Wild-type but not mutant p53 activates the hepatocyte growth factor/scatter
factor promoter
Wild-type but not mutant p53 activates the hepatocyte growth factor/scatter factor promoter
Anna M. J.
Metcalfe*
,
Ruth M.
Dixon
and
George K.
Radda
1
MRC Clinical and Biochemical Magnetic Resonance Unit, Department of
Biochemistry, University of Oxford, South Parks Road,
Oxford
OX1 3QU,
UK
and
1
MRC, 20 Park Crescent,
London
W1N 4AL,
UK
Received November 25, 1996;
Revised and Accepted January 14, 1997
ABSTRACT
p53 transactivates the expression of a variety of genes by binding to specific
DNA sequences within the promoter. We have investigated the ability of wild-type p53 and a non-DNA binding p53 mutant to activate the hepatocyte growth factor/scatter factor (HGF/SF) promoter using chloramphenicol acetyltransferase reporter
constructs. We also used deletion sequences of the HGF/SF promoter to identify
which regions, if any, were responsible for p53 binding. Our results show that
wild-type but not mutant p53 activates the HGF/SF promoter when using -3000 and -755 bp upstream of the HGF/SF gene. This activation is lost when promoter
sequences covering -365 and -239 bp are used. Analysis of the DNA sequence between -365 and -755 bp shows one putative p53 half-site with 80% homology to the consensus
sequence and another half-site 3 bases downstream of this with 100% homology to the consensus
sequence. In contrast to previously identified p53 binding DNA sequences, the
downstream half-site is inverted. We propose that the HGF/SF promoter can be activated by
wild-type p53
in vivo
and that this could be as a result of a novel form of sequence-specific DNA binding.
INTRODUCTION
Wild-type p53 is a transcription factor that serves a dual role in the control of cellular proliferation. It is probably best recognized for its tumour suppressor function whereby p53 can induce either a G
1
/G
2
arrest
(
1
,
2
) or apoptosis (
3
) in cells that have sustained DNA damage. The nature, extent and site of DNA damage all relate to the type of
p53-mediated response that ensues, as reviewed by Gottlieb and Oren (
4
), indicating that this tumour suppressor is a highly versatile and response-selective `guardian of the genome'.
p53 can also positively regulate cell division and differentiation by specific induction or repression of promoters for growth- associated factors and growth factor receptors (
5
). The cloning and sequencing of p53 cDNAs from several species has led to a
detailed understanding of structural features of the protein. The ability of
p53 to activate gene expression by sequence-specific DNA binding has been attributed to amino acid residues 102-292 (
6
), which lie in the core domain of the protein. To date, the number of growth-related genes which are known to be transcriptionally activated by p53 in this manner are relatively few.
When p53 is mutated, as it is in ~50% of all human cancers (
7
), it is usually a missense mutation occurring in the core domain of the
protein. Depending on the nature of the mutation, it can give rise to loss of
wild-type p53 function, thus allowing cell cycling to proceed unchecked and
inappropriate gene expression to occur. Wild-type p53 has been observed to stimulate activation of the epidermal growth
factor receptor (EGF-R) promoter (
8
) and could be an important regulator of its expression in normal development.
Notably, stimulation shows increased sensitivity to certain forms of mutant
p53, suggesting that cell proliferation could be enhanced by overexpression of
the EGF-R in cancers involving these p53 mutants.
Growth factors act in a paracrine, autocrine or endocrine fashion and interact
with specific cell surface receptors. p53 may influence the growth factor
network by regulating the expression of certain growth factor receptors, such
as the EGF-R and c-met (
9
,
10
), or other growth-related molecules, such as transforming growth factor-[alpha] (
11
) and insulin-like growth factor binding protein 3 (
12
).
HGF/SF is derived from a single chain molecule which is proteolytically cleaved
to form a biologically active heterodimer consisting of an [alpha] chain (69 kDa) and a [beta] chain (34 kDa) (
13
). It is produced by a wide variety of tissues, including the liver, pancreas,
salivary glands, thyroid, duodenum and kidney, and is thought to exert its
effects mainly by paracrine interaction with the cell surface receptor c-met (
14
). Studies indicate that expression of HGF/SF is restricted to cells of mesenchymal origin (
15
) in a manner that is highly controlled and developmentally regulated.
The biological effects of HGF/SF are numerous. It is the most potent stimulator
of DNA synthesis for mature hepatocytes (
16
). It also stimulates the invasiveness and movement of epithelial cells, but is
cytostatic to others (
17
-
19
). Because of its significant effect on the proliferation and invasive capacity
to some cell types, HGF/SF may be an important candidate for enhancing tumour
development and metastasis formation. Indeed, several tumour types have been
reported to show elevated expression of HGF/SF (
20
-
24
). It is of great interest to elucidate the mechanisms responsible for
controlling the expression of both HGF/SF and c-met with the aim of understanding neoplasia formation and progression
associated with abnormal HGF/SF and c-met expression.
In order to decipher positive and negative regulatory elements involved in
controlling HGF/SF expression, the promoter regions of human, mouse and rat
HGF/SF have been cloned and sequenced (
25
-
27
). Cell lines have also been used to measure HGF/SF expression in response to
various ligands. These include epidermal growth factor (
28
), platelet-derived growth factor (
28
), fibroblast growth factor (
28
), IL-1 (
29
), injurin (
30
) and injurin- like factor (
31
), all of which increase the expression of HGF/SF in human skin fibroblasts or
MRC-5 cells. Transforming growth factor-[beta] and some glucocorticoids down-regulate HGF/SF production in leukaemia cell cultures and human lung
fibroblasts (
32
).
Promoter sequence analysis has further characterized potential response elements
and identified a high degree of sequence conservation between species (95%
between mouse and rat and ~90% between rodent and human), implying common regulatory mechanisms of
HGF/SF gene expression. Whilst one putative p53 half-site has previously been identified in the rat HGF/SF promoter (
27
), no experimental evidence for transcriptional activation of the HGF gene by
p53 has been provided to date.
We observed several putative p53 half-sites within 1000 bp of the HGF/SF promoter that have not been previously
identified. Here we have investigated the ability of p53 to activate different
HGF/SF promoter constructs. We show that wild-type human p53, but not mutant p53, activates murine HGF/SF promoter-
CAT
constructs which contain the novel p53 half-sites. This suggests that wild-type p53 can regulate cell proliferation indirectly, via HGF/SF
promoter activation.
MATERIALS AND METHODS
Plasmids
Wild-type and mutant human p53 expression plasmids (kindly provided by Jo
Milner, YCRC p53 Laboratory, Department of Biology, York University, York, UK) used the human cytomegalovirus (CMV) major immediate-early promoter-enhancer (bases 209-864) in the vector pRc/CMV (Invitrogen). pRc/CMV wt hp53
contains wild-type human p53 cDNA, whilst pRc/CMV M237I contains mutant human p53 with a
Met -> Ile substitution at amino acid 237 (
33
).
CAT plasmids containing the HGF/SF promoter sequences (-3000, -755, -365 and -291 bp) were made by ligating restriction fragments
into the promoter-less plasmid pCAT-basic (Promega) as previously described by Plaschke-Schlutter
et al.
(
26
).
Cell culture and transfection
The cells used in these experiments were clone D4
ras
-transformed NIH 3T3 mouse fibroblasts (
34
) which were propagated in Dulbecco's modified minimum essential medium
containing 10% foetal bovine serum. In a typical experiment, 1 * 10
6
cells were plated into a 100 mm diameter Petri dish and co-transfected with 5 [mu]g HGF/SF-
CAT
construct and 5 [mu]g p53 expression plasmid. Transfections were carried out using the calcium
phosphate method as previously described (
35
). A control plasmid containing the Rous sarcoma virus promoter and the
Escherichia coli
lacZ
gene was co-transfected in each experiment.
CAT assay
Cells were harvested 48 h post-transfection and lysed by three cycles of freeze-thawing. Cell extract volumes used in the CAT assay were adjusted
according to the [beta]-galactosidase activity in each transfection. CAT enzyme activity was
then measured using
14
C-labelled chloramphenicol (Dupont) as described previously (
36
). Each transfection was repeated three times.
RESULTS AND DISCUSSION
Wild-type p53 transactivates -3000 and -755 bp HGF/SF promoter constructs
Inability of wild-type p53 to activate -365 and -239 bp of the HGF/SF promoter and identification of putative
p53 binding sites
To identify which regions of the HGF/SF promoter sequence are responsible for
p53 binding we performed transfections using truncated constructs of the HGF/SF
promoter, one being -365 and the other -239 bp upstream of the gene. The CAT activity was greatly reduced
using either the -365 bp construct (Fig.
2
) or the -239 bp construct (6% conversion of [
14
C]chloramphenicol in each case) to values almost identical to those obtained
with mutant p53 transfections (4 and 8% respectively). This indicates that
either all or part of the p53 response element in the HGF/SF promoter exists
between -755 and -365 bp of the promoter.
In our experiments, the mouse HGF/SF promoter DNA sequence between -755 and -365 bp included one p53 half-site (a decamer) with 80% homology to the consensus sequence
(5'-G/A G/A G/A C A/T A/T G T/C T/C T/C-3') (
45
). This putative half-site is situated three thymidines downstream of another half-site (100% homology to consensus sequence) which runs in the
opposite direction (inverted). The sequences of these are A
C
ACATGC
A
T (-394 to -385) and CCTGTTCAAA (-381 to -372) (bases which do not conform to the consensus
sequence are underlined). The rat promoter (
27
) contains identical sequences to this and the human sequence (
25
) has one base change in each half-site (Fig.
3
).
CAT activity was lost when a -365 bp construct was used (Fig.
2
). Analysis of the sequence between -365 bp and the start site for gene transcription failed to identify any
further p53 half-sites. However, we have noted that there is another half-site with 90% homology to the consensus sequence in the HGF/SF
promoter at ~-1000 bp. Thus our data imply that this half-site is not essential for p53 binding, as it is not contained within
the -755 bp promoter which gave the strongest CAT activity. A graphical
representation of our data is given in Figure
4
.
ACKNOWLEDGEMENTS
This work was funded by grants from the Medical Research Council and the
Deutsche Forschungsgemeinschaft. We thank Susan Metcalfe for critical review of the manuscript and Jo Milner for her support
throughout this work and for collaboration in gel shift experiments (funded in part by the European Community, Human Capital Mobility Grant CHRX-CT93-0180). We are also very grateful to Jürgen Behrens for expert practical advice in performing CAT assays, Walter Birchmeier for providing
laboratory facilities and Ermanno Gherardi for valuable discussions.
