ABSTRACT
Tissue-type plasminogen activator (t-PA) gene expression in human endothelial cells and HeLa cells is stimulated
by the protein kinase C activator phorbol 12-myristate 13-acetate (PMA) at the level of transcription. To study the mechanism of transcriptional regulation,
we have characterized a segment of the t-PA gene extending from -135 to +100 by
in vivo
footprinting analysis [dimethyl sulphate (DMS) method] and gel mobility shift assay.
In vivo
footprinting analysis revealed changes in cleavage pattern in five distinct
promoter elements in both endothelial cells and HeLa cells, including a PMA-responsive element (TRE), a CTF/NF-1 binding site and three GC-boxes, and an altered cleavage pattern of the TRE and CTF/NF-1 element after PMA treatment of HeLa cells. Although
endothelial cells and HeLa cells differed in the exact G residues protected by
nuclear proteins,
in vitro
bandshift analysis showed that nuclear protein binding to the t-PA promoter was qualitatively and quantitatively very similar in both cell
types, except for the TRE. Protein binding to the TRE under non- stimulated conditions was much higher in human endothelial cells than in HeLa cells, and this TRE-bound protein showed a lower dissociation rate in the endothelial cells
than in HeLa cells. In endothelial cells, the proteins bound to the TRE
consisted mainly of the AP-1 family members JunD and Fra-2, while in HeLa cells predominantly JunD, FosB and Fra-2 were bound. The proteins bound to the other protected
promoter elements were identified as SP-1 (GC-box II and III) and CTF/NF-1 (CTF/NF-1 binding site). After PMA treatment of the cells, AP-1 and SP-1 binding was increased two-fold in endothelial cell nuclear
extracts and >20-fold in HeLa nuclear extracts. In the endothelial cells, all Jun and Fos
forms (c-Jun, JunB, JunD, c-Fos, FosB, Fra-1 and Fra-2) were part of the AP-1 complex after PMA induction. In HeLa cells, the
complex consisted predominantly of c-Jun and the Fos family members FosB and Fra-2. In the light of previous studies involving mutational analysis of
the human and murine t-PA promoter our results underline an important role of the five identified
promoter regions in basal and PMA-stimulated t-PA gene expression in intact human endothelial cells and HeLa cells. The small differences in DMS protection pattern and differences in the individual AP-1 components bound in endothelial cells and HeLa cells point to subtle cell-type specific differences in t-PA gene regulation.
Tissue-type plasminogen activator (t-PA) plays a key role in the dissolution of the fibrin matrix of
thrombi and haemostatic plugs (
1
). t-PA catalyzes the conversion of the zymogen plasminogen into the active
serine proteinase plasmin, the enzyme that digests fibrin. Gene targeting and
gene transfer studies have confirmed the significant role of t-PA-mediated plasminogen activation in maintaining vascular patency (
2
). Regulation of t-PA expression, both
in vitro
and
in vivo
, has therefore been the focus of many studies (
3
).
t-PA in the circulation originates predominantly from the vascular
endothelium (
3
).
In vitro
studies using cultured human endothelial cells have demonstrated that
activation of protein kinase C (PKC) with vasoactive compounds such as thrombin
or histamine, or with the phorbol ester 4[beta]-phorbol 12-myristate 13-acetate (PMA) stimulates t-PA expression (
4
-
7
). Based on nuclear run-on transcription assays, t-PA expression is modulated by PMA at the level of transcription (
8
). A very strong induction of t-PA gene transcription with PMA was found in HeLa cells (
9
). Transient transfection experiments in HeLa cells using deletion mutants of
the t-PA gene promoter fused to the chloramphenicol acetyltransferase (CAT)
reporter gene revealed that two regions in the t-PA promoter (between positions -102 to -115 and +60 to +74) are critical for basal and PMA-stimulated t-PA promoter activity (
9
). To pursue the physiological significance of these studies, we performed
in vivo
footprinting analysis in control and PMA-treated human endothelial cells and HeLa cells to reveal the pattern of
protein-DNA interactions in the intact cell, where the nucleic acid is complexed
with chromosomal proteins to form chromatin. In addition, such studies may
reveal cell type-specific differences between primary human endothelial cells and the established human cervical carcinoma cell line, HeLa. Gel mobility shift assays were performed to identify the nuclear proteins which
interact with the various binding sites in the promoter region of t-PA as revealed by
in vivo
footprinting.
Dimethyl sulphate (DMS) and piperidine were obtained from Fluka (Bornem, The
Netherlands). T
4
kinase and DNA ligase were obtained from Promega (Madison, WI). dNTPs (ultra
pure),
Taq
polymerase and rATP were obtained from Pharmacia Biotech (Woerden, The
Netherlands). 4[beta]-phorbol 12-myristate 13-acetate (PMA) was obtained from Sigma (St Louis, MO). A
stock solution of PMA (100 [mu]M) was prepared in ethanol and kept at -20oC. The anti-c-Jun and anti-c-Fos rabbit polyclonal antibodies were a gift from Dr T. Oehler
(Massachusetts Institute of Technology, Cambridge, MA) and Dr H.J. Rahmsdorf (Kernforschungszentrum, Karlsruhe, Germany), respectively. These antibodies preferentially recognize c-Jun and c-Fos, but also bind other members of the Jun and Fos protein family.
Antibodies specific for the AP-1 proteins c-Jun, JunB, JunD, c-Fos and Fra-1 were obtained from Santa Cruz Biotechnology (Santa
Cruz, CA). Anti-FosB and anti-Fra-2 were a gift from Dr C. Pfarr and Dr M. Yaniv (Institut
Pasteur, Paris). Anti-CTF/NF-1 rabbit polyclonal antibody (
10
) was a gift from Dr W. van Driel (Laboratory for Physiological Chemistry,
Utrecht University, The Netherlands). Anti-SP-1 and anti-AP-2 rabbit polyclonal antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). 5[[gamma]-
32
P]triphosphate (3 Ci/[mu]mol), deoxycytidine 5[[alpha]-
32
P]triphosphate (3 Ci/mol), uridine 5[[alpha]-
32
P]triphosphate (>400 Ci/mol) and Sequenase (version 2) were obtained from
Amersham Nederland BV ('s Hertogenbosch, The Netherlands). Bradford protein reagent was obtained from from Bio-Rad (Veenendaal, The Netherlands).
The primers used for the
in vivo
DMS footprint analysis were synthesized and HPLC-purified by Isogen Bioscience (Amsterdam, The Netherlands), and had the following sequences:
elongation primer 1: 5'-CCCTTTTAAGCCTGGGACATAG-3';
PCR primer 2: 5'-GACTCTAAAGGAAGATGATTCTTAAGGTCCC-3';
elongation primer 3: 5'-GGAAGATGATTCTTAAGGTCCCATCCCACTCC-3';
25-mer linker: 5'-GCGGTGACCCGGGAGATCTGAATTC-3';
and 11-mer linker: 5'-GAATTCAGATC-3'.
The oligodeoxynucleotides used for the bandshift assays were synthesized by Isogen Bioscience (Amsterdam, The Netherlands), and correspond to the following regions of the t-PA promoter:
-120 to -98 (TRE-like): 5'-GATTCAATGACATCACGGCTGTG-3';
-95 to -72 (CTF/NF-1-like): 5'-TAATCAGCCTGGCCCGAAGCCAGGG-3';
-49 to -28 (GC-box I): 5'-TGAACTTCCTCCCCCTGCTTTA-3';
+30 to +52 (GC-box II): 5'-ACACAGAAACCCGCCCAGCCGG-3';
+57 to +78 (GC-box III): 5'-ACCGACCCCACCCCCTGCCTGG-3'.
As an aspecific competitor oligodeoxynucleotide, the following (random) sequence
was used: 5'-CTGAGGATTCTCCACTGCA-3'. An AP-2 consensus sequence oligodeoxynucleotide was
obtained from Santa Cruz Biotechnology (Santa Cruz, CA) and had the sequence: 5'-GATCGAACTGACCGCCCGCGGCCCGT-3'. The consensus CTF/NF-1 oligodeoxynucleotide used in competition
experiments, 5'-CCTTTGGCATGCTGCCAATATG-3', was obtained from Promega (Madison, WI).
Endothelial cells from human umbilical cord veins (HUVEC) were isolated by the
method of Jaffe
et al
. (
11
), and cultured as previously described (
12
). HUVEC were grown on fibronectin-coated dishes in Dulbecco's modified Eagle's medium (DMEM) supplemented
with N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES) (20 mM), newborn calf
serum (heat-inactivated; 10% v/v), human serum (10% v/v), heparin (5 IU/ml), endothelial cell
growth supplement (150 [mu]g/ml) (
13
), L-glutamine (2 mM), penicillin (100 IU/ml) and streptomycin (100 [mu]g/ml). HeLa cells were grown in DMEM supplemented with HEPES (20 mM),
fetal bovine serum (heat-inactivated; 8% v/v), L-glutamine (2 mM), penicillin (100 IU/ml) and streptomycin (100 [mu]g/ml). Both cell types were grown at 37oC under a 5% CO
2
/95% air atmosphere, and the medium was replaced every 2-3 days. Subcultures were obtained by trypsin/ethylenedinitrilo-tetraacetic acid disodiumsalt-dihydrate (EDTA) treatment at a split ratio of 1:3 for HUVEC
and of 1:10 for HeLa cells. HUVEC were cultured for maximally three passages.
For experiments, confluent cultures were used and the cells were always re-fed the day before the experiment with incubation medium, i.e. for HUVEC:
DMEM supplemented with human serum (10%), L-glutamine, penicillin and streptomycin; and for HeLa cells: DMEM
supplemented with L-glutamine, penicillin and streptomycin. After incubation of the cells with
incubation medium containing the appropriate concentration of PMA (i.e. 10 nM for HUVEC and 162 nM for HeLa cells) or stock solvent, the cells were used for
in vivo
footprint analysis or the preparation of nuclear extracts.
Confluent cultures of HUVEC (486 cm
2
) or HeLa cells (162 cm
2
) were washed once with phosphate-buffered saline (PBS) (0.15 M NaCl, 10 mM Na
2
HPO
4
, 1.5 mM KH
2
PO
4
, pH 7.4) at room temperature, and then incubated with DMEM supplemented with 10
mM HEPES (pH 7.5) and 0.5% (v/v) DMS for 2 min. The cells were washed with ice-cold PBS and lysed in 10 mM Tris pH 7.7, 400 mM NaCl, 2 mM EDTA and 0.2%
(w/v) SDS. DNA was isolated by digestion with proteinase K (300 [mu]g/ml, 37oC overnight), followed by phenol/chloroform extraction and ethanol precipitation (
14
). The DNA was dissolved in water to a final concentration of 1 [mu]g/[mu]l, and incubated with 10% (v/v) piperidine for 30 min at 90oC. After ethanol precipitation, the samples were processed for
ligation-mediated PCR (LMPCR) analysis.
In vitro
controls were obtained by the reaction of purified DNA with DMS as described by
Maxam and Gilbert (
15
).
LMPCR was performed by the method described by Mueller and Wold (
16
). Elongation primer 1 (0.6 pmol, see Materials section) was annealed to 10 [mu]g heat-denatured (3 min, 95oC) piperidine-cleaved DNA at 45oC for 30 min. Primer extension was then carried out
with Sequenase version 2.0 for 15 min at 45oC by adding 8.8 [mu]l of elongation mixture (20 mM MgCl
2
, 20 mM DTT, 200 [mu]M of dATP, dCTP, dGTP and dTTP, and 0.3 [mu]l Sequenase version 2). The DNA polymerase was heat-inactivated by incubation at 67oC for 15 min. Ligation of the universal linker (100 pmol of
annealed 25mer and 12mer oligodeoxynucleotide, see Materials section) to the
primer-extended molecules was done overnight at 15oC by adding 20 [mu]l of 17.5 mM MgCl
2
, 42.3 mM DTT and 125 [mu]g/ml bovine serum albumin (BSA) and 25 [mu]l of ligation mix (10 mM MgCl
2
, 20 mM DTT, 3 mM ATP, 50 [mu]g/ml BSA and 0.4 U/[mu]l T
4
DNA ligase). After ethanol precipitation, the pellets were dissolved in 60 [mu]l water. PCR amplification was performed in 10 mM Tris (pH 8.8), 40 mM NaCl,
5 mM MgCl
2
, 10 pmol primer 2, 10 pmol 25mer linker primer (see Materials section), 10 U
Taq
polymerase, 0.2 mM of dATP, dCTP, dGTP and dTTP and 0.01% (w/v) gelatin in a
total volume of 100 [mu]l on a Perkin Elmer Thermocycler 9600. Twenty cycles of PCR (1 min 95oC, 2 min 64oC and 3 min 75oC) were performed. Subsequently, linear PCR was done with 2 pmol end-labelled elongation primer 3 (see Materials section), 5
U
Taq
polymerase, 2 [mu]l 2.5 mM dNTP-mix. One PCR cycle (2 min 95oC, 2 min 66oC and 10 min 75oC) was performed. The PCR-amplified fragments were extracted with
phenol/chloroform, ethanol precipitated, and then separated on a 6% (w/v)
denaturating polyacrylamide-gel (
15
). The sequence gel was dried on Whatman 3MM paper, and radiolabelled DNA fragments were visualized by autoradiography.
For gel shift experiments confluent cultures of HUVEC (324 cm
2
) or HeLa cells (162 cm
2
) were rinsed twice with ice-cold PBS and lysed in 2 ml of lysis buffer (10 mM Tris pH 7.4, 10 mM NaCl,
3 mM MgCl
2
, 0.5% NP-40, 1 mM DTT, 0.25 mM vanadate and 1 [mu]g/ml of the protease inhibitors leupeptin, pepstatin and aprotinin).
The lysates were homogenized in a potter (20 strokes); nuclei were collected by centrifugation (5 min at 1000
g
, 4oC), and washed once with lysis buffer. The nuclear pellet was resuspended in 150 [mu]l of 20 mM HEPES (pH 7.9), 400 mM NaCl, 1 mM EDTA, 1 mM ethylene glycol-bis(oxyethylenenitrilo)tetraacetic acid (EGTA), 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride
(PMSF), 0.25 mM vanadate and 1 [mu]g/ml of leupeptin, pepstatin and aprotinin. Suspensions were incubated for
15 min at 4oC while being continuously shaken, and then centrifuged at 1000
g
, 4oC for 5 min. Supernatants were stored at -80oC until use. The protein concentrations in the nuclear extracts
were determined using the Bradford protein assay.
Oligodeoxynucleotides were end-labelled using T
4
-kinase and subsequently purified by phenol/chloroform extraction and ethanol
precipitation. For the electromobility shift assay (EMSA), 25 fmol (~10
4
c.p.m.) of labelled double-stranded oligodeoxynucleotide was mixed with nuclear extract (5 [mu]g protein) in a total volume of 20 [mu]l of 20 mM HEPES (pH 7.9), 20 mM KCl, 2 mM MgCl
2
, 20% glycerol, 2.5 mM EDTA, 2 mM spermidine, 1 [mu]g poly(dI-dC), 1 [mu]g BSA and 1 mM PMSF. The mixture was incubated at 4oC for 30 min. All bandshifts were performed in the presence of a 100-fold excess of unlabelled nonhomologous DNA in order to
prevent aspecific probe/protein interactions. Furthermore, all DNA- protein complexes were checked for the sequence specifity of the binding reaction by adding a 100-fold excess of the same, unlabelled, double-stranded oligodeoxynucleotide (100* competitor). For generation of supershifted complexes, the nuclear
extracts were preincubated for 1 h at 4oC with the appropriate antiserum prior to the binding reaction. The
antibodies against Fos and Jun were previously shown not to interfere with
nuclear protein binding to the jun1 TRE and the SV-40 enhancer, respectively (
17
,
18
). The antibodies against SP-1 and CTF/NF-1 did not inhibit nuclear protein binding to the human t-PA GC-box I and to the t-PA TRE, respectively (data not shown). DNA-protein complexes were separated from the non-bound oligodeoxynucleotide by electrophoresis on a 5%
polyacrylamide gel in 0.25* TBE buffer (22.5 mM Tris-borate, 0.5 mM EDTA) (
14
). Electrophoresis was carried out at room temperature at 150 V for 70 min, using 0.25* TBE as running buffer. The gel was dried on Whatman 3MM paper, and DNA-protein complexes were visualized by autoradiography.
To study the interactions between nuclear proteins and the human t-PA promoter sequence
in vivo
in intact HUVEC and HeLa cells, we performed genomic DMS footprinting. Using the appropriate primers, the DNA-protein interaction sites of the t-PA promoter region between -130 and +100 were mapped. As shown in Figure
1
and summarized in Figure
2
, only nine out of the 16 affected residues coincide in HUVEC and HeLa cells.
However, all affected residues are clustered around the same five consensus
sites for transcription factor binding. These five boxes consist of a PMA
responsive element (TRE) between positions -112 and -104; a consensus site for the family of CCAAT-binding transcription factors, also referred to as nuclear factor 1 (CTF/NF-1) binding site, between positions -92 and -77; and three GC-boxes between positions -43 and -34, +39 and +45, and
+62 and +68, which have homology to SP-1 and activator protein-2 (AP-2) binding sites. Although the affected residues are clustered
around the same five boxes in HUVEC and HeLa cells, the observed differences in
the pattern of protection suggest that the proteins bound in HeLa and HUVEC may be similar but are not necessarily identical.
To identify the nature of the proteins bound to the t-PA promoter, the DNA-protein interactions were studied
in vitro
by using the electromobility shift assay (EMSA). All five regions identified
with the
in vivo
footprinting assay bound nuclear protein, and protein binding was qualitatively
and quantitatively comparable in HUVEC and HeLa cells, except for the TRE-like binding site (Figs
3
-
9
). The specific DNA-protein complex formed with the TRE-like sequence was far more abundant with nuclear extracts from
HUVEC than from HeLa cells (Fig.
3
), which is in agreement with the higher protection of this region in HUVEC in
the
in vivo
footprint analysis (Fig.
1
). Dissociation experiments showed that the association of protein to the TRE-like site was stable over a 20 min period when using nuclear extracts from
HUVEC, while with nuclear extracts from HeLa cells the protein rapidly
dissociated from the DNA (Fig.
4
). PMA strongly induced protein binding with nuclear extracts from HeLa cells
(>20-fold) but hardly further increased protein binding with nuclear extracts
from HUVEC (~2-fold) (Fig.
3
). This is consistent with the
in vivo
footprint data, which showed a stronger induction of protection in PMA-treated HeLa cells than in PMA-treated HUVEC. PMA did not alter the protein dissociation rate with
nuclear extracts of either cell type (Fig.
4
), indicating that the strong induction of protein binding activity in the
nuclear extracts from PMA-treated HeLa cells is probably the result of an increase in protein
levels.
TRE-like consensus sites bind transcription factors belonging to the families
of the activator protein-1 (AP-1) or the cAMP responsive element binding (CREB) proteins, and also heterodimers formed between these two families like c-Jun/ATF-2 (
17
). We found that protein binding to the TRE-like region of the t-PA promoter in both HUVEC and HeLa nuclear extracts was strongly
inhibited with polyclonal antibodies directed against the AP-1 family members
Jun
and
Fos
(Fig.
3
). Apparently, this region of the t-PA promoter is bound by
Jun/Fos
heterodimers. Similar results were obtained with nuclear extracts of PMA-treated HUVEC (Fig.
3
). The protein-DNA complex formed with nuclear extracts from PMA-treated HeLa cells was fully inhibited with anti-Fos polyclonal antibody, but only partially (about 50%) with
anti-Jun polyclonal antibody (Fig.
3
). To further characterize the composition of the AP-1 complexes bound to the t-PA TRE, we performed bandshift experiments with antibodies specifically recognizing the different AP-1 family members (Fig.
5
). With nuclear extract from non-stimulated endothelial cells a marked decrease in AP-1 binding was observed after pre-incubation of the nuclear extract with anti-JunD and anti-Fra-2 antibodies, suggesting that complex formation of the antibody with
JunD or Fra-2- containing dimers mostly interfered with DNA binding but hardly yielded the
formation of `supershifted' complexes. A slight but significant decrease was seen with anti-c-Jun, anti-JunB, anti-c-Fos, anti-FosB and anti-Fra1 antibodies (Fig.
5
). After PMA treatment of the HUVEC, all AP-1 family members appear to contribute to a similar extent to the increase
in AP-1 binding activity. The DNA binding of the AP-1 proteins present in untreated HeLa nuclear extracts could be
reduced predominantly by anti-JunD, anti-FosB, and anti-Fra-2 antibodies, and, to a minor extent, by anti-c-Fos and anti-Fra1 antibodies, as measured by inhibition of binding in the presence
of antiserum. Upon incubation of HeLa cells with PMA, a strong induction of the binding of c-Jun, FosB and Fra-2 was observed. These results show that in both untreated and PMA-stimulated human endothelial and HeLa cells, AP-1 binding to the t-PA TRE is not only quantitatively but also
qualitatively different.
Previous investigations of the regulation of the human t-PA gene by transient transfection assays in HeLa cells employing deletion mutants of the t-PA gene promoter demonstrated that the DNA elements which regulate constitutive and PMA-stimulated expression are encoded by sequences downstream of position -115 of the t-PA gene (
9
). In this study, we further characterized the -135 to +100 region of the human t-PA promoter for persistent and PMA-inducible DNA-protein interactions in cultured vascular endothelial
cells and HeLa cells.
In vivo
genomic footprinting analysis revealed five distinct protein binding elements
in both endothelial cells and HeLa cells, corresponding to a PMA responsive
element (TRE; -112 to -104), a CTF/NF-1 binding site (-92 to -77) and three GC-boxes (-43 to -34, +39 to +45, and +62 to
+68). After PMA treatment of HeLa cells, the G residues of the TRE consensus
sequence (-113, -107 and -104) were less susceptible to methylation, reflecting
enhanced protein binding. In contrast to HeLa cells, in HUVEC these residues
were already fully occupied by protein under non-stimulated conditions, and PMA treatment had no marked effect on their
methylation and subsequent cleavage. The subtle differences between endothelial
and HeLa cells in the DMS footprint pattern of the TRE consensus sequence are reflected in subtle differences in the composition of the bound
protein complexes. In human endothelial cells, the proteins bound to the TRE
consisted mainly of the AP-1 family members JunD and Fra-2, while in HeLa cells predominantly JunD, FosB and Fra-2 were bound. Also, the t-PA TRE sequence was bound much more efficiently by the
AP-1 complexes from endothelial cells than from HeLa cells.
After PMA treatment of the endothelial cells, all Jun and Fos forms (c-Jun, JunB, JunD, c-Fos, FosB, Fra-1 and Fra-2) contributed to the AP-1 complex, while in HeLa cells, the AP-1 complex consisted predominantly of c-Jun and the Fos family members FosB
and Fra2. Since the various members of the Jun and Fos families of transcription factors differ in their transactivation potential (
19
), these differences in promoter-associated factors may be involved in cell-type specific differences in t-PA expression. Our results show also subtle differences in
binding of nuclear proteins distinct from AP-1 to the t-PA promoter in human endothelial and HeLa cells, both in respect to
the DMS protection pattern and to the transcription factors bound. The
identified transcription factors bound to the CTF/NF-1 site and GC-boxes II and III included CTF/NF-1 and SP-1, respectively, in both cell types. Like for the AP-1 binding site, the presence of subtle differences
in the DMS footprint pattern of the CTF/NF-1 binding site may point at cell-type specific differences in the CTF/NF-1 family members bound, since the CTF/NF-1 family exists of a large group of related
transcription factors (
20
).
The lower dissociation rate of the DNA-bound AP-1 complex in human endothelial cells as compared to HeLa cells may
be due to differences in posttranslational modification (i.e. phosphorylation) of the AP-1 proteins (
19
) or to a different activity of the AP-1 inhibitory protein IP-1 (
21
). IP-1 function is blocked by protein kinase A (PKA) activation, which also
further enhances t-PA transcriptional induction by PMA (
22
,
23
). We observed, however, no increase in AP-1 binding activity after treatment of both cell-types with the cAMP-raising compound forskolin (Arts
et al
., unpublished data), which argues against a role of IP-1 in the regulation of AP-1 binding activity. However, at present the existence of other cellular proteins modulating the DNA binding
activity of AP-1 which are not regulated by PKA-dependent signalling cannot be excluded.
Two of the five protein binding sites, the TRE and GC-box III, were also reported by Medcalf
et al
. (
9
) to be essential for basal and PMA-induced t-PA promoter activity in HeLa cells on the basis of mutational
analysis. We found, using gel-shift assays, that nuclear protein binding to each of these sites was
strongly induced in PMA-treated HeLa cells (~20-fold) and ~2-fold in human endothelial cells, which parallels the difference in
transcriptional induction of t-PA by PMA in HeLa cells and HUVEC (
8
,
9
). In contrast to the increase in nuclear protein binding to the TRE, the
enhanced nuclear protein binding to GC-box III was not reflected in a change in the cleavage pattern of the G
residues of this region in the
in vivo
footprint, possibly because these G residues were already optimally accessible
for methylation, or because of a high rate of exchange of bound proteins to
this sequence.
Our observation that GC-boxes II and III bind SP-1 protein is in line with previous reports that human t-PA transcription predominantly initiates from a TATA-less promoter at position +110 (
24
). Such TATA-less promoters depend on SP-1 for the recruitment of the transcription initiation complex (
25
). Additional evidence for an important role of SP-1 in t-PA transcription is provided by the study of Medcalf
et al
. (
9
) who reported a strict correlation between nuclear protein binding to GC-box III (i.e. SP-1) and t-PA expression in different cell-types: a high nuclear protein binding and t-PA expression in Bowes melanoma cells,
intermediate in HeLa cells and hardly detectable nuclear protein binding and no t-PA expression in HepG2 cells (
9
).
Our finding that the GC-box at +60 in the human t-PA promoter binds SP-1, is in contrast to reports suggesting that GC-box III is an AP-2 binding site. This suggestion, however, was
based on experiments which showed competition of GC-box III nuclear protein binding by a consensus AP-2 binding site (
9
). Since a consensus AP-2 binding site is also capable of binding SP-1 protein (Arts and Kooistra, unpublished data), these experiments are not directly indicative of AP-2 binding. Similarly, the GC-boxes in the murine t-PA promoter have also been reported to lack
affinity for AP-2 (
26
).
Our structural analysis of the human t-PA promoter extends previous studies by the identification of three
additional protein binding sites (CTF/NF-1 and GC-boxes I and II), which were previously not detected by
transfection and
in vitro
footprinting techniques (
9
,
27
,
28
). The exact role of these elements in human t-PA transcription remains unclear at present and needs to be established by
mutational analysis and transfection experiments. They are likely to be
important in t-PA expression, however, since deletion and/or mutation of these sites in
the very homologous murine t-PA promoter hampered t-PA transcription (
20
,
29
-
31
).
In conclusion, our studies on the identification of DNA-protein interactions at the t-PA promoter in intact human cells did not only confirm the
in vivo
presence of interactions previously detected
in vitro
, but also identified three additional protein binding sites. In addition, our
experiments showed the existence of subtle differences in DNA-protein
interactions between HUVEC and HeLa cells, which may be essential for
appropriate transcriptional control of t-PA in different physiological contexts.
We would like to thank Dr T. Oehler and Dr H. J. Rahmsdorf for the gift of anti-Jun and anti-Fos antisera, respectively, Dr C. Pfarr and Dr M. Yaniv for the gift
of of anti-FosB and anti-Fra2 antisera, and Dr W. van Driel for the gift of anti-CTF/NF1 antiserum. This study was supported by grants from the
Netherlands Heart Foundation (90.267 and 92.324) and from the Deutsche
Forschungsgemeinschaft (An 182/6-2).
*To whom correspondence should be addressed. Tel: +31 71 518 1450; Fax: +31 71
518 1904
REFERENCES
Return