, notably in formation of the placenta and neural ectoderm [21] In contrast, Ctbp1–⁄ – mice are viable although they exhibit reduced fitness and fertility [21] CtBP1 is both cytoplasmic and nuc-lear, and this subcellular localization is regulated by interplay between post-transcriptional modifications (SUMOylation and phosphorylation) and binding to neural nitric oxide synthase, a PDZ domain-containing protein [22,23] In contrast, CtBP2 is mainly nuclear, due to a specific N-terminal 20 amino acid region containing Lys residues acetylated by P300 [24,24a] Finally, CtBP1 is SUMOylated by PIAS1 and PIASxb
on Lys428, which is not conserved in CtBP2 [23], whereas SUMOylation of CtBP2 on a nonconsensus targeting motif requires the presence of Pc2 as an E3 ligase [25]
In this article, we demonstrate that HIC1 can inter-act with both CtBP1 and CtBP2 This interinter-action relies
on a GxDLS motif that is slightly divergent from the PxDLS consensus motif found in most CtBP-interact-ing proteins Furthermore, point mutation of two CtBP2 residues forming part of the structure of the recognition cleft for a PxDLS motif also ablates the
PLCGLDLSKKSPPGSAAP PLCGLDLSKKSPPGSSVP PPCGLDLSKKSPTGPSAQ SVYGLDLSKKSPNSQSQL PNYGLDLSKKSPSPNSQT
*********
Hu HIC1
714 BTB/POZ
Hu HIC1
Mu HIC1
Ck FBPB
Zf HIC1b
Fu HIC1
Cons
1
Zinc Fingers
225
CID
Fig 1 Schematic drawing of the human HIC1 protein The
BTB ⁄ POZ domain and the five C 2 H 2 zinc fingers are represented as
dotted and grey boxes, respectively The evolutionarily conserved
CtBP interaction domain (CID) [14] is represented as a dotted line,
and its sequence in the various HIC1 proteins is shown The central
Leu (residue 225 in human HIC1), which is the sole invariant
resi-due in all the CID motifs described so far, is highlighted in bold In
the consensus lane (Cons), identical residues are indicated by *
under the aligned sequences Hu, Human; Mu, Murine; Ck,
Chicken; Zf, zebrafish (Danio rerio); Fu, Fugu rubipres.
Trang 3interaction with a GxDLS motif Conversely, mutation
of the central Leu residue (Leu225 in HIC1), which is
the sole invariant residue in all CtBP-interacting motifs
known so far, also abolishes the interaction between
HIC1 and CtBP1 or CtBP2 In close agreement with
the corepressor activity of CtBPs, this mutation
impairs the HIC1-mediated transcriptional repression
These results thus demonstrate a strong conservation
in the recognition of the CtBP-interacting motifs
despite great divergence in their amino acid sequences
Results
HIC1 can interact with both CtBP1 and CtBP2
in vivo
The vertebrate genomes contain two different genes,
CtBP1 and CtBP2, that encode two highly related
corepressors, CtBP1 and CtBP2, but with only
parti-ally redundant functions It is thus important to
deter-mine if a given transcription factor can interact with
both corepressors The interaction between HIC1 and
CtBP1 was first demonstrated through transient
trans-fection assays and in a stably tranfected HIC1 cell line
with inducible HIC1 expression [14] We have recently
shown that endogenous HIC1 proteins can be detected
in a human medulloblastoma cell line, DAOY [15] To
fully validate the interaction between endogenous
HIC1 and CtBPs, we performed
coimmunoprecipita-tion experiments using DAOY total cell extracts and
either normal rabbit immunoglobulins or the
anti-HIC1 antibody The immunoprecipitated proteins were
divided into two equal parts, separated by SDS⁄ PAGE
and then immunoblotted with monoclonal antibodies
to CtBP1 or CtBP2 CtBP1 was detected with the
spe-cific CtBP1 monoclonal antibodies in the anti-HIC1
immunoprecipitates (Fig 2A, top panel, lane 3) but
not in control IgG immunoprecipitates (lane 2) In
contrast, analyses of equal amounts from the same
immunoprecipitates with the CtBP2 monoclonal
anti-bodies yield only nonspecific bands, presumably due
to the quality of the antibodies used (Fig 2A, middle
panel, lanes 2 and 3) Coimmunoprecipitation of
DAOY nuclear extracts with the monoclonal or
another commercial polyclonal CtBP2 antibody
fol-lowed by western blotting with an anti-HIC1 antibody
did not give better results, probably because CtBP2 is
associated with numerous partners (data not shown)
As control, HIC1 is detected in the anti-HIC1
immu-noprecipitates but not in control IgG
immunoprecipi-tates (Fig 2A, bottom panel)
To confirm the interaction observed between
endog-enous HIC1 and CtBP1 (Fig 2A, top panel), we
per-formed another coimmunoprecipitation experiment using a similar amount of DAOY total cell extracts and another HIC1 antibody, the polyclonal 325 anti-body directed against a C-terminal peptide of HIC1 or normal rabbit immunoglobulins The total immunopre-cipitated proteins were separated by SDS⁄ PAGE After transfer, the membrane was separated into two parts The upper part (proteins above 60 kDa) and the lower part were directly immunoblotted, respectively, with the HIC1 antibody and with the CtBP1 monoclo-nal antibody CtBP1 was detected with the specific CtBP1 monoclonal antibodies in the anti-HIC1 immu-noprecipitates (Fig 2B, lane 6) but not in control IgG immunoprecipitates (lane 5) As a further control, pro-teins immunoprecipitated from DAOY total extracts in stringent conditions (RIPA buffer) with anti-CtBP1, normal rabbit immunoglobulins or HIC1 polyclonal antibodies were loaded on the same gel (Fig 2B, lanes 1–3)
To determine if HIC1 can also interact with CtBP2
in vivo, we thus performed coimmunoprecipitation assays in COS7 cells transiently transfected with expression vectors encoding a FLAG version of human HIC1 and CtBP1 or CtBP2, alone or in combination First, western blot analyses of total cell lysates with monoclonal antibodies directed against CtBP1 or CtBP2 demonstrated the specificity and the reactivity
of these antibodies (Fig 2C, top and middle panels; compare lanes 3 and 4) Immunoprecipitation of total cell extracts with the FLAG monoclonal antibody fol-lowed by immunoblot with the CtBP1 or CtBP2 monoclonal antibodies detected specific binding of HIC1 not only to CtBP1 as previously described (Fig 2C, lane 11) but also to CtBP2 (Fig 2C, lane 12)
in cotransfected cells
CtBP has been proposed to be a redox sensor link-ing gene expression and metabolism Indeed, CtBP binding to some, but not all, of its partners is potenti-ated by hypoxia or treatment with agents such as CoCl2, both of which increase the level of free NADH [26,27] We therefore investigated the effect of CoCl2 treatment on the interaction between HIC1 and CtBP1 COS7 cells were transfected with expression vectors for HIC1 and CtBP1 and treated with 200 lm CoCl2 for 3 h before lysis After immunoprecipitation with the HIC1 antibody, a significant increase in the amount of CtBP1 coimmunoprecipitated was observed
in the presence of CoCl2 (Fig 2D; compare lanes 6 and 8), indicating that the interaction between HIC1 and CtBP1 is favoured in the presence of higher NADH levels
Thus, HIC1 can interact in vivo with both the CtBP1 and CtBP2 corepressors
Trang 4The integrity of the substrate-binding domain
of CtBP2 is required for recognition of
the GLDLSKK motif
Crystal structure analyses of the substrate-binding
domain of CtBP1 have highlighted crucial residues
involved in the interaction with a PIDLSKK-like
peptide Notably, point mutations A41E or V55R, involving two residues that directly contact the PID-LSKK peptide, are sufficient to abolish the interaction between the E1A C-terminus and CtBP1 [28]
HIC1 is the first transcriptional factor known to recruit CtBP via a divergent GLDLSKK motif in which the Pro, which was considered to be an invariant
A
Input 2% IgG
WB: CtBP2
HIC1 WB: HIC1
(325)
B
C G
IP (RIPA Buffer)
Co-IP (IPH Buffer)
HIC1
CtBP1
WB: HIC1 (325)
WB: CtBP1 (E-12)
C
WB: CtBP1
WB: CtBP2
WB: HIC1 (325)
Input 5%
1 2 3 4 5 6 7 8 9 10 11 12
IP FLAG
WB: CtBP1
WB: FLAG
1 2 3 4 5 6 7 8
Fig 2 In vivo, HIC1 can interact with C-terminal binding protein 1 (CtBP1) or CtBP2 (A) Endogenous HIC1 and CtBP1 interact DAOY cells were lysed with IPH buffer, and 2% of the lysates were kept as input (lane 1) Equal amounts of lysate were then immunoprecipitated with normal rabbit immunoglobulins (lane 2: IgG) or the rabbit HIC1 antibodies (lane 3: HIC1 (2563)) The immunoprecipates were divided in two and analysed in parallel by immunoblot with CtBP1 (top panel) or CtBP2 (middle panel) monoclonal antibodies The anti-CtBP1 membrane was stripped and analysed with the HIC1 polyclonal antibodies as control (bottom panel) *Nonspecific band (B) Endogenous HIC1 and CtBP1 interact DAOY cells were lysed with RIPA buffer in stringent conditions, and lysates were immunoprecipitated with the rabbit CtBP antibodies (lane 1: aCtBP1(1128)) [38], with normal rabbit immunoglobulins (lane 2: IgG) or another rabbit HIC1 antibody (lane 3: aHIC1 (325)) The same amount of DAOY cells as used in (A) were lysed with IPH buffer, and 2% of the lysates were kept as input (lane 4) Equal amounts of lysate were then immunoprecipitated with normal rabbit immunoglobulins (lane 5: IgG) or the other rabbit HIC1 antibodies (lane 6: aHIC1 (325)) The immunoprecipates were loaded and analysed directly by immunoblot with the polyclonal HIC1 325 (top panel) or the monoclonal CtBP1 E-12 (bottom panel) antibodies (C) In vivo, HIC1 interacts with either CtBP1 or CtBP2 COS7 cells were transfected with the above indicated expression vectors Five per cent of each lysate was directly resolved by SDS ⁄ PAGE and immunoblotted with the indica-ted antibodies (input 5%, lanes 1–6) Each cell lysate was immunoprecipitaindica-ted with the FLAG M2 antibody (IP FLAG, lanes 7–12) The immu-noprecipitates were then analysed by western blotting with a CtBP1 monoclonal antibody (top panel) The membrane was then stripped and reprobed with a CtBP2 polyclonal antibody (middle panel) Finally, the presence of HIC1 in the immunoprecipitates was confirmed by immu-noblotting with the HIC1 polyclonal antibody (lower panel) (D) Treatment with CoCl 2 increases the interaction between HIC1 and CtBP1 COS7 cells were transfected with the indicated expression vectors Forty-eight hours after transfection, cells were either treated with
200 l M CoCl2or mock-treated for 3 h and lysed in IPH buffer Immunoprecipitates were analysed by western blotting as described above with CtBP1 or FLAG monoclonal antibodies.
Trang 5residue, is replaced by a Gly [14] The interaction
between HIC1 and CtBPs could thus involve residues
different from those involved in the interaction via a
consensus PIDLSKK peptide To address this question,
the equivalent point mutations were introduced by
site-directed mutagenesis into CtBP2 [29] to yield A58E
and V72R Like wild-type (wt) CtBP2, these mutants
are able to dimerize normally (data not shown) We
first used the yeast two-hybrid assay to investigate the
interaction between the central region (CR) of HIC1
(amino acids 135–422) and wt or mutant CtBP2 Yeast
cotransformed with the HIC1 CR fused to the Gal4
DNA-binding domain (DBD) and the CtBP2 point
mutants fused to the Gal4 activation domain (AD)
were unable to grow on His-selective medium (Fig 3A,
right panel), similar to those transfected with the empty
Gal4 vectors as negative control As expected, the
inter-action between the HIC1 CR and wt CtBP2 restores
the growth on selective medium
These results were confirmed in the context of the full-length proteins by coimmunoprecipitation analyses after transient transfection in COS7 cells In this assay,
a strong interaction is observed between wt HIC1 and CtBP2 (Fig 3B, lane 12), whereas the A58E and V72R point mutants fail to interact with HIC1 (Fig 3B, lanes 14 and 16)
Thus, binding of a CtBP-interacting partner contain-ing a PxDLS or a GxDLS motif is mediated by the same peptide recognition cleft in the CtBP N-terminal region
Point mutation of the only invariant residue
in the CtBP-interacting domain is sufficient to abolish the HIC1–CtBP interaction
Having established that the interaction between the HIC1 GxDLS motif and CtBP1 or CtBP2 occurs by binding to the same peptide cleft that binds typical
HIC1 CR
Gal4AD
Gal4AD CtBP2
Gal4AD CtBP2 A58E
Gal4AD CtBP2 V72R
Gal4DBD Gal4DBD
HIC1 CR
Gal4AD
Gal4AD CtBP2
Gal4AD CtBP2 A58E
Gal4AD CtBP2 V72R
B
*
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
WB:CtBP2 WB: HIC1
Fig 3 The GLDLSKK motif of HIC1
con-tacts the same residues in CtBP2 as the
consensus PIDLSKK motif (A) In yeast
two-hybrid assays, CtBP2 A58E and CtBP2
V72R fail to interact with the central region
of HIC1 The yeast two-hybrid assay was
used to assess the interaction between the
central region of HIC1 containing the
GLDLSKK motif and murine CtBP2 mutants
A58E and V72R Yeast cotransformed with
the plasmids shown grew on SD-Leu-Trp
media (left panel) These transformants
were patched onto SD-Leu-Trp-His selective
media (right panel) Growth after 48 h at
30 C is shown (B) Effect of the CtBP2
mutants A58E or V72R on HIC1–CtBP2
inte-raction in vivo COS7 cells were transfected
with the indicated expression vectors
Forty-eight hours after transfection, cells were
directly lysed in IPH buffer Five per cent of
each lysate was directly resolved by SDS ⁄
PAGE and immunoblotted with the indicated
antibodies to control for HIC1 and CtBP2
protein expression (input 5%, lanes 1–8).
Lysates were immunoprecipitated with
anti-HIC1 (lanes 9–16) and analysed by western
blotting with anti-CtBP2 (upper panel) and
with HIC1 polyclonal antibody (lower panel).
*Nonspecific band detected by anti-CtBP2
in each extract under the conditions used.
Trang 6PxDLS-containing partners, we next investigated the
role of the central Leu in the HIC1 GxDLS motif,
which is now the sole invariant residue in the CID
This Leu was thus mutated to Ala (mutation L225A)
to yield a GxDAS motif (Fig 4A) We first conducted
mammalian two-hybrid assays in rabbit kidney cells
(RK13) Chimeras between the Gal4 DBD and a
trun-cated CR of HIC1 (amino acids 135–296) [14]
contain-ing or not containcontain-ing the L225A point mutation were
tested for their ability to interact with full-length
CtBP1 fused to the VP16 AD A chimera between the
Gal4 DBD and the HIC1 BTB⁄ POZ domain was used
as a negative control (Fig 4A, lane 2) As previously
described [14], this truncated CR of HIC1 strongly
interacts with CtBP1 (Fig 4A, lane 3), whereas the
point mutation of the central Leu to Ala abolished this
interaction (Fig 4A, lane 4)
These results were obtained using the isolated CR of
HIC1 To confirm the essential role of the central
Leu225 in the context of the full-length protein, we
next performed coimmunoprecipitation experiments
As shown above, the full-length HIC1 protein can
interact with CtBP1 (Fig 4B, lane 4) or CtBP2
(Fig 4C, lane 5), whereas the L225A point mutant is
unable to interact with CtBP1 (Fig 4B, lane 6) or
CtBP2 (Fig 4C, lane 6)
In conclusion, mutation of Leu225 of the
GLDLSKK motif in the HIC1 CID domain into an
Ala (GLDASKK) is sufficient to abolish the
interac-tion between HIC1 and CtBP1 or CtBP2 in vivo
The HIC1 central region functions as a
CtBP-dependent and CtBP-independent
autonomous repression domain
We next investigated the importance of this L225A
point mutation for the transcriptional repression
prop-erties of the whole CR of HIC1 To that end, the
entire domain (amino acids 135–422) located between
the BTB⁄ POZ domain and the first zinc finger motif
was cloned in frame with the Gal4 DBD and tested in
the Gal4 repression assay Upon transient transfection
into RK13 cells, Gal4-HIC1 (135–422) efficiently
repressed the expression of a reporter gene containing
five Gal4-binding sites (Fig 5A, lane 2) Notably, the
L225A mutation, which abolishes the interaction with
CtBP1 and CtBP2 (Fig 4B,C), significantly reduced
but did not totally abolish the repression potential of
the CR (Fig 5A, lane 3)
CtBP is found in a large multiprotein complex
con-taining HDACs [30] When these experiments were
performed in the presence of TSA, a specific inhibitor
of class I and class II HDACs, the repression exhibited
both by the wt and the L225A chimeras was signifi-cantly reduced (Fig 5A)
A similar effect was observed, albeit to a lesser extent, when the repression of the wt and L225A full-length HIC1 proteins were tested on a Luc reporter gene driven by the recently defined HIC1-responsive element, 5xHiRE [15] (Fig 5B) This relatively mild effect on repression is probably explained by the pres-ence of the BTB⁄ POZ domain, which is another potent autonomous repression domain
Thus, the HIC1 CR represses transcription through CtBP-dependent and CtBP-independent mechanisms and involves the recruitment of class I or class II HDACs
Discussion
In this report, we demonstrate that the tumour sup-pressor gene and transcriptional resup-pressor HIC1 inter-acts with the related but not fully functionally redundant CtBP1 and CtBP2 corepressors in vivo and that this interaction is regulated by CoCl2, hence connecting the HIC1-mediated repression to NAD+⁄ NADH levels and hypoxia Moreover, mutation of the invariant Leu residue, L225A, is sufficient to disrupt the interaction between HIC1 and CtBPs
CtBPs have been previously shown to bind to repres-sion domains in a number of transcription factors and other regulatory proteins In the vast majority of CtBP partner proteins, a PxDLS motif has been shown to be the primary determinant of CtBP binding, which has led to the hypothesis that the PxDLS motif slots into
an ‘accepting’ pocket in CtBP However, some variant motifs have been reported, such as a bipartite motif located in the C-terminus of the viral EBNA3A pro-tein, where two nonconsensus sites, ALDLS and VLDLS, synergize to produce very efficient binding to CtBP [31] Other variants include a GLDLS motif in HIC1 [14] and a GLELE motif in ArpNa, a brain-spe-cific actin-related protein found in many chromatin-remodelling and histone acetyltransferase complexes [32] Using the HIC1 GxDLS motif as a paradigm for these nonconsensus motifs, we investigated the proper-ties of its binding to CtBPs In their elegant work on the crystal structure of rat CtBP⁄ BARS in a ternary complex with NADH and a PIDLSKK peptide, Nard-ini et al [28] identified the peptide-binding site as a hydrophobic surface cleft in the N-terminal part of the substrate-binding domain As this cleft is lined with hydrophobic residues that are fully conserved between CtBP1 and CtBP2, two different point mutants were constructed in the PIDLS-accepting pocket of CtBP2
In directed yeast two-hybrid assays or in transient
Trang 7tranfection assays in mammalian cells, these two point
mutants are unable to interact with HIC1 (Fig 3)
These results indicate that this cleft can accommodate
peptides with some variability in their amino acid sequences and could be considered to be a general binding site for (P⁄ G ⁄ V ⁄ A)xDLS-containing partners
A
VP16 VP16-CtBP1
296
296
DBD-Gal4 1
Fold activation
135 GLD L SKK
135 GLDASKK
POZ 2
3
4
WB: CtBP1
WB: FLAG
I I
PI
IP HIC1(2563)
1 2 3 4 5 6
+ CtBP1
CtBP2 FLAG-HIC1 FLAG-HIC1 L225A
IP: FLAG
WB: CtBP2
WB:HIC1 (325)
WB: CtBP2
WB: HIC1 (325)
Input 5%
C
WB: CtBP1
Input 5%
WB: FLAG
7 8 9
Fig 4 Mutation of the central invariant Leu225 to Ala in the GLDLSKK motif abolishes the interaction between HIC1 and C-terminal binding protein 1 (CtBP1) (A) In mammalian two-hybrid assays, HIC1 mutant L225A does not interact with CtBP1 Left panel: schematic structures
of the Gal4 DNA-binding domain (construct 1) and of the various Gal4-HIC1 chimeras (constructs 2–4) Numbering refers to HIC1 residues Black box, GLDLSKK motif; hatched box, mutated GLDASKK motif Right panel: Luc and b-galactosidase assays were performed on total extracts from RK13 cells that had been transiently transfected with 250 ng of the pG5-luc reporter (schematically drawn), 50 ng of the pSG5-lacZ construct as a control of transfection efficiency, 50 ng of the indicated Gal4 construct and 150 ng of the VP16 activation domain (grey bars) or VP16 activation domain-tagged murine CtBP1 (black bars) After normalization to b-galactosidase activity, the data were expressed as Luc activity relative to the activity of the pG5-luc with empty control vectors, which was given an arbitrary value of 1 Results presented are the mean values and standard deviations from two independent transfections in triplicate (B) HIC1 L225A mutation disrupts its interaction with murine CtBP1 COS7 cells were transfected with expression vectors for CtBP1 and the indicated FLAG construct Forty-eight hours after transfection, lysates were split into two and immunoprecipitated with rabbit preimmune serum (PI: lanes 1, 3 and 5) or the rabbit HIC1 2563 polyclonal antibody (I: lanes 2, 4 and 6) The resulting immunoprecipitates were analysed by western blotting with the indi-cated monoclonal antibodies (anti-CtBP1, upper panel; and anti-FLAG M2, lower panel) Five per cent of each total cell extract (input) was similarly analysed with CtBP1 and FLAG M2 antibodies to control for CtBP1 and HIC1 expression *Nonspecific band (C) HIC1 L225A muta-tion disrupts its interacmuta-tion with CtBP2 COS7 cells were transfected with expression vectors for CtBP2, FLAG-HIC1 and FLAG-HIC1 L225A alone or in combination as indicated and the indicated FLAG Input (lanes 7–12) and immunoprecipitates (lanes 1–6) were analysed as des-cribed above with the CtBP2 or HIC1 (325) antibodies.
Trang 8Mutagenesis of the first Pro, of the Pro-Leu or of
the Asp-Leu residues in the PLDLS CtBP-binding
motif from various proteins has been widely used to
severely impair their interaction with CtBP, at least in
some assays Based on the cocrystal structure [28],
binding of a PLDLS motif can be driven by docking
of the two Leu side chains into the conserved
hydro-phobic surface groove However, to the best of our
knowledge, our work provides the first demonstration
that a unique point mutation of the central Leu
resi-due, which is the sole invariant residue in CID motifs,
is sufficient to ablate the interaction between a
transcription factor and CtBPs Our results are thus not only in perfect agreement with the structural data but also provide experimental clues to the universal conservation of this Leu residue in all CtBP-binding motifs described so far
The HIC1 CR appears to be a second repression domain exhibiting both dependent and CtBP-independent repression mechanisms, both of which are sensitive to TSA Indeed, the L225A point mutation (Fig 5A) as well as the deletion of the GLDLS motif [14] impaired but did not fully abolish the transcrip-tional repression mechanisms of the CR The other corepressors and complexes interacting with the HIC1
CR are currently being investigated Along these lines,
we have recently identified a SUMOylation site at Lys314 that plays a role in the repression potential of this region (Stankovic-Valentin et al., unpublished results) In addition, a yeast two-hybrid screen per-formed with the HIC1 CR as bait identified not only full-length CtBP1 and CtBP2 as HIC1 partners but also some proteins clearly associated with HDAC-containing complexes (C Fleuriel and D Leprince, unpublished results)
Finally, we are currently generating conditional mouse knock-in (KI) mutants harbouring the critical Leu-to-Ala substitution at position 225, in the CtBP-interacting domain of HIC1 These animal models could be very useful for directly addressing in vivo the role played by the HIC1–CtBP interaction in two mutually nonexclusive pathways: the HIC1 tumour suppressor properties, which are altered in many human cancers, and normal development
CtBP appears to be a potential modulator of apop-tosis and epithelial-to-mesenchymal transition (EMT),
an important feature of embryonic development and tumorigenesis [33] The heterozygous Hic1+⁄ –mice are viable, and have no obvious developmental defects, but spontaneously develop tumours late in their life with a predominance of sarcomas and lymphomas in females and of carcinomas in males [2] It would thus
be interesting to determine if heterozygous mice carry-ing the L225A point mutation in the CtBP-bindcarry-ing domain, Hic1+⁄ KI, will also develop tumours, especi-ally the males
More importantly, the homozygous Hic1–⁄ – mice are embryonic lethal and display severe developmental anomalies, some of which are found in MDS patients [9] In Drosophila, CtBP is a corepressor for several short-range repressors essential for early embryonic development, such as Kru¨ppel, Giant, Knirps and Snail [34] In vertebrates, Ctbp1- or Ctbp2-deficient embryos exhibit a large variety of developmental defects, consis-tent with the fact that they interact with a multitude of
DBD-Gal4
1
GLDLSKK 422
135
2
GLDASKK 422
135
DMSO TSA
3
Fold repression
FLAG
FLAG-HIC1
FLAG-HIC1 L225A
Fold repression
Fig 5 Mutation L225A in the CtBP-binding motif impaired the
HIC1-mediated transcriptional repression (A) The HIC1 central
region contains C-terminal binding protein (CtBP)-dependent and
CtBP-independent repression activities which are both trichostatin
A (TSA)-sensitive Left panel: schematic structure of the Gal4
DNA-binding domain and the two Gal4-HIC1 chimeras Numbering refers
to HIC1 residues Black box, GLDLSKK motif; hatched box,
mutated GLDASKK motif Right panel: RK13 cells were transiently
transfected in triplicate with 100 ng of the indicated constructs,
350 ng of the pG5-Luc reporter and 50 ng of the pSG5-LacZ vector.
The cells were treated 24 h later with 300 n M TSA (dissolved in
dimethyl sulfoxide (DMSO)) (white boxes) or mock-treated with an
equal volume of DMSO (black boxes) for a further 24 h before
har-vesting Luc and b-galactosidase assays were conducted as
des-cribed in Fig 4A (B) The L225A mutation affects the transcriptional
repression potential of full-length HIC1 U2OS cells were
transfect-ed with 225 ng of the indicattransfect-ed expression vector, 225 ng of the
reporter 5xHiRE-luc plasmid and 50 ng of the pSG5-lacZ construct
as control of transfection efficiency Luc and b-galactosidase assays
were done as described above.
Trang 9transcription factors involved in many signalling
path-ways [21] For all these reasons, it is tempting to
specu-late that the Hic1 KI⁄ KI embryos could phenocopy, at
least in part, the abnormalities found in the Hic1–⁄ –
mice For example, GATA2 KI⁄ KI mice carrying a
Val-to-Gly point mutation in the GATA2 N-terminal
zinc finger that ablates the interaction with cofactors of
the Friend of GATA (FOG) family have been
gener-ated These mutant mice display complete
megakaryo-poietic failure, a phenocopy of Fog1–⁄ –mice [35]
Focusing on CtBP, the best example in favour of
our working hypothesis is holoprosencephaly (HPE),
the most common structural defect of the developing
forebrain in humans, frequently accompanied by
cra-niofacial anomalies HPE4, one of the loci associated
with this disease, maps to TGIF, a gene encoding a
homeodomain transcriptional repressor modulating
NODAL (a member of the transforming growth
factor-b family) signalling Individuals with HPE carry
heterozygous mutations either in the DNA-binding
domain, the SMAD-binding domain, or a
CtBP-inter-action motif [36] This latter S28C mutation converting
a canonical PLDLS motif into a PLDLC motif is
suffi-cient to disrupt the interaction between TGIF and
CtBP [37], as does the L225A in HIC1
In summary, we have shown that HIC1 is able to
interact with CtBP1 and CtBP2 We have
demonstra-ted that point mutation of the central Leu225 in the
CtBP-interacting motif is sufficient to abolish the
inter-action with CtBP, which is in perfect agreement with
the structural data and the universal conservation of
this residue in all CID motifs This point mutation
could also provide useful KI animal models to study
the role of the HIC1–CtBP interaction in
tumorigen-esis as well as in development
Experimental procedures
Constructs
The full-length FLAG-HIC1 L225A point mutant was
gen-erated by the two-round PCR mutagenesis strategy using
the following two mutagenic oligonucleotides, which
introduced an Nsi1 restriction site (underlined) (5¢-CGG
CGGGCTCTTCTTGGATGCATCCAGGCC-3¢ and 5¢-GG
convenient flanking oligonucleotides The StuI–XhoI
frag-ment containing the L225A mutation was exchanged with
the same restriction fragment in the previously described
FLAG-HIC1 wt expression vector [14]
The Gal4-HIC1 135–296 wt and L225A were generated
by PCR using the wt or L225A full-length clones as
matri-ces and the following oligonucleotides: sense 5¢-GGAATT
CGGGATCCCAAAGTACTGCCACCTGCGG-3¢ with a BamH1 site, and antisense 5¢-AGTGGTACCGTCGACTC ATCCCGGGCTGCCGCT-3¢, with a KpnI site The Gal4-HIC1 135–422 wt and L225A were generated by PCR as described above with the same sense oligonucleotide and
an antisense oligonucleotide containing an SstI site 5¢-AT GCACACACGTAAGGCACTCAGCTGAGATCTCGAG-3¢ The BamHI–KpnI and BamHI–SstI restriction fragments were cloned in frame with the Gal4 DBD in the pSG5424 vector A58E and V72R mutations were introduced into mCtBP2 by overlap PCR mutagenesis (mCtBP2.A58E.F, GACCTGGCCACTGTGGAATTCTGTGATGCACAG; mCtBP2.A58E.R, CTGTGCATCACAGAATTCCACAGT GGCCAGGTC; mCtBP2.V72R.F, GAAATCCATGAGA AGCGGTTGAATGAAGCTGTG; and mCtBP2.V72R.R, CACAGCTTCATTCAACCGCTTCTCATGGATTTC) BglII- and SalI-digested mutant inserts were ligated into the BamHI–SalI sites of pGAD10 vector to generate Gal4AD-CtBP2-A58E and Gal4AD-CtBP2-V72R Second,
wt CtBP2, CtBP2-A58E and CtBP2-V72R mutant inserts were reamplified by PCR using appropriate primers and cloned into the NotI–SalI sites of pMT3 (derived from pMT2)
PCR fragments were systematically verified by sequen-cing on both strands All clones were checked by appropri-ate restriction enzyme digestion, and the vector–insert junctions were verified by sequencing
Yeast two-hybrid system
The yeast two-hybrid system was used as described in the manufacturer’s protocol (Clontech, Saint Quentin en Yve-lines, France) Briefly, the CR of HIC1 [14] was cloned in-frame into the Gal4 DBD plasmid pGBT9 These plasmids were cotransfected with the Gal4 AD plasmid pGAD10 fused to mCtBP2, A58E and V72R mutants into the yeast strain HF7c, and transformants were selected on Trp⁄ Leu-deficient media (SD-Leu-Trp) Colonies were patched onto Trp⁄ Leu ⁄ His-deficient media (SD-Leu-Trp-His)
Cell culture and transfection
COS7, DAOY, U2OS and RK13 cells were maintained in Dulbecco medium supplemented with 10% fetal bovine serum
Cells were transfected in OptiMEM (Gibco, Paisley, UK)
by the PEI (Euromedex, Souffelweyersheim, France) method
as previously described [14] in either 100 mm diameter dishes (in vivo interaction in COS7 cells) with 2.5 lg of DNA or in 12-well plates (repression and mammalian two-hybrid assays) with 500 ng of DNA for RK13 cells and for U2OS cells Cells were transfected for 6 h and were then incubated in fresh complete medium They were rinsed in cold NaCl⁄ Pi 48 h after transfection and processed for
Trang 10coimmunoprecipitation, repression or mammalian
two-hybrid assays as described below
Immunoprecipitation and coimmunoprecipitation
Total DAOY cell extracts were immunoprecipitated in
stringent conditions using RIPA buffer as previously
des-cribed [14]
For coimmunoprecipitation experiments, DAOY cells or
COS7 cells (48 h after transfection) were rinsed twice in
cold NaCl⁄ Pi and lysed in cold IPH buffer (50 mm Tris
pH 8, 150 mm NaCl, 5 mm EDTA, 0.5% NP-40, protease
inhibitor cocktail (Roche)) Cell lysates were cleared by
cen-trifugation (20 000 g, 30 min) The supernatants were
incu-bated overnight with 4 lL of antibody Then, protein A
Sepharose beads (Amersham Biosciences, Orsay, France)
were added for 1 h The beads were washed three times
with IPH buffer Proteins were eluted by boiling in
Lae-mmli loading buffer and separated by SDS⁄ PAGE before
western blotting For experiments using CoCl2, 3 h before
lysis, cells were either treated with 200 lm CoCl2diluted in
fresh medium or mock-treated with fresh medium
Repression and mammalian two-hybrid assays
Forty-eight hours after transfection, cells were rinsed in
NaCl⁄ Piand lysed with the Luc assay buffer (25 mm
glycyl-glycine, pH 7.8; 15 mm MgSO4; 4 mm EGTA; 1% Triton
X-100) Luciferase and b-galactosidase activities were
meas-ured by using, respectively, beetle luciferin (Promega,
Char-bonnieres, France) and the Galacto-light kit (Tropix,
Bedford, MA, USA) with a Berthold (Thoiry, France)
chemioluminometer After normalization to b-galactosidase
activity, the data were expressed as Luc activity relative to
the activity of pG5-Luc with empty control vector, which
was given an arbitrary value of 1 Results represent the
mean values and standard deviations from two independent
transfections in triplicate [12]
Western blot and antibodies
Western blots were performed essentially as previously
des-cribed [14] The HIC1 2563 and HIC1 325 antibodies have
been previously described [14] For CtBP1, the CtBP1
monoclonal antibody (E-12) raised against amino acids 1–
440 of human CtBP1 (Santa Cruz Biotechnology, Le perray
en Yvelines, France) was used To detect endogenous
CtBP2, we used two different antibodies: a monoclonal
antibody raised against amino acids 361–445 of mouse
CtBP2 (BD Biosciences, Pharmingen, Le Pont de Claix,
France) or a goat polyclonal antibody raised against a
pep-tide near the C-terminus of CtBP2 of human origin
(sc-5966; Santa Cruz Biotechnology) In the
immunoprecip-itation experiment, we used the CtBP1 rabbit polyclonal
1128 antibodies raised against the ovalbumin-coupled pep-tide corresponding to amino acids 352–374 of murine CtBP1 [38] and kindly provided by B Wasylyk Anti-Flag M2 is a monoclonal antibody (F3165; Sigma, Lyon, France) The secondary antibodies were horseradish peroxi-dase-linked antibodies raised against either rabbit or mouse immunoglobulins (Amersham Biosciences)
Acknowledgements
We thank Drs Se´bastien Pinte and Brian R Rood for critically reading the manuscript This work was sup-ported by funds from CNRS, the Pasteur Institute, the Ligue Nationale contre le Cancer (Comite´ Interre´gio-nal du Septentrion), the Association pour la Recherche contre le Cancer (ARC) and a grant from the Austra-lian National Health and Medical Research Council
N Stankovic-Valentin was supported by a fellowship from the Ministe`re de la Recherche et de la Technolo-gie and by the Association pour la Recherche contre le Cancer (ARC)
References
1 Wales MM, Biel MA, el Deiry W, Nelkin BD, Issa
JP, Cavenee WK, Kuerbitz SJ & Baylin SB (1995) p53 activates expression of HIC-1, a new candidate tumour suppressor gene on 17p13.3 Nat Med 1, 570–577
2 Chen WY, Zeng X, Carter MG, Morrell CN, Chiu-Yen RW, Esteller M, Watkins DN, Herman JG, Mankowski JL & Baylin SB (2003) Heterozygous disruption of Hic1 predisposes mice to a gender-dependent spectrum of malignant tumors Nat Genet
33, 197–202
3 Guerardel C, Deltour S, Pinte S, Monte D, Begue A, Godwin AK & Leprince D (2001) Identification in the human candidate tumor suppressor gene HIC-1 of a new major alternative TATA-less promoter positively regulated by p53 J Biol Chem 276, 3078–3089
4 Britschgi C, Rizzi M, Grob TJ, Tschan MP, Hugli B, Reddy VA, Andres AC, Torbett BE, Tobler A & Fey
MF (2005) Identification of the p53 family-responsive element in the promoter region of the tumor suppressor gene hypermethylated in cancer 1 Oncogene 25, 2030– 2039
5 Chen W, Cooper TK, Zahnow CA, Overholtzer M, Zhao Z, Ladayni M, Karp JE, Gokgoz N, Wunder JS, Andrulis IL et al (2004) Epigenetic and genetic loss of Hic1 function accentuates the role of p53 in tumorigen-esis Cancer Cell 6, 387–398
6 Chen WY, Wang DH, Yen RC, Luo J, Gu W & Baylin
SB (2005) Tumor suppressor HIC1 directly regulates