Comparative and functional analysis of the AP2 promoter indicates that conserved
octamer and initiator elements are critical for activity
Comparative and functional analysis of the AP2 promoter indicates that conserved octamer and initiator elements are critical for activity
Peter C.
Creaser
,
David A.
D'Argenio
and
Trevor
Williams*
Department of Biology, Yale University, PO Box 208103,
New Haven
, CT 06520-8103,
USA
Received February 5, 1996;
Revised and Accepted May 17, 1996
GenBank accession nos X95234-X95236 (incl.)
ABSTRACT
AP-2 is a developmentally-regulated transcription factor expressed in ectodermal cell
lineages. The AP-2 protein is essential for neural tube, craniofacial and body wall
morphogenesis and has been implicated in oncogenesis. Here we report the
isolation of the AP-2 promoter from human, mouse and chicken. The initiation sites for the
human gene have been mapped in a variety of cell lines, including several
derived from breast tumours. Initiation occurs just upstream of an IR3-like repetitive element, present in the human and mouse genes, but absent
in chicken. The
cis-
acting elements responsible for promoter activity in human HeLa cells have been
mapped both
in vivo
and
in vitro
. The proximal promoter contains binding sites for transcription factors AP-2, NF-1 and octamer proteins, but lacks a TATA box motif. Functional
analysis demonstrates that the octamer binding site is the critical component
of basal promoter activity. In addition, the promoter relies on an initiator
element for efficient start site utilization. There is an excellent correlation
between the requirement for the initiator and octamer elements in transcription
assays and the conservation of these
cis-
acting sequences between chicken, mouse and human.
INTRODUCTION
The transcription factor AP-2 is a critical regulatory molecule required for vertebrate development.
The AP-2 protein binds as a homodimer to the consensus recognition sequence
GCCNNNGGC that is an important
cis-
regulatory element for a variety of cellular genes and viral genomes, including
keratin, proenkephalin and MMTV (
1
-
3
). In addition, the pattern of AP-2 expression is regulated both temporally and spatially during mouse embryogenesis beginning around day E8.5 (
4
). The majority of AP-2 RNA and protein are found in ectodermal cell lineages, including the
neurepithelium and neural crest (
4
-
6
). The tissue-specific distribution of AP-2 correlates with the differential expression of the gene in various
cell lines. For example, AP-2 mRNA and protein are present in epithelial-derived HeLa cells, but absent from HepG2 hepatoma cells (
7
). Furthermore, AP-2 expression is induced in teratocarcinoma cell lines undergoing
differentiation in response to the morphogen retinoic acid (
6
-
9
). The importance of this transcription factor for normal mammalian embryogenesis was recently revealed by generating mice containing a homozygous disruption of the AP-2 gene. These AP-2-null mice die at birth of severe developmental abnormalities
including exencephaly, craniofacial defects and a failure of ventral body wall
closure (
10
) .
The AP-2 protein is also associated with cellular transformation. In PA1 human teratocarcinoma cells, the addition of an AP-2 expression vector leads to anchorage independent growth (
11
). Furthermore, the AP-2 protein is highly expressed in a particular class of human breast cancer
cell lines. Approximately 30% of human breast cancers contain high levels of c-erbB2, a tyrosine kinase receptor that causes mammary carcinoma in animal
model systems (
12
-
14
). The region of the c-erbB2 promoter responsible for increased expression in human breast cancer
cell lines has been mapped (
15
) and identified as an AP-2 responsive element (
16
). Moreover, there is a correlation between the presence of AP-2 and elevated levels of c-erbB2 protein in human breast cancer cell lines, strongly suggesting
that AP-2 may be directly responsible for c-erbB2 overexpression (
16
). Therefore, the mechanism of regulation of the transcription factor AP-2 may provide an important key to understanding the aetiology of mammary
carcinoma. In this study we have begun to address this question by
characterizing the AP-2 promoter from three vertebrate species and identifying the
cis-
regulatory elements critical for its basal level of expression.
MATERIALS AND METHODS
Isolation and sequencing of genomic clones
An
Eco
RI-
Nco
I fragment of the human AP-2 cDNA spanning from nucleotides (nt) 1-1289 (
7
) was used to probe a human male circulating lymphocyte genomic cosmid library
and a mouse C57 Black/6 female liver genomic cosmid library (Stratagene).
Restriction fragments surrounding the 5'-end of the AP-2 cDNA were subcloned from positive cosmids and sequenced on
both strands. The chicken promoter sequence was obtained by PCR from genomic
DNA (gift of H. Belting) using primers based on mouse/human homology. Flanking
sequence was then obtained using inverse PCR as follows. Genomic DNA was
digested with the restriction enzymes
Nla
III,
Taq
I, or
Pst
I plus
Nsi
I, followed by religation at low DNA concentration. PCR was then performed using the primers CK6 (GGATC GGACC CTCTC CCGCC GACCC) and CK7 (GCTTT
ACCCG CAGCC GGAGC GCCTC ATTAG C). The sequence was obtained directly by cycle sequencing (Perkin
Elmer) and was identical for all three enzyme digests.
Plasmid constructs
The plasmids designated +275 were made by fusing the
Sau
3AI site at +275 of the human genomic sequence, located 8 nt upstream of the
translational initiator, to the
Bgl
II site of pBLCAT3[Delta] (
8
,
17
). The plasmids designated +37 were generated by fusing an
Ecl
136II site at +37 of the human genomic sequences to the T4 DNA polymerase
repaired
Xho
I site in pBLCAT3[Delta]. The extent of upstream sequences is as follows: -500 is an
Xho
I partial digest and -190 is an
Xho
I digest into the
Sal
I site of pBLCAT3[Delta]; -151 is an
Xba
I digest into the
Xba
I site of pBLCAT3[Delta]; -122 and -99 are
Stu
I and
Hae
III digests into the T4 DNA polymerase repaired
Sal
I site of pBLCAT3[Delta]; -82, -49 and -10 were made using PCR and introduce a
Sal
I site that was ligated into the
Sal
I site of pBLCAT3[Delta]. The relevant sequences at the junction sites are; gtc gac TAC TGG CGA (-82), gtc gac tTA ATG AGG (-49); gtc gac GGA AAA GTT (-10). The [Delta]oct, [Delta]ctf and LS templates were made in the -151 backbone using PCR or
direct insertion of double-stranded oligonucleotides. All constructs were confirmed by sequencing.
In vitro
transcription reactions
HeLa nuclear extract was prepared according to Dignam (
18
) as modified by Wildeman (
19
). Each
in vitro
transcription reaction (20 [mu]l) contained 4 [mu]l nuclear extract in 10 mM HEPES-KOH (pH 7.9), 0.5 mM DTT, 0.1 mM EDTA, 1 mM MgCl
2
, 10 mM KCl, 10% glycerol, 4 mM spermidine, 0.5 mM each rNTP. Where appropriate,
each reaction contained 200 ng template DNA linearized with
Nco
I. [alpha]-Amanitin (5 [mu]g/ml) was included where indicated. Reactions proceeded for 60
min at 30oC and were then analyzed by S1 nuclease mapping or primer extension. All
transcription reactions were performed independently at least twice. Primer
extension analysis was conducted using a kinase labelled 24 nt CAT primer 5'- GCCAT TGGGA TATAT CAACG GTGG-3' which was annealed in water to the transcription
products at 37oC for 10-20 min. Primer extension reactions proceeded for 60 min at 37oC in 1* buffer (Gibco BRL), 10 mM DTT, 0.5 mM each dNTP, 20 [mu]g/ml actinomycin D, 0.5 U/[mu]l RNasin (Promega) and 2.4 U/[mu]l superscript II reverse transcriptase (Gibco BRL).
S1 nuclease mapping
Total RNA was prepared by the guanidine isothiocyanate procedure followed by
centrifugation through a caesium step gradient (
20
). RNA from
in vitro
transcription reactions was isolated as above. S1 probe used for mapping either
endogenous transcripts, or RNA generated by
in vitro
transcription of templates extending to +275, was derived from the -49/+275 plasmid. This plasmid was kinase labelled at an
Ngo
MI site at +195, followed by an
Sph
I restriction digest, which cuts in the pBLCAT3 polylinker sequences upstream of
the AP-2 insert. The S1 probe used for mapping transcripts derived by
in vitro
transcription of templates extending to +37 was generated from the -190/+37 plasmid. This plasmid was kinase labelled at a
Pvu
II site in the CAT gene, followed by a second digest with
Sph
I as above. In both instances, the strands were then separated on a standard
sequencing gel. Hybridization between probe and RNA was performed at 42oC for at least 8 h in 50% formamide, 40 mM Pipes pH 6.4, 1 mM EDTA and 0.4
M NaCl before S1 nuclease digestion at 37oC for 30 min in the absence of carrier DNA (
20
).
DNase I footprinting
The plasmid used for footprinting was the human genomic clone from the
Xho
I site at -190 to a
Bam
HI site in the transcribed region of the gene, cloned into pBluescript II
(Stratagene). This plasmid was kinase labelled at the
Acc
65I site in the vector polylinker, then restricted with
Ecl
136II (+37) and the footprinting probe subsequently purified by PAGE. DNAse I footprinting with HeLa nuclear
extract and purified HeLa AP-2 or CTF protein was as described previously (
7
).
EMSA
HeLa nuclear extracts were made according to the procedure of Hurst (
21
). Assays were performed as described previously (
3
). Where appropriate, the reactions were incubated with either the competitor oligonucleotides for 10 min, or with 1 [mu]l of an anti-oct1 or anti-oct2 antiserum (Santa Cruz Biotechnology) for 60 min on ice,
before addition of the kinase labelled AP octamer probe.
Transfections
Co-transfections were performed by the calcium phosphate method as previously
described (
3
) using 18 [mu]g of AP-2 /+37 CAT3[Delta] reporter construct and 12 [mu]g pUC118 DNA as carrier. Transfections were performed at
least twice in duplicate, and the extracts normalized for protein concentration prior to the CAT assay.
RESULTS
Identification of the AP-2 promoter
The human AP-2 coding region was used to isolate four independent human genomic cosmid
clones, and the sequence encompassing the 5' boundary of the AP-2 cDNAs was then determined (shown diagrammatically in Fig.
1
; see also Fig.
10
). This analysis revealed the presence of an ~140 nt copy of a CT-rich repeat sequence in the 5' non-coding region of the AP-2 transcript. This repeat, based on the triplet
TCC, is a member of the IR3-like repetitive element family that was initially described in the genome
of EBV, but has since also been identified in a number of cellular genes (
22
). It might be expected that this repetitive element could complicate the
analysis of the AP-2 promoter. First, repetitive elements often harbour internal promoter elements and so it could function as an integral part of the AP-2 upstream regulatory region. Secondly, the presence of an IR3-like repeat would make it more difficult to analyze
transcription of the gene by techniques such as RNase protection because of the
presence of other cellular transcripts containing related sequences. Therefore, to determine which region corresponded to the AP-2 promoter, S1 nuclease mapping was utilized to locate the 5'-end of the AP-2 mRNA from human HeLa cells. In contrast to a
uniformly labelled probe, typical of an RNase protection, any of the S1 probe
annealing to related repeats in other transcripts will have the label present in the unique sequence removed by nuclease digestion.
Thus, the spurious IR3 hybrids will be eliminated from the analysis and only genuine AP-2 transcripts will be observed.
Characterization of AP-2 promoter activity in cell-free extracts
The functional organization of the AP-2 promoter was further analyzed by using a series of 5' deletion mutants as templates for
in vitro
transcription. Figure
4
lane 2 shows the basal amount of endogenous AP-2 RNA present in the HeLa nuclear extract used for the transcription
reactions in the absence of any added template DNA. Inclusion of the longest template tested, -190/+275, caused a marked increase in the 185-195 nt products derived from the AP-2 promoter (Fig.
4
; compare lanes 2 and 8). These AP-2 transcripts are sensitive to the low concentrations of [alpha]-amanitin that specifically inhibit RNA polymerase II (compare
lanes 1 and 8). In contrast, the 244 nt products initiating upstream of the AP-2 sequences are not eliminated by this concentration of [alpha]-amanitin, indicating that they are products of other RNA polymerases. Successive deletion of upstream sequences
between -190 and -82 had no effect on the amount of
in vitro
transcription observed (Fig.
4
, lanes 4-8). However, deletion to -49 caused a severe reduction of transcription (Fig.
4
, lane 3). These experiments map the major determinants for
in vitro
transcription of the AP-2 promoter downstream of -82.
Identification of proteins interacting with the AP-2 promoter
DNase I footprinting assays were performed using the nuclear extract employed in
the
in vitro
transcription studies to identify proteins that interacted with the basal
promoter region (Fig.
6
). The most noticeable footprint occurred between nucleotides -39 and -56, which maps to the critical region of the promoter identified
in the transcription assays (Fig.
6
, left). This region conforms strongly to the octamer consensus sequence,
differing by only 1 nt from the canonical binding site (
23
).
The activity of the AP-2 promoter relies on octamer and initiator elements
Next, a series of specific point mutations was constructed that disrupted
discrete
cis-
acting elements in the context of the intact promoter. In light of the DNA
binding data, the marked loss of activity seen
in vitro
with the -49 template (Figs
4
and
5
) might result from mutation of either the CTF/NF-1 site or the octamer motif. In the -49 template, the CTF/NF-1 site is completely removed and the octamer sequence
GATATGCTAATGA is replaced with GTCGACTTAATGA by the juxtaposition of upstream vector sequences. Therefore, to determine the importance of the octamer and CTF/NF-1 sites in the normal context of surrounding AP-2 sequences, more specific substitution mutations of these sites were engineered into the -151 template. These constructs were then used for
in vitro
transcription assays and the products analyzed by S1 nuclease mapping. As shown in Figure
8
, the alteration of the CTF/NF-1 site showed no decrease in the level of transcription. In contrast, the
disruption of the octamer site at -49 severely diminished transcription of the normally active -151 promoter template
in vitro
(Fig.
8
, [Delta]oct), demonstrating that the oct sequence is a critical component of the
AP-2 promoter.
In vivo
activity of the AP-2 promoter
To confirm and extend the results obtained with
in vitro
transcription, the AP-2 promoter constructs were transfected into HeLa cells. Varying lengths of
the AP-2 promoter extending to +37 were fused to the CAT gene in the pBLCAT3[Delta] background. Figure
9
shows that when the AP-2 promoter spanning from -151/+37 is inserted into this vector there is an ~5-fold increase in the level of CAT activity relative to
pBLCAT3[Delta] alone. Analysis of the reporter gene transcripts from the transfected
cells indicated that they utilized the correct initiation sites (data not
shown). The further addition of upstream sequences to the -151 template, up to -5 kb, did not cause any significant increase in the level of CAT
expression. Similarly, deletion of sequences to -82 does not affect relative expression levels (data not shown).
Therefore, the AP-2 sites are not required for basal activity, in agreement with the data
obtained in cell free extracts. In contrast, mutation of the octamer sequence abolished expression of the CAT gene (Fig.
9
). Therefore, the upstream sequences that appear to be critical for
in vitro
transcription of the AP-2 gene also appear to be the most important in the HeLa co-transfection assay.
Comparison of AP-2 promoter sequences between species
We compared the sequence of the AP-2 promoter from different vertebrate species to determine which
cis-
acting elements were conserved. The human AP-2 coding region was used to isolate a mouse genomic cosmid clone. Subsequently, sequences conserved between these two species were used to design primers to isolate the
chicken genomic sequence. A comparison between the human, mouse and chicken
data is shown in Figure
10
.
Both the human and mouse 5' untranslated regions contain an IR3-like repeat, but this element is absent from the chicken. Upstream
of this repetitive element there is a high degree of sequence identity between all three species, especially surrounding the initiation site and octamer element. Just upstream from the CTF/NF-1 site the homology between human and mouse is still very high; the
proximal AP-2 site is identical in human and mouse, but is less conserved in chicken.
Further upstream from this site, the degree of sequence identity between mouse
and human is greatly reduced (data not shown). These comparative data indicate
that the
cis-
acting DNA sequences critical for transcription are highly conserved between
these three species. Therefore, this comparison strongly supports the role of
the octamer and initiator elements as fundamental control sequences for the AP-2 promoter.
DISCUSSION
In this report we have isolated and compared the AP-2 promoter from human, mouse and chicken species. In addition, the
critical elements for AP-2 expression in HeLa cells have been identified and characterized in cell
free extracts or in the context of intact cells. There is a high degree of
conservation between the promoter sequences, especially in the vicinity of the
start site. Both human and mouse promoters contain a CT-rich IR3-like repetitive element (
6
,
22
,
24
) between the transcriptional start site and the translational initiation site.
The conservation of the large IR3-like element between these two species suggests that it was inserted into
the transcribed region of AP-2 before the divergence of these organisms and raises the possibility that
it may perform some function for this transcription unit. However, the IR3-like repeat is absent from the chicken AP-2 gene. Furthermore, this repeat is not necessary for AP-2 promoter function as it can be deleted from the human gene
without affecting initiation
in vitro
and we find that no transcripts originate from within this element either
in vivo
or
in vitro
. Alternatively, this repetitive element, which will be present in the 5' untranslated region of the AP-2 RNA, could be important for regulating message stability or
translational efficiency.
Upstream of the IR3-like repeat the promoters display a high degree of sequence identity,
especially around the start site and octamer motif. None of the promoters
contain a recognisable TATA box sequence upstream of the initiation site.
Furthermore, AP-2 does not resemble TATA-less GC-rich promoters that use a number of dispersed start sites (
25
). Instead, AP-2 is similar to promoters that do not contain an obvious TATA-box or Sp1 binding sites, yet initiate at only a few discrete sites
(
25
-
30
). These promoters, typified by the terminal deoxynucleoside transferase gene,
contain a
cis-
acting initiator element (Inr) centered around the start site which is important
for both the accuracy and level of transcription initiation (
25
,
30
). Similarly, mutation of sequences encompassing the AP-2 start site leads to a loss of activity, indicating that this sequence is
an integral component of the promoter and not simply a passive recipient of
upstream information. Indeed, the region between +1 and +10, CTACCATTAG, has
limited homology to the consensus initiator sequence PyPyPyPyCANTPyPy which
acts as a binding site for the TFIID complex (
25
,
31
-
33
). The AP-2 start site differs from the classical Inr element in that AP-2 transcription does not begin at the C of the CANT consensus but at
a C residue 4 nt upstream. However, the potential Inr in the AP-2 promoter is consistent with previous experiments in which initiator
elements can direct polymerase to commence transcription at adjacent
nucleotides (
28
,
34
). In the future, a detailed analysis of this region of the AP-2 promoter in a heterologous context will help to define its potential
function as an initiator element.
ACKNOWLEDGEMENTS
We are grateful to Zhiling Jiang for the synthesis of oligonucleotides and Helen Hurst for RNA samples. We thank Rick Austin, Mark Biggin, Adrian
Hayday, Helen Hurst, Bill Segraves, Timothy Nottoli and Jian Zhang for critical
reading of the manuscript and encouragement throughout. We are also indebted to
Steve Smale for helpful discussion. This work was funded in part by grant GM
46770 from the National Institutes of Health. T.W. is a Pew Scholar in the
Biomedical Sciences. The accession numbers for the chicken, human and mouse
promoter sequences are X95234, X95235 and X95236, respectively.
REFERENCES
1 Mitchell,P.J., Wang,C. and Tjian,R. (1987) Cell, 50, 847-861.MEDLINE Abstract
2 Imagawa,M., Chiu,R. and Karin,M. (1987) Cell, 51, 251-260.MEDLINE Abstract
3 Williams,T. and Tjian,R. (1991) Genes Dev., 5, 670-682.MEDLINE Abstract
20 Ausubel,F.M., Brent,R., Kingston,R.E., Moore,D.D., Seidman,J.G., Smith,J.A. and Struhl,K. (1989) Current Protocols in Molecular Biology. Greene Publishing Associates and Wiley-Interscience.
21 Hurst,H.C. (1991) In Collins,M. (eds) Methods in Molecular Biology, Humana Presss, NY, pp. 141-151.
23 Herr,W. (1992) Oct-1 and Oct-2, In McKnight,S.L. and Yamamoto,K.R. (eds) Transcriptional Regulation, 2. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 1103-1135.
