p53-targeted LSD1 functions in repression of chromatin structure and transcription in vivo.
ABSTRACT Despite years of study focused on the tumor suppressor p53, little is understood about its functions in normal, differentiated cells. We found that p53 directly interacts with lysine-specific demethylase 1 (LSD1) to alter chromatin structure and confer developmental repression of the tumor marker alpha-fetoprotein (AFP). Chromatin immunoprecipitation (ChIP) and sequential ChIP of developmentally staged liver showed that p53 and LSD1 cooccupy a p53 response element, concomitant with dimethylated histone H3 lysine 4 (H3K4me2) demethylation and postnatal repression of AFP transcription. In p53-null mice, LSD1 binding is depleted, H3K4me2 is increased, and H3K9me2 remains unchanged compared to those of the wild type, underscoring the specificity of p53-LSD1 complexes in demethylation of H3K4me2. We performed partial hepatectomy of wild-type mouse liver and induced a regenerative response, which led to a loss of p53, increased H3K4me2, and decreased LSD1 interaction at AFP chromatin, in parallel with reactivation of AFP expression. In contrast, nuclear translocation of p53 in mouse embryonic fibroblasts led to p53 interaction with p21/CIP1 chromatin, without recruitment of LSD1, and to activation of p21/CIP1. These findings reveal that LSD1 is targeted to chromatin by p53, likely in a gene-specific manner, and define a molecular mechanism by which p53 mediates transcription repression in vivo during differentiation.
- [show abstract] [hide abstract]
ABSTRACT: Precise control of transcriptional programmes underlying metazoan development is modulated by enzymatically active co-regulatory complexes, coupled with epigenetic strategies. One thing that remains unclear is how specific members of histone modification enzyme families, such as histone methyltransferases and demethylases, are used in vivo to simultaneously orchestrate distinct developmental gene activation and repression programmes. Here, we report that the histone lysine demethylase, LSD1—a component of the CoREST-CtBP co-repressor complex—is required for late cell-lineage determination and differentiation during pituitary organogenesis. LSD1 seems to act primarily on target gene activation programmes, as well as in gene repression programmes, on the basis of recruitment of distinct LSD1-containing co-activator or co-repressor complexes. LSD1-dependent gene repression programmes can be extended late in development with the induced expression of ZEB1, a Krüppel-like repressor that can act as a molecular beacon for recruitment of the LSD1-containing CoREST-CtBP co-repressor complex, causing repression of an additional cohort of genes, such as Gh, which previously required LSD1 for activation. These findings suggest that temporal patterns of expression of specific components of LSD1 complexes modulate gene regulatory programmes in many mammalian organs.Nature 03/2007; 446(7138):882-887. · 38.60 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: The tumour suppressor p53 becomes activated in response to upstream stress signals, such as DNA damage, and causes cell-cycle arrest or apoptosis. Here we report a novel role for p53 in the differentiation of mouse embryonic stem cells (ESCs). p53 binds to the promoter of Nanog, a gene required for ESC self-renewal, and suppresses Nanog expression after DNA damage. The rapid down-regulation of Nanog mRNA during ESC differentiation correlates with the induction of p53 transcriptional activity and Ser 315 phosphorylation. The importance of Ser 315 phosphorylation was revealed by the finding that induction of p53 activity is impaired in p53(S315A) knock-in ESCs during differentiation, leading to inefficient suppression of Nanog expression. The decreased inhibition of Nanog expression in p53(S315A) ESCs during differentiation is due to an impaired recruitment of the co-repressor mSin3a to the Nanog promoter. These findings indicate an alternative mechanism for p53 to maintain genetic stability in ESCs, by inducing the differentiation of ESCs into other cell types that undergo efficient p53-dependent cell-cycle arrest and apoptosis.Nature Cell Biology 03/2005; 7(2):165-71. · 20.76 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: There is growing evidence that the p53 tumor suppressor protein not only can function to activate gene transcription but also to repress the expression of specific genes. Although recent studies have implicated the transcriptional repression function of p53 in the pathway of apoptosis, the molecular basis of this activity remains poorly understood. This study takes a first step toward elucidating this mechanism. We report that trichostatin A (TSA), an inhibitor of histone deacetylases (HDACs), abrogates the ability of p53 to repress the transcription of two genes that it negatively regulates, Map4 and stathmin. Consistent with this finding, we report that p53 physically associates in vivo with HDACs. This interaction is not direct but, rather, is mediated by the corepressor mSin3a. Both wild-type p53 and mSin3a, but not mutant p53, can be found bound to the Map4 promoter at times when this promoter preferentially associates with deacetylated histones in vivo. Significantly, inhibition of p53-mediated transcriptional repression with TSA markedly inhibits apoptosis induction by p53. These data offer the first mechanistic insights for p53-mediated transcriptional repression and underscore the importance of this activity for apoptosis induction by this protein.Genes & Development 11/1999; 13(19):2490-501. · 12.44 Impact Factor
MOLECULAR AND CELLULAR BIOLOGY, Sept. 2008, p. 5139–5146
Copyright © 2008, American Society for Microbiology. All Rights Reserved.
Vol. 28, No. 17
p53-Targeted LSD1 Functions in Repression of Chromatin Structure
and Transcription In Vivo?†
Wen-Wei Tsai,1Thi T. Nguyen,1Yang Shi,2and Michelle Craig Barton1*
Department of Biochemistry and Molecular Biology, Program in Genes and Development, Graduate School of Biomedical Sciences,
University of Texas M. D. Anderson Cancer Center, Houston, Texas 77030,1and Department of Pathology,
Harvard Medical School, 77 Avenue Louis Pasteur, Boston, Massachusetts 021152
Received 20 February 2008/Returned for modification 10 April 2008/Accepted 10 June 2008
Despite years of study focused on the tumor suppressor p53, little is understood about its functions in
normal, differentiated cells. We found that p53 directly interacts with lysine-specific demethylase 1 (LSD1) to
alter chromatin structure and confer developmental repression of the tumor marker alpha-fetoprotein (AFP).
Chromatin immunoprecipitation (ChIP) and sequential ChIP of developmentally staged liver showed that p53
and LSD1 cooccupy a p53 response element, concomitant with dimethylated histone H3 lysine 4 (H3K4me2)
demethylation and postnatal repression of AFP transcription. In p53-null mice, LSD1 binding is depleted,
H3K4me2 is increased, and H3K9me2 remains unchanged compared to those of the wild type, underscoring the
specificity of p53-LSD1 complexes in demethylation of H3K4me2. We performed partial hepatectomy of
wild-type mouse liver and induced a regenerative response, which led to a loss of p53, increased H3K4me2, and
decreased LSD1 interaction at AFP chromatin, in parallel with reactivation of AFP expression. In contrast,
nuclear translocation of p53 in mouse embryonic fibroblasts led to p53 interaction with p21/CIP1 chromatin,
without recruitment of LSD1, and to activation of p21/CIP1. These findings reveal that LSD1 is targeted to
chromatin by p53, likely in a gene-specific manner, and define a molecular mechanism by which p53 mediates
transcription repression in vivo during differentiation.
Numerous analyses of p53 functions, in response to cellular
stress, underscore its important, pleiotropic roles in arrest of
the cell cycle and promotion of apoptosis (5, 24, 33). Less
understood are mechanisms of p53-mediated regulation during
differentiation and in maintenance of cellular homeostasis.
Previous work from our laboratory established p53 as a major
repressor of AFP transcription during hepatic development
(60). Developmental repression of AFP, an onco-fetal tumor
marker gene, is delayed more than 2 months after birth in
p53-null mice, compared to cessation within 2 to 3 weeks in
wild-type (WT) mice (45). Transactivating p73 (TA-p73), but
not TA-p63, partially compensates for an absence of p53 in
p53-null mice, characterized by elevated methylation of his-
tone H3 lysine 4 (H3K4) at the p53/p73 response element of
alpha-fetoprotein (AFP) chromatin (12). Therefore, we hy-
pothesized that an H3K4 demethylase is recruited by p53 dur-
ing hepatic differentiation and that combinatorial alterations in
chromatin structure during development dictate the develop-
mental timing of transcription regulation.
Enzymatic removal of methyl groups from lysine residues of
histone N-terminal tails correlates with either repression or
activation of transcription, as determined by the specific his-
tone substrate, amino acid position, and level of methylation.
Loss of methylated H3K4 (H3K4me) is generally associated
with repression or cessation of active transcription (7, 14, 55).
The first enzyme with the capacity to remove methyl groups
from histone lysine residues was identified as BHC110 or his-
tone lysine-specific demethylase 1 (LSD1; recently renamed
KDM1) (42), a corepressor of nerve-specific genes in nonneu-
ral tissues, where LSD1 is targeted by the repressor proteins
REST and Co-REST (50, 51). LSD1 is restricted to demethyl-
ation of di- and monomethylated substrates due to chemical
limitation of the flavin-dependent oxidation reaction (50, 51).
Target specificity of LSD1 is determined by interacting pro-
teins, e.g., interactions with nuclear receptors to demethylate
H3K9me2 and effect activation (21, 39, 59, 61) or interactions
with REST/Co-REST to demethylate H3K4me2 and repress
transcription (50, 51).
Similar to the case for modifications of histone proteins, the
specific types, extent, and sites of posttranslational modifica-
tions of p53 offer a “combinatorial code” and determine p53
interactions with protein partners as effectors of downstream
regulation and p53 protein stability (2, 24). Additionally, the
consequences of methylation are degree and residue specific
for p53, as with histones. Methylation of p53 alters its ability to
regulate transcription and is either repressive, e.g., mono-
methylation at K370 by Smyd2 (28) or at K382 by SET8 (49),
or activating, e.g., dimethylation at K372 by SET9 (9) or at
K370 by an unidentified methyltransferase (29). Recently,
Berger and colleagues uncovered LSD1 as a demethylase of a
nonhistone substrate, i.e., dimethylated p53 (29). They found
that LSD1-mediated demethylation of p53K370me2 prevents
p53 interaction with 53BP1 as a coactivator, blocking p53 bind-
ing to DNA and activation of p21/CIP1 and MDM2 in cultured
cells. In this manner, LSD1 functions as a repressor of p53-
activated gene expression in the absence of stress stimuli and
does so independently of its ability to modify chromatin.
* Corresponding author. Mailing address: Dept. of Biochemistry
and Molecular Biology, University of Texas M. D. Anderson Cancer
Center, 1515 Holcombe Blvd., Box 1000, Houston, TX 77030. Phone:
(713) 834-6268. Fax: (713) 834-6273. E-mail: firstname.lastname@example.org.
† Supplemental material for this article may be found at http://mcb
?Published ahead of print on 23 June 2008.
In the current study, we found that LSD1 is recruited by p53
to demethylate H3K4me2 of active chromatin and to repress
transcription of AFP during liver development. Interaction
with p53 is direct and independent of the enzymatic activity of
LSD1. The binding of p53 and recruitment of LSD1 to chro-
matin are gene specific and reversible, concomitant with deg-
radation of p53 and reactivation of AFP expression during liver
regeneration. These results reveal a role for LSD1 complexes
in developmental and tissue-specific regulation of nonneural
genes, a role that is mediated by p53 in normal cells in vivo.
MATERIALS AND METHODS
ChIP and RNA analyses of solid tissue and cultured cells. Liver tissue was
isolated and flash-frozen from WT and p53?/?mice of a mixed C57BL/6J:129sv
background. Chromatin immunoprecipitation (ChIP) and re-ChIP assays of liver
tissue or cultured cells were performed as described previously (12, 45), with
minimal modifications. The fragmented, precleared chromatin lysate was
incubated overnight with the following specific antibodies for ChIP: antibod-
ies to histone H3 (Abcam), H3K4me2 (Upstate/Millipore), H3K4me3 (Ab-
cam), H3K4me1 (Abcam), H3K9me2 (Upstate/Millipore and Abcam), acety-
lated H3K9 (H3K9ac) (Upstate/Millipore), H4ac (Upstate/Millipore), p53
(Oncogene and Novocastra), p73 (Santa Cruz), LSD1 (Abcam), Foxa1 (Abcam),
and normal sheep immunoglobulin G (IgG) (Upstate/Millipore). The specificity
of the LSD1 antibody was confirmed by immunoblotting of mouse liver nuclear
extract (see Fig. S1 in the supplemental material).
To analyze specific and antibody- and protein-bound DNAs, we performed
conventional PCR, real-time quantitative PCR (qPCR), and quantitative Taq-
Man (Applied Biosystems, ABI, Foster City, CA) real-time PCR. Conventional
PCR primers were generated to detect the AFP Smad binding element/p53
response element (SBE/p53RE) and calbindin RE-1 regions as described previ-
ously (4, 45). The following TaqMan real-time PCR primers and probe were used
for the AFP SBE/p53RE region: forward primer, 5?-CTACATATGAAGCCTT
AGCAAACATGT-3?; reverse primer, 5?-ACTCAGACGTTGGCGTGTCA-3?;
and probe, 6-carboxyfluorescein (FAM)-CCTCTAGACACACAGACT-MGB.
The following real-time PCR primers were used to detect the p21 gene 5? p53
binding element: forward primer, 5?-CCTTTCTATCAGCCCCAGAGGATAC
C-3?; and reverse primer, 5?-GACCCCAAAATGACAAAGTGACAA-3?. Prim-
ers and reverse transcription-PCR (RT-PCR) determinations of RNA expression
were performed as previously described (45). The following primers were used to
detect expression of the LSD1 gene: forward primer, 5?-GGAATCCCATGGC
TGTCGTCA-3?; and reverse primer, 5?-GATATCTCTGGGCGGCTTCACTT-
3?. qPCRs were conducted in a model 7500 FAST ABI instrument.
Liver regeneration. Partial hepatectomy (PH) to remove 60% to 70% of total
liver tissue or control, sham surgery of three to five mice for each experimental
condition was performed by IACUC-approved procedures, as previously de-
scribed (20). Mice were sacrificed 24 h or 7 days following PH; remnant liver
tissue was harvested, flash-frozen, and processed for RNA and ChIP analyses.
Cell culture and coimmunoprecipitation (co-IP) assays. Hepa1-6 murine hep-
atoma cells, HEK293 human embryonic kidney cells, and U2OS human osteo-
sarcoma cells were obtained from ATCC and cultured under suggested condi-
tions. AML12 cells were obtained from J. Clark and N. Fausto (University of
Washington, Seattle). Val5 mouse embryonic fibroblasts (MEFs) were obtained
from M. Murphy (Fox Chase Cancer Center, Philadelphia, PA) (44). Plasmids
encoding human LSD1, LSD1?C, and TA-p73? have been described (15, 50, 51).
HEK293 cells were transfected with 2 ?g of total DNA in six-well plates by
standard Ca(PO4) methodology (22). U2OS cells were transfected with 1 ?g of
total DNA in six-well plates, using Effectene (Qiagen) as recommended by the
Transfected or control HEK293 cells, U2OS cells, and AML12 cells were lysed
in NTEP buffer (150 mM NaCl, 25 mM Tris-HCl, pH 7.5, 5 mM EDTA, 0.5%
NP-40 plus freshly added 1? protease inhibitor cocktail set I [Calbiochem] with
1 mM phenylmethylsulfonyl fluoride [Sigma]) and sonicated for a few seconds.
The cell lysate was incubated with 100 ?g/ml ethidium bromide for 1 h and
precleared by incubation with 20 ?l protein G beads (50% slurry; Sigma) for 1 h.
The precleared lysate was incubated overnight with anti-Flag M2 beads (Sigma)
or the following specific antibodies: anti-p53 (Santa Cruz, Oncogene, and No-
vocastra), anti-LSD1 (Abcam), anti-rabbit IgG (Upstate/Millipore), anti-RBP2
(Bethyl Laboratories), and anti-Flag M2 (Sigma). Next, 25 ?l protein G beads
(50% slurry) was added and incubated for 2 h at 4°C. Protein-bound beads were
recovered by centrifugation and washed three times with NTEP, first with 500
mM NaCl, then with 0.5% sodium dodecyl sulfate, and then with NTEP buffer
alone. Input lysate, equivalent to 1/20 of the immunoprecipitation lysate, was
analyzed alongside bead-bound proteins by sodium dodecyl sulfate-polyacryl-
amide electrophoresis and immunoblot analyses, as previously described (60).
The primary antibodies for immunoblotting were as follows: anti-Co-REST (Up-
state), anti-RB (Santa Cruz), antihemagglutinin (Roche), and anti-Flag M2
RNA interference assays. Small interfering RNA (siRNA) oligonucleotide
pools targeting murine LSD1 and nonspecific and nontarget siRNAs were pur-
chased from Dharmacon/Thermo Fisher Scientific (Chicago, IL). Transient
transfection of siRNAs into AML12 cells was carried out using Lipofectamine
2000 (Invitrogen) as recommended by the manufacturer. RNAs were isolated
48 h after transfection, and protein lysates were prepared 72 h after transfection.
Primers and RT-PCR determinations of RNA expression were as previously
described (45). qPCRs were conducted in a model 7500 FAST ABI instrument.
Statistical analyses. GraphPad Prism5 software (GraphPad Software, Inc.)
was used for analysis of P values based on at least two independent experiments
with three independent PCRs. The two-tailed paired t test was used to compare
the differences in relative changes between two groups (see Fig. 1B, 3B, 4A, 5D,
and 6A). The two-tailed unpaired t test was used to compare the differences in
actual percentages between two groups (see Fig. 3A and 6B). P values of ?0.05
were considered statistically significant.
Developmental repression of AFP is marked by p53 and
LSD1 interaction with chromatin. We performed ChIP anal-
yses of developmentally staged mouse liver tissue to determine
relative levels of endogenous p53, LSD1, and H3K4me2
present at the developmental repressor region (SBE/p53RE)
(60) of AFP chromatin. Our previous work showed that this
intercalated Smad/p53 response element is essential for p53-
mediated repression of AFP in cultured cells and by in vitro
transcription (11, 34). Shortly after birth, when hepatic AFP
expression is robust (Fig. 1A, lane 8d), H3K4me2 levels are
high at the SBE/p53RE repressor region, but p53 and LSD1
are barely detectable (Fig. 1B). Quantitative real-time PCR
analyses of multiple ChIP assays showed that a 10-fold de-
crease in H3K4me2 level occurs alongside a 4- to 6-fold in-
crease in p53 and LSD1 binding at 2 months of age (Fig. 1B),
when AFP expression is completely repressed (Fig. 1A). Ab-
solute levels of bound DNA amplified by nonquantitative PCR
and the lack of nonspecific DNA associated with the IgG
control are shown compared to undiluted input DNA (Fig. 1B,
bottom panel). Overexpression of Flag-tagged p53 or LSD1 in
U2OS cells, which express endogenous p53 (1), does not cause
a generalized decrease in cellular H3K4me2 levels (see Fig.
S2A and B in the supplemental material).
p53 interacts with LSD1 to mediate AFP repression. To
determine if p53 and LSD1 simultaneously bind to the same
DNA element in vivo, we performed sequential ChIP or re-
ChIP experiments to analyze 2-month-old mouse liver tissue.
The results show that p53 and LSD1 cooccupy chromatin at
this stage of development, as revealed by sequential ChIP with
an antibody recognizing mouse p53 protein, followed by pre-
cipitation of p53-bound chromatin with an LSD1-specific an-
tibody. These antibodies differ in their relative affinities for
protein-bound chromatin, so elution is not likely quantitative;
however, reversal of the order of antibodies in re-ChIP yielded
equivalent results, with no nonspecific background binding to
IgG (Fig. 1C). These results suggest that p53 and LSD1 act
together to mediate repression of AFP during development.
We used co-IP analysis to determine if LSD1 and p53 inter-
act and associate as a soluble protein complex. We found a
5140TSAI ET AL.MOL. CELL. BIOL.
specific, DNA-independent interaction between endogenous
p53 and LSD1 proteins in p53-positive HEK293 cells (Fig. 2A;
see Fig. S1B in the supplemental material). We detected no
interaction between p53 and RBP2 (Fig. 2A), a member of the
JmjC family that can demethylate H3K4me2 (8, 32). Protein-
protein interactions between LSD1 and p53 do not require
enzymatic activity of the demethylase, as endogenous p53 as-
sociates with either a full-length or C-terminally truncated
(LSD1?C) version of LSD1 (Fig. 2B). We next determined if
p53 interacts directly with LSD1 by using recombinant pro-
teins. We performed in vitro co-IP experiments with recombi-
nant His-tagged p53 and recombinant Flag-tagged LSD1?C
proteins (Fig. 2C). The results show that the p53-LSD1 inter-
action is direct and does not require the catalytic domain of the
LSD1- and p53-mediated chromatin modification is re-
versed during liver regeneration. Two-thirds PH of adult
mouse liver induces a complex network of signal transduction
pathways, which trigger hepatocytes to reenter the cell cycle in
a synchronized wave of proliferation and compensatory growth
until the liver mass is fully regenerated (20, 40). During liver
regeneration, differentiation-associated repression of AFP
gene expression is reversed, and transcription is reactivated
(Fig. 3A) (23, 37, 53). Within 24 h of PH, p53 and LSD1
FIG. 1. p53 and LSD1 act together to repress AFP during liver development. (A) RT-PCR analysis of AFP and GAPDH RNA expression in
livers excised from 8-day (8d)- and 2-month (2mos)-old WT mice and in Hepa1-6 cells. (B) ChIP analysis of WT livers taken from mice at the ages
of 8 days and 2 months. Quantitative real-time PCR (upper panel) and conventional PCR analysis (bottom panel) (28 cycles) were performed to
measure relative antibody-bound DNA fragments of the SBE/p53RE region in 8-day- and 2-month-old livers. H3K4me2 levels were normalized
to H3 recovery. Each bar represents the average result for three independent ChIP experiments. Error bars show standard deviations.*, P ? 0.05;
**, P ? 0.01. (C) Re-ChIP assay of 2-month-old WT livers. Conventional PCR (32 cycles) analysis was performed with reciprocal re-ChIP assays
(primary antibody left of arrow and secondary antibody right of arrow) of p53 and LSD1 interaction at the SBE/p53RE.
FIG. 2. p53 interacts with LSD1 in vivo and in vitro. (A) Co-IP of endogenous p53 and endogenous LSD1 in HEK293 cells. Normal rabbit IgG
was used as a negative control. IP lysate (5%) was used as input. (B) Co-IP of expressed Flag-LSD1 protein or Flag-LSD1 with a carboxy-terminal
deletion (LSD1?C) and endogenous p53 protein in HEK293 cells. Expression of enhanced green fluorescent protein (EGFP) served as a negative
control. IP lysate (5%) was used as input. (C) Purified recombinant p53 and LSD1?C proteins interact directly, as shown by co-IP experiments.
VOL. 28, 2008 p53 RECRUITS LSD1 TO REPRESS CHROMATIN5141
interactions with AFP chromatin are significantly reduced, co-
incident with derepression of AFP expression (Fig. 3A and B).
Along with a loss of repressive p53, H3K4me2 levels increase
and Foxa1, a transactivator that opposes p53-mediated repres-
sion of AFP expression (34, 45), binds its response element,
intercalated within the SBE/p53RE (Fig. 3C).
The inverse relationship between binding of p53 and LSD1
and H3K4me2 modification of chromatin at the repressor re-
gion of AFP is further supported by ChIP analyses of fully
regenerated liver 7 days after PH, when AFP expression is
repressed once more (Fig. 3A and B). The SBE/p53RE repres-
sor is a complex regulatory element where Foxa factors, trans-
forming growth factor beta effectors Smad and SnoN, and p53
family members interact to regulate AFP expression. When
p53 and LSD1 are significantly depleted at the SBE/p53RE
repressor, during the regenerative response, multiple changes
in histone modifications occur in this region of chromatin,
including demethylation of H3K9me2, a ?2-fold increase in
H3K4me1 and H4ac, and relatively no change in H3K4me3 at
this distal regulatory site (data not shown; see Fig. S3 in the
The physiological roles of p53 during liver regeneration are
poorly understood, as the driving forces of regeneration re-
store full liver mass even in p53-null mice and p53-compro-
mised livers (18; our unpublished data). We found, by immu-
noblotting of liver nuclear extract with an antibody that detects
normal p53, that p53 is undetectable within 24 h of PH (Fig.
3D, 24-h PH extract overloaded for detection of p53). We are
currently investigating the signaling pathways that inactivate
and/or degrade p53 at peak times of proliferation during liver
regeneration (24 to 36 h after PH) (20; data not shown) and
those that restore p53 function once the liver mass is restored.
p53 is required for LSD1-mediated repression of AFP. We
determined if p53 is essential for recruitment of LSD1 in vivo
by ChIP analysis of liver tissue from p53-null mice. These mice,
disrupted in Trp53 by homologous recombination and lacking
expression of p53 (17, 38), maintain expression of AFP at 2
months of age, well beyond the developmental window of re-
pression observed in WT mice (45). Comparison of WT and
p53-null livers showed that in the absence of p53, LSD1 is
significantly reduced (Fig. 4A). Nonquantitative PCR analysis
of SBE/p53RE and the transcription start site supports the
sequence specificity of the p53 and LSD1 interaction with
chromatin and the lack of nonspecific binding compared to
FIG. 3. p53- and LSD1-mediated AFP repression and chromatin modification are reversed during liver regeneration. (A) (Right) RT-PCR was
conducted to detect the RNA levels of AFP and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) at different time points after PH and in
sham control tissue. (Left) qPCR was performed to quantify RNA levels. AFP levels were normalized to GAPDH levels. Each bar represents the
average result for three independent RT-PCR experiments. Error bars show standard deviations.*, P ? 0.05; ns, no statistical significance.
(B) ChIP analysis of mouse livers undergoing regeneration. qPCR was performed to quantify antibody-bound protein-DNA of SBE/p53RE from
livers at 24 h post-PH and 7 days post-PH compared to sham levels. Each bar represents the average result for at least six independent PCRs from
two independent ChIP experiments. Error bars show standard deviations.*, P ? 0.05;**, P ? 0.01. (C) ChIP analysis of mouse livers undergoing
regeneration. Conventional PCR (29 cycles) of the SBE/p53RE region was performed after ChIP at 24 h post-PH (24h PH) and with sham-treated
liver (24h Sham). (D) Protein levels of p53 were analyzed by immunoblotting of liver nuclear extracts isolated at different stages of development
and at 24 h post-PH. Recombinant His-p53 was used as a positive control for antibody detection, and the blot was reprobed for albumin (ALB)
expression for loading of each extract. The nuclear extract from 24 h post-PH was overloaded to attempt detection of p53; albumin levels remain
unchanged in liver at 24 h post-PH (57a). The slower migrating band positive for p53 (*) in the 24-h PH lysate is likely nonspecific, as it did not
appear with a different p53 antibody (CM5) (data not shown).
5142 TSAI ET AL.MOL. CELL. BIOL.
undiluted input DNA (Fig. 4B). The decrease in LSD1 inter-
action with AFP chromatin is not due to decreased expression
of LSD1, which is not altered in p53-null mice (Fig. 4C).
Previously, we saw that p53 and its family member TA-p73
are bound at the developmental repressor of AFP, concomi-
tant with developmental repression of AFP chromatin struc-
ture and expression. In p53-null liver, repression of AFP is
greatly delayed, from 2 to 3 weeks to 4 months, when chroma-
tin-bound TA-p73 levels increase compared to those in the
WT. When AFP is fully repressed at 4 months in the p53-null
mouse, H3K4me2 levels at the SBE-p53RE region remain
higher than those in 4-month-old WT mice, but H3K9me2
levels are equivalent to those of WT mice (12). Partial com-
pensation by p73 for loss of p53, which leads to a significant
temporal delay but not a reversal of repression, may be due in
part to significantly decreased but measurable recruitment of
LSD1 in p53-null liver (Fig. 4A). We saw that p73 and LSD1
interact in a protein complex when they are expressed ectopi-
cally in p53-positive U2OS cells (Fig. 4D), though an endoge-
nous association was undetectable (data not shown).
LSD1 is required for AFP repression in hepatocytes. We
used a nontransformed hepatocyte cell line to manipulate lev-
els of LSD1 expression and determine whether LSD1 is re-
quired for AFP repression in isolated hepatocyte-derived cells.
AML12 cells were established by continuous growth of primary
hepatocytes derived from liver tissue of a mouse transgenic for
human transforming growth factor alpha expression (62).
These immortalized hepatic cells recapitulate repression of
AFP and expression of albumin (ALB), as occurs in fully dif-
ferentiated liver tissue (Fig. 5A). The results of co-IP experi-
ments showed that endogenous p53 and LSD1 interact in these
cells (Fig. 5B), and ChIP analysis showed binding of p53 and
LSD1 at the SBE/p53RE region of endogenous AFP chromatin
(Fig. 5C). We transiently transfected siRNA oligonucleotides
to target LSD1 expression and found that depletion of LSD1
by siRNA led to derepression of AFP and no change in ex-
pression of ALB (Fig. 5D), reflecting the target specificity of
LSD1 depletion and its requirement in repression of AFP.
Dysfunction of p53 occurs in more than 50% of all human
cancers (6, 52, 58). We therefore determined if expression of
AFP as a tumor marker in hepatoma cells (Fig. 1A) was
consistent with a lack of p53 and LSD1 interaction with
chromatin at the repressor region of AFP. We performed
ChIP analyses to detect binding of p53 and LSD1 on AFP
chromatin in Hepa1-6 cells, which were originally cultured
from a clonally derived mouse hepatoma tumor (13) and
express signal-responsive p53 (see Fig. S4 in the supplemen-
tal material). We found that p53 and LSD1 do not interact
with AFP chromatin and that H3K4me2 levels are readily
detectable at the SBE/p53RE (Fig. 5E). As a positive con-
trol, LSD1 did associate with a REST response element of
calbindin chromatin (Fig. 5E, lower panel). Previously,
FIG. 4. p53 is required for LSD1 recruitment to the developmental repressor region of AFP. (A) ChIP analyses of 2-month-old WT and p53?/?
mouse liver tissues. qPCR was performed to assay relative antibody-bound SBE/p53RE from WT and p53?/?mouse liver tissues. Each bar is an
average result for three independent ChIP experiments. Error bars show standard deviations.**, P ? 0.01. (B) ChIP analyses of 2-month-old WT
and p53?/?mouse liver tissues. Conventional PCRs (30 cycles) of the SBE/p53RE region and transcription start site of AFP were conducted to
determine the binding of p53 and LSD1. (C) RT-PCR (left) and immunoblot (right) analyses were performed with RNAs and nuclear extracts
isolated from liver tissues excised from 2-month-old WT and p53?/?mice to determine the relative levels of LSD1 and GAPDH RNA and protein
expression. (D) Co-IP of expressed Flag-LSD1 protein and hemagglutinin-p73 (HA-p73) protein in U2OS cells. Expression of enhanced green
fluorescent protein (EGFP) served as a negative control. IP lysate (5%) was used as input.
VOL. 28, 2008p53 RECRUITS LSD1 TO REPRESS CHROMATIN 5143