Evidence for three major transcription activation elements in the proximal mouse
pro
[alpha]2(I) collagen promoter
Evidence for three major transcription activation elements in the proximal mouse pro [alpha]2(I) collagen promoter
Tadao
Hasegawa
,
Xin
Zhou
,
Lee Ann
Garrett
,
E. Cristy
Ruteshouser
,
Sankar N.
Maity
and
Benoit
de Crombrugghe*
Department of Molecular Genetics, The University of Texas M. D. Anderson Cancer
Center, 1515 Holcombe Boulevard,
Houston
, TX 77030,
USA
Received March 21, 1996;
Revised and Accepted June 7, 1996
ABSTRACT
In vivo
transient expression and
in vitro
transcription experiments indicated that a segment between -170 and -40 bp upstream of the start of transcription of the mouse pro
[alpha]
2(I) collagen gene was essential to activate transcription. DNase I protection
experiments identified three strong footprints in this segment. Experiments
with deletion mutants encompassing the sequences defined by these three
footprints indicated that each of the three elements contributed to the
transcriptional activity of the promoter. All three elements are GC-rich, redundant sites for a complex set of DNA binding proteins that
includes SP1, other proteins that bind to an SP1 consensus site and proteins
that bind to a Krox consensus site. In addition, the segment corresponding to
the most proximal footprint also binds the multimeric CCAAT binding protein
CBF. Addition of an excess amount of oligonucleotides corresponding to either
of the two distal footprints significantly inhibited
in vitro
transcription of the -350 bp pro
[alpha]
2(I) collagen promoter. Anti-SP1 antibodies that completely inhibited transcription of the early SV40
promoter had little effect on transcription of the wild-type -350 bp promoter, suggesting that SP1 has only a minor role in
activity of this promoter. Our results show that the segment between base pairs
-170 and -40 of the pro[alpha]
2(I) collagen promoter, which contains redundant binding sites for a complex set
of nuclear proteins, is essential in the transcriptional activity of this
promoter in fibroblasts.
INTRODUCTION
Expression of the genes for the pro[alpha]1(I) and pro[alpha]2(I) chains in type I collagen is coordinately regulated in a
variety of physiological and pathological situations (
1
). In intact animals and embryos, type I collagen is synthesized by a discrete
number of cell types, including osteoblasts, odontoblasts, fibroblasts,
mesenchymal cells and smooth muscle cells. It over-accumulates in fibrotic diseases such as lung fibrosis, cirrhosis and scleroderma. In many of these cases it is likely that control of pro[alpha]1(I) and pro[alpha]2(I) expression is likely exerted at the transcriptional
level, but the precise mechanisms involved during development and in adult
tissues are still poorly understood.
In earlier experiments we showed that a -2000 bp promoter of the mouse pro[alpha]2(I) collagen gene linked to reporter genes was sufficient for
tissue-specific expression in transgenic mice and mimicked the pattern of
endogenous gene expression during embryonic development. Similarly, a -350 bp promoter, clearly less active than the -2000 bp segment, showed an analogous pattern in transgenic mice,
but with practically no osteoblast expression (
2
-
4
). Our previous studies of the -350 bp promoter identified several functional
cis
-acting elements (
5
) and their cognate DNA binding proteins, including a CCAAT binding factor (CBF)
binding site between -95 and -75 bp (
6
-
8
), a CTF/NF1 binding site between -310 and -285 bp (
9
) and poorly characterized binding sites for unknown factors around -250 and -160 bp (
10
,
11
). Recent studies identified three short GC-rich segments in the human pro[alpha]2(I) gene between -323 and -264 bp (
12
), whereas other studies presented evidence that a complex of proteins binding
in this area participated in the TGF-[beta] response of this promoter (
13
).
This study, using deletion analysis, transient expression and
in vitro
transcription experiments, asked whether additional
cis
-acting elements were important for the activity of the -350 bp proximal mouse pro[alpha]2(I) collagen promoter and whether such elements were
binding DNA binding proteins.
MATERIALS AND METHODS
Cell culture and DNA transfections
714 Balb 3T3 fibroblasts were cultured in Dulbecco's modified Eagle's medium
containing 10% calf serum. Transient transfections were done as described
previously (
3
) except that SV[beta]gal (
14
) was used as internal control. Briefly, 714 Balb 3T3 fibroblasts were co-transfected with 15 [mu]g luciferase plasmid and 5 [mu]g SV[beta]gal plasmid (
15
). The cells were harvested after 48 h. Luciferase levels were measured using D-luciferin (Sigma) as substrate with a Monolight 2010 Luminometer
(Analytical Luminescence Laboratory). [beta]-Galactosidase activity was measured with a resorufin-[beta]-D-galactopyranoside substrate (Boehringer
Mannheim) at 30oC.
Reagents
Anti-SP1 antibodies were purchased from Santa Cruz Biotechnology Inc. and the
SP1 consensus oligonucleotide from Promega Corp.
DNA constructs
All wild-type and promoter deletion mutants were constructed in plasmid pA3LUC (
16
). In pH 6, pH 4 and pH 5, mouse pro[alpha]2(I) gene sequences -350 to +54 bp, -500 to +54 bp and -2000 to +54 bp respectively were cloned in the
Hin
dIII site of the vector. pH39, containing the minimal promoter -40 to +54 bp and 5'-deletion mutants of the -350 to +54 promoter, has been described previously (
3
). Internal deletion mutants of the -350 to +54, -500 to +54 and -2000 to +54 promoters were constructed by PCR (
17
,
18
).
In vitro
transcription assay
The assay was performed as described previously (
19
). All reactions contained 300 ng plasmid containing pro[alpha]2(I) promoters and 100 ng control plasmid (pLAG5, RSVLuc or pSV2CAT).
DNase I footprinting
The procedure was as described previously (
20
). Ten femtomoles of end-labeled
Bam
HI-
Nar
I fragments of the -350 to +7 pro[alpha]2(I) promoter were used. Nuclear extracts were prepared according
to standard procedures (
21
).
Gel retardation assays
One microliter of 714 Balb 3T3 fibroblast nuclear extract (5 [mu]g protein/[mu]l) was incubated with 5 fmol end-labeled double-stranded oligonucleotides in 10 [mu]l at room temperature for 20 min. All binding reactions contained 25 mM HEPES-NaOH, pH 7.9, 10% glycerol, 75 mM KCl, 0.25 mM
EDTA, 0.5 mM DTT, 0.5 mM PMSF, 0.05% NP-40 and 1 [mu]g poly(dA[middot]dT). The reactions were electrophoresed on 4-20% gradient polyacrylamide gels (Novex) in 0.5* TBE.
RESULTS
DNase I footprint experiments
Previous transient transfection experiments using a series of 5'-deletion mutants of the -350 bp promoter showed that a substantial decrease in
promoter activity occurred between -171 and -133 and also between -108 and -41 (
3
). To determine whether these active promoter sites interacted with DNA binding
proteins,
in vitro
DNase I footprints of the -350 bp proximal promoter of the mouse pro[alpha]2(I) collagen gene were performed using nuclear extracts of 714
Balb 3T3 fibroblasts. Three strong DNase I footprints from -176 to -152, from -131 to -114 and from -98 to -75 (a previously identified CBF
binding site) were identified (Fig.
1
A, lanes 3 and 7, and B, lanes 3 and 7). This CBF binding site contained a
characteristic CCAAT motif from -84 to -80 on the antisense strand. No other CBF binding site was
identified. Each of the three protected regions was clearly demonstrated on
both DNA strands. Weaker but reproducible footprints were detected between -285 and -258. A protected region around -300, previously identified as a CTF/NF1 binding site, was
also detected. The sequences around the CBF binding site, in the segments
around -120 and -160 and between -258 and -285, were GC-rich (Fig.
1
C). An excess of the double-stranded -105 to -65 oligonucleotide, containing the CBF binding site,
competed both the -98 to -75 and -131 to -114 footprints, suggesting that the GC-rich sequences near the CBF binding site might
compete with the -131 to -114 footprint. In contrast, the -105 to -65 oligonucleotide competed with the -176 to -152 footprint only very weakly (Fig.
1
A, lane 4, and B, lane 4). An oligonucleotide containing a consensus SP1 binding site competed with the -131 to -114 footprint and the weaker -285 to -258 footprint, only weakly with the -176 to -152 footprint and almost not at all
with the footprint over the CBF binding site (Fig.
1
A, lane 8, and B, lane 8). Hence, in the -350 bp mouse promoter there are several regions other than the CBF and
NF1 binding sites that are binding sites for other proteins. Some DNA binding
proteins that bound to the -131 to -114 region also bound to the region of the CBF binding site.
Proteins that bound an SP1 consensus site also bound to the -131 to -114 segment, to a lesser extent to the -176 to -152 segment and also to the GC-rich segment near the NF1 binding site.
DNA transfections with internal deletion mutants of the pro
[alpha]
2(I) collagen promoter
Several internal deletion mutants were generated. Surprisingly, the promoter
activity of a -108 to -40 deletion was as high in transient transfection experiments as the wild-type promoter (Fig.
2
A). This mutant is missing the CBF binding site and the DNA segment around it,
but not the TATA element and the transcription initiation site. In contrast, a
point mutation in the CCAAT motif (-84 to -80) significantly decreased promoter activity. Thus, the segment
upstream of -108 compensated for the deleted segment. However, when -130 to -40 was deleted, the promoter activity dropped to 14% of the
wild-type -350 bp promoter and -108 to -40 deletion mutant activities. Deletion of -170 to -40 reduced promoter activity further to
the level obtained with a minimal promoter extending from -40 to +54 (
3
). Deletion of -220 to -40 also reduced promoter activity to the level obtained with the -170 to -40 deletion. Other internal deletion mutants
retained the segment around the CBF binding site but lacked -130 to -110, -170 to -130 or -170 to -110. The first two deletion mutants
had 75 and 55% activities respectively, versus the wild-type -350 bp promoter; the third deletion mutant had ~25% activity. These results indicate that deletion mutants -170 to -110 and -130 to -110 are more active than deletion
mutants -170 to -40 and -130 to -40 respectively, hence providing evidence for a role of the -110 to -40 sequence in promoter activity.
In vitro
transcription of 5
'
-deletion mutants and internal deletion mutants of the -350 bp promoter
A series of 5'-deletion mutants were tested in
in vitro
transcription experiments using Balb 3T3 fibroblast nuclear extracts (Fig.
3
). We observed a marked decrease in promoter activity from -171 to -108 and from -108 to -40, in good agreement with the results of previous
transient transfection experiments using the same mutants (
3
). Both types of experiments suggested that active transcription factors bind
between -171 and -108 and to the region bound by CBF.
DNA binding experiments
To obtain additional information about the DNA binding proteins that interact with the -131 to -114 and -176 to -152 segments, gel shift experiments were performed. An
oligonucleotide from -140 to -86, including the GC-rich sequences immediately upstream of the CCAAT motif and
the -131 to -114 footprint, was used as probe with nuclear extracts of 714 Balb
3T3 fibroblasts. Three prominent DNA-protein complexes were observed (Fig.
5
A, lane 1, complexes 1-3), which were completely competed by sequences that included the
upstream footprint (-176 to -152), i.e. by oligonucleotide -180 to -136 (lane 5) and by the shorter oligonucleotide -176 to -152 (lane 6). Competition by
oligonucleotides -180 to -136 and -176 to -152 at equal molar ratios was clearly stronger than
competition by either the probe itself (lane 3) or by oligonucleotide -135 to -105 (lane 4). This indicated that the proteins which bind between -140 and -86 can also bind to the segment between -180 and -136 and that they can bind more
efficiently. The three complexes were also partially competed for by the -105 to -65 oligonucleotide (lane 2), containing the CBF binding site,
suggesting the existence within this segment of binding sites for proteins
different from CBF. These results agree with the DNase I protection experiments
in which oligonucleotide -105 to -65 competed for the -131 to -114 footprint. Because the -140 to -86 segment is very GC-rich, consensus oligonucleotides
for SP1 DNA binding and Krox DNA binding (
20
) were used in competition experiments. The SP1 oligonucleotide competed for all
three major complexes (lanes 8 and 11); the Krox oligonucleotide competed for
only complex 3 (lanes 9 and 12). The supershift of part of complex 1 produced
by anti-SP1 antibodies (see lane 10, complex 4) indicated the presence of SP1.
However, compared with the extensive competition by an SP1 consensus oligonucleotide, the SP1 antibodies mainly supershifted only
complex 1, with little effect on the other complexes. In summary, several
proteins apparently bound the promoter between -140 and -86, including SP1, a protein(s) different from SP1 that bound an
SP1 consensus oligonucleotide and a protein(s) that bound to a Krox consensus
oligonucleotide. Our competition experiments with two other promoter segments
also strongly suggested the existence of redundant binding sites for these
proteins.
Effects of oligonucleotides and SP1 antibodies during
in vitro
transcription
In vitro
transcription experiments were performed in the presence of an excess of
oligonucleotides -180 to -136 and the -140 to -86 to compete for binding of proteins interacting
with the promoter segments. An excess of oligonucleotide -180 to -136 inhibited transcription of the -350 bp promoter by 70% (Fig.
6
A, lane 5). An excess of oligonucleotide -140 to -86 decreased transcription by ~50% (lane 3). This agrees with the relative potency of each
of these oligonucleotides in competing for the DNA-protein complexes they form. The SP1 oligonucleotide inhibited
transcription by ~30% (lane 7). It is likely that the -180 to -136 oligonucleotide was able to remove from the template a
larger number of transcriptionally active DNA binding proteins than the SP1
oligonucleotide and the -140 to -86 oligonucleotide.
DISCUSSION
Transcriptional control of the type I collagen genes is a complex phenomenon
that likely involves both cell-specific and ubiquitous transcription factors. Here, we examined the
proximal promoter of the mouse pro[alpha]2(I) collagen gene for elements critical to transcriptional activation.
Our DNase I protection experiments identified three strong footprints in the
proximal promoter: -176 to -152, -131 to -114 and -98 to -75. All three segments are GC-rich and the third also contains
a CCAAT motif previously shown to bind CBF (
6
-
8
). The role of each of these three redundant elements in promoter activity was
examined by deletion analysis in transient expression experiments and by
in vitro
transcription experiments. Deleting only the most proximal element had no
effect, only the middle element, a relatively small effect and only the most
distal element, a more pronounced effect. Deletion of two of the three had a
much more pronounced effect, whereas an internal deletion from -170 to -40, removing all three footprints, resulted in very low activity,
about the same as for the basal -40 to +54 promoter. Hence, the -170 to -40 segment is essential for activated transcription above
the basal level. The results imply that potential transcription elements
upstream of -170 cannot activate transcription above the basal level in the absence of
-170 to -40. Nonetheless, each of these three proximal elements has an
important role in transcription activation. Indeed, deletion of either -170 to -130 or -130 to -110 alone decreased promoter activity. The -130 to -40 deletion mutant had much less
activity than the -110 to -40 mutant and the -170 to -40 deletion mutant completely lost its activity.
Thus both the -170 to -130 and -130 to -110 regions play important roles in promoter
activity. Although the -110 to -40 deletion mutant had essentially full promoter activity in transient transfection and
in vitro
transcription experiments, this segment itself contains an activating element,
since a point mutation in the CCAAT motif that abolished CBF binding caused a 2- to 4-fold decrease in promoter activity in our experiments. Furthermore,
the overall enhancing activity of the -110 to -40 element was also indicated by the much higher activity of the -170 to -110 deletion versus the -170 to -40 deletion mutant and of the -130 to -110 deletion mutant
versus the -130 to -40 deletion. It may be that removing the -110 to -40 sequence brings the other elements (from -170 to -110) and the transcription factors
that bind them closer to the general transcription factors anchored at the TATA
box of the promoter, thus providing more efficient interactions between the
proteins binding to these upstream sites and the general transcription factors
anchored at the TATA box. It may also be that between -110 and -40 lies a repressor binding site whose removal along with the CBF
binding site would result in a null effect. A potential inhibitory site has
previously been postulated to interact with a putative repressor present in an
SV40-transformed cell line but not an untransformed fibroblast line (
22
). We favor the former because, as our data indicate, the -110 to -40 element has an overall enhancing activity.
Several classes of proteins present in nuclear extracts bind to segments -180 to -136, -140 to -86 and -105 to -65 (see Fig.
7
for summary). These proteins include SP1, proteins different from SP1 that bind
an SP1 consensus binding site and could include other members of the SP1
family, such as SPR2, SPR3 and SPR4 (
23
), and proteins that bind to a Krox consensus binding site. All proteins that
bind the -140 to -86 segment and all proteins, except CBF, that bind the -105 to -65 segment can also bind the -180 to -136 sequence, however, the latter
segment appears to bind additional proteins that show a more complex pattern of
DNA binding in gel shift assays. Hence, redundant functionally active DNA
elements in the proximal pro[alpha]2(I) collagen promoter bind several classes of DNA binding proteins. Our
mutual competition experiments indicated that proteins that bind the -140 to -86 segment bind with higher efficiency to the -180 to -136 segment. Since many proteins can bind to the
redundant DNA GC-rich sequences, site-specific mutagenesis of the binding sites was not helpful in
identifying these proteins. Because their precise identification may require
cloning their cDNAs, we have recently cloned the cDNAs for two such proteins
from a mouse embryo fibroblast cDNA library using the yeast one hybrid system (
24
) in which the mouse pro[alpha]2(I) promoter sequence between -180 and -136 was used as bait. One cDNA encoded SPR2, an SP1 family
member, while the other encoded a new zinc finger protein (
25
).
ACKNOWLEDGEMENTS
We thank Sandra McKinney and Heidi Eberspaecher for excellent technical
assistance and Patricia McCauley for editorial assistance. This work was
supported by National Institute of Health grant HL41264 (to B. de C.) and the
Japan Foundation for Aging and Health (to T. H.).
REFERENCES
1 Vuorio,E. and de Crombrugghe,B. (1990) Annu. Rev. Biochem., 59, 837-872.MEDLINE Abstract
2 Khillan,J.S., Schmidt,A., Overbeek,P.A., de Crombrugghe,B. and Westphal,H. (1986) Proc. Natl. Acad. Sci. USA, 83, 725-729.MEDLINE Abstract
3 Goldberg,H., Helaakoski,T., Garrett,L.-A., Karsenty,G., Pellegrino,A., Lozano,G., Maity,S. and de Crombrugghe,B. (1992) J. Biol. Chem., 267, 19622-19630.MEDLINE Abstract
4 Niederreither,K., D'Souza,R.N. and de Crombrugghe,B. (1992) J. Cell. Biol., 119, 1361-1370.MEDLINE Abstract
5 Schmidt,A., Rossi,P. and de Crombrugghe,B. (1986) Mol. Cell. Biol., 6, 347-354.
6 Hatamochi,A., Paterson,B. and de Crombrugghe,B. (1986) J. Biol. Chem., 261, 11310-11314.MEDLINE Abstract
7 Hatamochi,A., Golumbek,P.T., Van Schaftingen,E. and de Crombrugghe,B. (1988) J. Biol. Chem., 263, 5940-5947.
8 Maity,S.N., Golumbek,P.T., Karsenty,G. and de Crombrugghe,B. (1988) Science, 241, 582-585.MEDLINE Abstract
9 Oikarinen,J., Hatamochi,A. and de Crombrugghe,B. (1987) J. Biol. Chem., 262, 11064-11070.MEDLINE Abstract
10 Karsenty,G., Golumbek,P. and de Crombrugghe,B. (1988) J. Biol. Chem., 263, 13909-13915.MEDLINE Abstract
11 Karsenty,G. and de Crombrugghe,B. (1991) Biochem. Biophys. Res. Commun., 177, 538-544.MEDLINE Abstract
12 Tamaki,T., Ohnishi,K., Hart,L.C., LeRoy,E.C. and Trojanowska,M. (1995) J. Biol. Chem., 270, 4299-4304.MEDLINE Abstract
13 Inagaki,Y., Truter,S. and Ramirez,F. (1994) J. Biol. Chem., 269, 14828-14834.MEDLINE Abstract
30 Chen,S.S., Ruteshouser,E.C., Maity,S.N. and de Crombrugghe,B. (1996) submitted.
31 Bou-Gharios,G., Garrett,L.A., Rossert,J., Niederreither,K., Eberspaecher,H., Smith,C., Black,C. and de Crombrugghe,B. (1996) J. Cell Biol., in press.
