(Received for publication, September 13, 1996, and in revised form, October 25, 1996)
From the Renal Division, Beth Israel Hospital and Harvard Medical School, Boston, Massachusetts 02215
We performed deletion analysis of
WT1-reporter constructs containing up to 24 kilobases of
5-flanking and first intron WT1 sequence in stably
transfected cultured cells as an unbiased approach to identify cis
elements critical for WT1 transcription. Although not a
tissue-specific element, a proximate 9-base pair CTC repeat accounted
for ~80% of WT1 transcription in this assay. Enhancer activity of the element and mutated versions correlated completely with
their ability to form a DNA-protein complex in gel shifts. Antibody
supershift, oligonucleotide competition, and Southwestern studies
indicated that the CTC-binding factor is the transcriptional activator
Sp1. Sp1 binds the CTC repeat with an affinity, KD = 0.37 nM, at least as high as the consensus GC box.
Similar CTC repeats are found in promoters of other growth-related
genes. Because Sp1 is important for WT1 expression, we
examined Sp1 immunohistochemistry in fetal and adult kidney. In a
pattern that precedes that of WT1 message, Sp1
immunostaining was highest in uninduced mesenchyme, early tubules,
developing podocytes, and mature glomeruli, but was minimal in mature
proximal tubules. This work suggests abundant Sp1 may be a prerequisite
for WT1 expression, and that Sp1 may have a wider role in
nephrogenesis.
The Wilms' tumor-1 (WT1) gene is essential for development of the genitourinary tract. Homozygous germline WT1 disruption in mice causes renal agenesis and gonadal failure (1). The knockout mice also have hypoplastic hearts and lungs, likely due to failed interaction with the mesothelium (1), which normally expresses WT1 (2). In humans, germline single-allele WT1 defects cause urogenital abnormalities and strongly predispose to Wilms' tumor. Male genital ambiguity and a nephropathy that progresses to renal failure are frequent features of the Denys-Drash syndrome (3-5), due to single-allele WT1 point mutations (6-8). Less profound genitourinary malformations, such as cryptorchidism and hypospadias, are associated with aniridia and Wilms' tumor (9-11), which with mental retardation comprise the WAGR syndrome, and result from one WT1 allele loss in a chromosome 11p13 deletion (12). Both syndromes are also characterized by persistent nodules of embryonic kidney tissue, called nephrogenic rests, from which malignant Wilms' tumors may arise (13), and their histopathology is suggestive of aberrant kidney development. The WT1 gene encodes a putative tumor suppressor (14), and biallelic intragenic WT1 defects have been found in a small percentage of Wilms' tumors (15). The WT1 protein is a zinc finger transcription factor (16-18) that in most contexts acts as a transcriptional repressor (Ref. 19 and below). Many candidate WT1 target genes have been identified, such as insulin-like growth factor 2 (igf2) (20), platelet-derived growth factor (pdgf) A-chain (21, 22), igf 1 receptor (23), epidermal growth factor receptor (egfr) (24), and Pax2 (25), among others. An attractive hypothesis for Wilms' tumor pathogenesis is that loss of the negative regulator WT1 results in overexpression of one or more growth-promoting target genes (reviewed in Ref. 26).
Little is known about regulation of the WT1 gene, but part appears to be transcriptional. The WIT1 gene is adjacent to WT1 and transcribed in the opposite direction from a shared ~2-kb1 intergenic region (27). Because WIT1 and WT1 are often coexpressed (27), shared cis elements may be important in their regulation. Furthermore, nuclear run-on studies from our laboratory in WT1-expressing and non-expressing cell lines support that a major component of the gene's regulation is transcriptional.2
WT1 is tightly regulated. WT1 message is tissue-restricted and expressed in a striking pattern in nephrogenesis (29). Little WT1 message is detected in undifferentiated metanephric mesenchyme (30-32), but levels increase prominently in the structures formed as the mesenchyme condenses, the renal vesicle, comma-shaped, and S-shaped body, where expression becomes localized to the developing glomerular podocytes (29). Throughout nephrogenesis, WT1 protein is expressed coordinately with WT1 message, supporting that WT1 regulation is not translational (32, 33). In addition, the WAGR genitourinary defects are most likely due to reduced WT1 expression from haplotype insufficiency. The above observations suggest that WT1 message levels are important in development and that transcriptional control of the WT1 gene is likely.
WT1 is transcribed from a 5-flanking region that is ~70%
GC-rich and lacks a consensus TATA box (34, 35). Multiple
WT1 transcription start sites (TSS) have been identified
over a ~300-bp promoter region (35-38). These data indicate that the
human WT1 major TSS is in the region 393-413 bp upstream of
the initial ATG (35, 37) (Fig. 1B, dotted
overline), and this TSS is conserved in the mouse gene (36). While
no consensus initiator element resides in this region, a candidate TATA
box (CTTATTTGA) (Fig. 1, A and B) that resembles
the functional SV40 early promoter TATA box (CTTATTTAT) (39) is well
positioned upstream of the major TSS and may direct initiation. Other
regulatory regions of the WT1 promoter remain
unexplored.
We report here the identification and characterization of a critical WT1 promoter enhancer element and the corresponding transcriptional activator.
A human lung fibroblast (WI38) genomic
library in bacteriophage FIX (Stratagene) was screened using two
probes to obtain genomic WT1 sequence. The first probe was
5
human WT1 cDNA sequence (gift of K. Call, Harvard
School of Public Health), a 1.8-kb EcoRI-EcoRI fragment, and the second was murine WT1 5
-flanking region
(bp
593 to +200, relative to the major TSS; Ref. 36) generated by
polymerase chain reaction. One clone screening positive with both
probes contained 9.4 kb of upstream and 2.4 kb of downstream sequence
(designated here
9.4/+2.4 kb) with respect to the WT1 major TSS. The clone was sequenced over ~1.5 kb of the proximate promoter using a deazaG and deazaA chain-termination sequencing kit
(Pharmacia Biotech Inc.). The 11.8-kb WT1 genomic fragment was cloned into pBluescript II SK+ (Stratagene), which was converted into a reporter construct by placing a blunted
XbaI-BamHI fragment from pCAT-Basic (Promega),
containing the chloramphenicol acetyltransferase (CAT) gene and SV40 t
intron and poly(A) addition site, into a unique BbrPI site
in the WT1 clone 5
-untranslated region at position +206. A
still larger
22/+2.4 kb WT1-reporter construct was made by
placing a
2.5/+2.4 kb BamHI fragment from the
9.4/+2.4
kb construct into cosmid pWE15 (Stratagene) and inserting upstream a
19.4-kb WT1 BamHI fragment (
22/
2.5 kb) from
WT1 cosmid L156 (16) (gift of D. Haber, Massachusetts
General Hospital). Serial deletions were made from the
9.4/+2.4 kb
construct using restriction sites SnaBI (
4.3 kb) and
BamHI (
2.5 kb). In addition, a WT1 PstI-PstI fragment (
2.4 kb/+191 bp) was cloned into
pCAT-Basic to yield a reporter construct with the CAT gene in a
location similar to the larger constructs. In the latter case, fine
deletions were made using polylinker and WT1 restriction
sites, yielding 5
deletions at the following locations:
SacI (
1.1 kb), HindIII (
453 bp),
BpmI (
319 bp), AvaII (
168 bp),
BssHII (
122 bp), EagI (
81 bp), and
NaeI (
48 bp). Series of
81/+191 bp linker scanning
reporter constructs were made by inserting EagI(
86
bp)-NgoMI(
50 bp) sticky-ended double-stranded
oligonucleotides (see below) (Oligos Etc.) into a HindIII-
and NgoMI-cut
453/+191 bp vector, blunting with T4
polymerase, and ligating. The
453/+191 bp linker scanning constructs
were made by placing the EagI-NgoMI oligonucleotides into an
EagI- and NgoMI-cut
453/+191 bp vector. To make
the minimal promoter linker scanning constructs,
EagI-XbaI sticky-ended double-stranded oligonucleotides (see below) were cloned into the XbaI site
in E1b TATA-CAT vector (40) such that the final constructs contained from
81 to
39 of WT1 sequence. The
81/+191 bp,
48/+191 bp, and all linker scanning constructs were sequenced to
confirm their authenticity.
293, HeLa, HeLa S3, and
Drosophila Schneider S2 cells were obtained from ATCC. JMN
cells (41) were the gift of J. Rheinwald, Brigham and Women's
Hospital. Human cells were maintained in Dulbecco's modified Eagle's
medium with 10% fetal bovine serum (HyClone) and Schneider cells in
Schneider medium with 10% heat-inactivated serum
(Sigma). Cells were transfected by calcium-phosphate
precipitation with a total of 20 µg of plasmid DNA/100-mm dish.
Transient transfections included 20-50% reporter plasmid, 1-5%
transfection control plasmid (CMV-driven -galactosidase), and the
remainder pBluescript II SK+, while stable transfections comprised 95%
reporter construct and 5% neo resistance gene vector
(pSV2-neo). Duplicate tissue culture dishes for each construct were
transfected in each experiment. Transiently transfected cells were
harvested at 48 h, and stably transfected cells were passaged from
100- to 150-mm dishes into medium containing 100-600 µg/ml G418
(Life Technologies, Inc.) at 48 or 72 h (42). Stably transfected
cells were maintained in culture under selection for 2-4 weeks, until
colonies were obvious and non-resistant cells had died. Stable
transfections were considered valid only if >100 colonies of roughly
equal size resulted, to negate integration position and copy number
effects. CAT assays were then performed on the entire pool of stable
colonies and normalized for protein concentration by Bradford assay
(Bio-Rad). Percent chloramphenicol acetylation was determined by
thin-layer chromatography and scintillation counting of
autoradiographed spots.
Nuclear extracts were prepared from cultured cells (43) and from tissue, with modifications (44). For further purification, HeLa S3 nuclear extracts were prepared according to modifications (45) of another standard protocol (46). HeLa S3 crude nuclear extract, ~150 mg of protein from 10 liters of culture (Cell Culture Center), was partially purified over a 2-ml wheat germ agglutinin-agarose affinity column (Vector Laboratories) (45).
EMSA reactions (25 µl) contained 20 mM HEPES, pH 8.4, 100 mM KCl, 20% glycerol, 0.1 mM EDTA, 0.25 mM ZnSO4, 0.05% Nonidet P-40. Thirty ng to 2 µg of protein and an equal amount of poly(dA·dT) (dA·dT) were
added at room temperature 10 min prior to addition of 0.1-7 ng of
radiolabeled oligonucleotide probe, after which room temperature
incubation continued for 18 min. One µg of acetylated bovine serum
albumin was used as carrier with purified protein. Four or 5%
acrylamide gels were run in 1 × TAE buffer (40 mM
Tris acetate, pH 8.5, 2 mM EDTA) at room temp at 160 V. For
quantitation of the EMSA bands and KD determination,
a constant amount of extract was incubated with an increasing amount of
probe until saturation of factor binding occurred (42). The fraction of probe in a protein-DNA complex was determined by Cerenkov counting of
excised, autoradiographed bound and free EMSA bands, and the amount of
protein by assuming a protein:DNA molar ratio of 1. The 31-bp
WT1 wild-type enhancer or 26-bp Sp1 consensus
oligonucleotides were used to determine KD values,
which were calculated as slope
1 in Rosenthal plots
(47). For EMSA supershift studies, 1-2 µg of monoclonal or
polyclonal Sp1 antibody (Santa Cruz Biotechnology) was added for 40 min
at room temperature following standard binding reaction above.
Synthesized oligonucleotides (Oligos Etc.) are shown below. The two
series of 45-bp WT1 6-bp and 2-bp linker scanning
oligonucleotides were designed with EagI-XbaI
ends. The EagI site is found in human WT1, but
the XbaI site was created by changing the first 2 nucleotides of the lower strands. The full-length 45-bp
oligonucleotides were used for the E1b TATA constructs only. For the
WT1 context constructs and for EMSAs, the 45-bp
oligonucleotides were inserted into pBluescript II SK+, excised with
EagI and NgoMI, and gel-purified to yield 36-bp
WT1 oligonucleotides (see Fig. 1B). The 31-bp
WT1 oligonucleotides contain a single base change so they
could concatamerize. The 26-bp Sp1 consensus oligonucleotide was
designed to contain the same 18-bp Sp1 consensus sequence used by
Letovsky and Dynan (48) with EagI-NgoMI ends.
Another Sp1 consensus oligonucleotide, also containing the 18-bp
sequence, was obtained from Promega. Synthesized oligonucleotides are
shown with mutations underlined and regions excised in parentheses:
WT1 wild-type enhancer (45 bp), upper strand (US)
5-GGCCGAGCCTCCTGGCTCCTCCTCTTCCCCGCGCCG(CCGGCCCCT)-3
and lower
strand (LS)
5
-(
AGAGGGG)CCGGCGGCGCGGGGAAGAGGAGGAGCCAGGAGGCTC-3
; WT1 A enhancer (45 bp), (US)
5
-GGCCG
CTGGCTCCTCCTCTTCCCCGCGCCG(CCGGCCCCT)-3
and (LS)
5
-(
AGAGGGG)CCGGCGGCGCGGGGAAGAGGAGGAGCCAG
C-3
; WT1 B enhancer (45 bp), (US)
5
-GGCCGAGCCTC
CCTCCTCTTCCCCGCGCCG(CCGGCCCCT)-3
and
(LS)
5
(
AGAGGGG)CCGGCGGCGCGGGGAAGAGGAGG
GAGGCTC-3
; WT1 C enhancer (45 bp), (US)
5
-GGCCGAGCCTCCTGGCT
CTTCCCCGCGCCG(CCGGCCCCT)-3
and (LS)
5
-(
AGAGGGG)CCGGCGGCGCGGGGAAG
AGCCAGGAGGCTC-3
; WT1 D enhancer (45 bp), (US)
5
-GGCCGAGCCTCCTGGCTCCTCCT
CGCGCCG(CCGGCCCCT)-3
and
(LS)
5
-(
AGAGGGG)CCGGCGGCGCG
AGGAGGAGCCAGGAGGCTC-3
; WT1 E enhancer (45 bp), (US)
5
-GGCCGAGCCTCCTGGCTCCTCCTCTTCCC
G(CCGGCCCCT)-3
and (LS)
5
-(
AGAGGGG)CCGGC
GGGAAGAGGAGGAGCCAGGAGGCTC-3
; WT1 B1 enhancer (45 bp), (US)
5
-GGCCGAGCCTC
GGCTCCTCCTCTTCCCCGCGCCG(CCGGCCCCT)-3
and (LS)
5
-(
AGAGGGG)CCGGCGGCGCGGGGAAGAGGAGGAGCC
GAGGCTC-3
; WT1 B2 enhancer (45 bp), (US)
5
-GGCCGAGCCTCCT
CTCCTCCTCTTCCCCGCGCCG(CCGGCCCCT)-3
and (LS)
5
-(
AGAGGGG)CCGGCGGCGCGGGGAAGAGGAGGAG
AGGAGGCTC-3
; WT1 B3 enhancer (45 bp), (US)
5
-GGCCGAGCCTCCTGG
CCTCCTCTTCCCCGCGCCG(CCGGCCCCT)-3
and (LS)
5
-(
AGAGGGG)CCGGCGGCGCGGGGAAGAGGAGG
CCAGGAGGCTC-3
; WT1 C4 enhancer (45 bp), (US)
5
-GGCCGAGCCTCCTGGCT
TCCTCTTCCCCGCGCCG(CCGGCCCCT)-3
and (LS)
5
-(
AGAGGGG)CCGGCGGCGCGGGGAAGAGGA
AGCCAGGAGGCTC-3
; WT1 C5 enhancer (45 bp), (US)
5
-GGCCGAGCCTCCTGGCTCC
CTCTTCCCCGCGCCG(CCGGCCCCT)-3
and (LS)
5
-(
AGAGGGG)CCGGCGGCGCGGGGAAGAG
GGAGCCAGGAGGCTC-3
; WT1 C6 enhancer (45 bp), (US)
5
-GGCCGAGCCTCCTGGCTCCTC
CTTCCCCGCGCCG(CCGGCCCCT)-3
and
(LS)
5
-(
AGAGGGG)CCGGCGGCGCGGGGAAG
GAGGAGCCAGGAGGCTC-3
; WT1 wild-type enhancer (31 bp), (US)
5
-CCGAGCCTCCTGGCTCCTCCTCTTCCCCGCG-3
and (LS)
5
-
CGGCGCGGGGAAGAGGAGGAGCCAGGAGGC-3
; WT1 C
enhancer (31 bp), (US)
5
-CCGAGCCTCCTGGCT
CTTCCCCGCG-3
and (LS)
5
-
CGGCGCGGGGAAG
AGCCAGGAGGC-3
. Sp1
consensus oligonucleotide (26 bp) was (US)
5
-GGCCGATTCGATCGGGGCGGGGCGAGC-3
and (LS)
5
-CCGGCTCGCCCCGCCCCGATCGAATC (48). egfr sequences were (US) 5
-CGGCCGCCTGGTCCCTCCTCCTCCCGCCCTGCCTGCCGGC-3
and (LS) 5
-GCCGGCAGGCAGGGCGGGAGGAGGAGGGACCAGGCGGCCG-3
(49), and
vav was (US)
5
-CGGCCGCCGCCCCATGGCTCCTCCTCCTCCACCCCCTCTAGA-3
and (LS)
5
-TCTAGAGGGGGTGGAGGAGGAGGAGCCATGGGGCGGCGGCCG-3
(50).
HeLa S3 cell crude nuclear extract (120 µg of protein/lane) was electrophoresed on 6.5% SDS-containing polyacrylamide gels (Bio-Rad). Size-separated proteins were transferred semidry to polyvinylidene difluoride membrane (Millipore). Membrane-bound proteins were denatured in 6 M guanidine in hybridization buffer (20 mM HEPES, pH 7.9, 100 mM KCl, 0.1 mM EDTA, 1 mM dithiothreitol) and stepwise renatured (51). Membrane strips were blocked with 5% nonfat dry milk in hybridization buffer, incubated overnight with monomeric radiolabeled 31-bp WT1 enhancer or 26-bp Sp1 consensus oligonucleotides (2.7 × 106 cpm/ml) in hybridization buffer with 0.25% milk at 4 °C for 15 h, washed six times with hybridization buffer with 0.25% milk at 4 °C, and autoradiographed.
Immunohistochemistry AnalysisKidneys from adult or fetal
(gestation day 18) Wistar rats (Charles River Laboratories) were
snap-frozen in OCT compound (Miles, Inc.) by immersion in liquid
nitrogen. Frozen sections (8 µm thick) were fixed in 20 °C
acetone for 30 min. Endogenous peroxidases were quenched with
H2O2, and biotin binding sites blocked
according to kit manufacturer (Vector Laboratories). After blocking
with normal goat serum, the sections were incubated at 4 °C
overnight with rabbit anti-human Sp1 serum AHP312 (Serotec USA) diluted 1:2000. Binding of Sp1 antiserum to kidney tissue was detected with an
ABC elite kit and peroxidase substrate kit (Vector Laboratories) using
the protocols recommended by the manufacturer. Some Sp1-stained sections were subsequently treated with periodic acid for 10 min and
stained with Schiff reagent to visualize basement membranes.
To identify key WT1 enhancer
sequences, we made 22/+2.4 kb,
9.4/+2.4 kb, and
2.5/+0.2 kb human
WT1-CAT reporter constructs and deletions (see
"Experimental Procedures") and tested their relative activities in
pooled stable transfection assays. We have designated as the
WT1 transcription start site (+1) the first G of the GGGG
sequence located 393 bp upstream of the WT1 ATG, as
determined by primer extension start site mapping in the mouse gene
(+1, Figs. 1, A and B) (36), and
confirmed for human WT1 as in the region
20 to +1 bp shown
in Fig. 1B (35, 37). The CAT gene was placed in the
WT1 5
-untranslated region, at position +206 in the
constructs derived from the
9.4/+2.4 kb construct and at +191 bp in
the
2.5 kb and smaller constructs. Transfections were carried out in
human lines, WT1-expressing 293 (embryonic kidney) and JMN
(mesothelioma) cells, and non-WT1 expressing HeLa (cervical
carcinoma) cells. Relative CAT activities of the reporter constructs
were normalized to that of the
48/+191 bp construct.
The intrinsic relative CAT activities of the large
WT1-reporter constructs and derived broad deletion
constructs were comparable in the three cell lines tested (Fig.
2). In addition, their absolute levels of CAT activity
were similar as well (data not shown). No large decrements in relative
CAT activity (>3-fold) were observed until the WT1 region
from 453 to
48 bp was deleted, whereupon CAT activity fell up to
10.6-fold, with the greatest effects observed in the
WT1-expressing 293 and JMN cells (Fig. 2). Although the
22/+2.4 kb construct was less active than the
9.4/+2.4 kb
construct, little transcriptional activity was lost with subsequent
4.3/2.4 kb,
2.4/0.2 kb,
1.1/0.2 kb, and
453/+191 bp deletions.
In fact, the latter proximate promoter construct retained from 30 to
79% activity of the most active construct (
9.4/+2.4 kb) and was more active than the the largest WT1-reporter construct
containing a full 24.4 kb of WT1 locus sequence (Fig. 2).
These findings support that over the region tested, the proximate
promoter contributes a large fraction of transcriptional activity to
the WT1 gene. An important non-tissue-specific
WT1 enhancer therefore resides in the proximate promoter,
between positions
453 and
48.
To obtain functional clues about the WT1 proximate promoter,
we compared the human (GenBankTM accession no. U77682[GenBank]) and mouse
sequences (36) (shown in part in Fig. 1A). This computer (MacVector®, Kodak Corp.) and hand aligned sequence
comparison reveals high homology (88.5%) over the 106 to +242 bp
promoter region, which falls off to 68% from
107 to
453 bp (data
not shown), with islands of conserved sequence. These observations
suggest that the highly conserved WT1 immediate 5
-flank may
have particular regulatory importance.
We then tested a series of fine deletion constructs spanning from
453/+191 bp to
48/+191 bp to localize the enhancer within this
~400-bp region. Nearly all enhancer activity, 52-220% of the
453/+191 bp construct and 41-66% of the
9.4/+2.4 kb construct, resides in a 33-bp region, from
81 to
48 with respect to the major
TSS (Fig. 3), supporting that the 33-bp enhancer is a
major contributor to WT1 transcription over the 24.4-kb
region examined. This enhancer functioned in all cell lines tested and
therefore does not appear to directly contribute to tissue-specific
gene expression. It conferred 5.5-12.5-fold activation in the stable transfection studies shown in Fig. 3. In transient transfection studies
in the same lines, activation by the enhancer was 2-3-fold less (data
not shown), suggesting that the enhancer plays a role in disrupting
inhibitory effects of chromatin. The enhancer was active as well in
mouse fibroblasts (NIH 3T3 cells) and was equipotent in the reverse
orientation (data not shown). The position of the enhancer, centered
45-66 bp upstream of the major transcription start site, indicates
that the enhancer likely serves a regulatory function and is not a core
promoter element.
"Linker Scanning" Analysis of the 33-bp Region
To
determine which part of the 33-bp WT1 enhancer was
responsible for its activity, a series of 81/+191 bp constructs were made with 6-bp tandem mutations introduced to "scan" the enhancer. Five double-stranded oligonucleotides were placed 5
of the
enhancerless
48/+191 bp WT1-reporter constructs to
generate a series of
81/+191 bp constructs with conversion of the
following nucleotides to 6 adenines, with mutations designated as A
(
80 to
75), B (
74 to
69), C (
68 to
63), D (
62 to
57),
and E (
56 to
51) (see Fig. 1B). The same pattern of
results was seen in each of the cell lines and was observed whether
cells were stably or transiently transfected. Mutations in the 6-bp B
or C regions completely eliminated enhancer activity, reducing reporter
activity to the level of the enhancerless
48/+191 bp construct (Fig.
4). The D mutation allowed partial activation, while the
A and E region mutations did not greatly affect enhancer activity (Fig.
4). These studies indicate that enhancer activity depends on the
integrity of a small subelement.
To determine the contribution of the enhancer to overall transcription
of the promoter region, we introduced the B, C, and D linker scanning
mutations in context into the active 453/+191 bp proximate promoter
construct and tested their effects on reporter activity in pooled
stable transfection assays. The
453/+191 bp construct contains all
identified forward-direction WT1 transcription start sites
(35-37). In both 293 and HeLa cells, mutations in the B or C regions
in the context of the active proximate promoter fragment brought
reporter activity down about 5-fold (Fig. 5), indicating
that the enhancer is responsible for the majority, perhaps 80%, of the
transcriptional activity of the promoter. It also suggests that in
these larger mutated WT1 promoter constructs, additional
GC-rich sequence, to which many transcriptional activators might bind,
and a full complement of transcription start sites cannot compensate
for loss of the enhancer. As in the
81/+191 bp linker scanning
constructs, the
453/+191 bp D mutation still allowed partial
activation (Fig. 5).
To further confirm the importance of the enhancer, single copies of the
wild-type and B, C, and D mutated enhancers were tested in front of a
heterologous minimal promoter, at a distance from the initiation site
comparable to the WT1 context. The base vector was the 13-bp
E1b TATA region linked to the CAT gene (40). The same pattern of
results was observed in transiently transfected 293 and HeLa cells. As
shown in Fig. 6, the enhancerless construct had no CAT activity above
background, but the addition of the wild-type enhancer brought reporter
activity up remarkably. Fold activation is infinite (Fig.
6), and the level of CAT activity is within 2-fold of
the 453/+191 bp construct (data not shown). The D mutated enhancer
had about half the activity of the wild-type. These observations
indicate that this small WT1 enhancer is itself sufficient
to confer a high level of transcriptional activation and, in the
WT1 context, likely does so through the major WT1 TSS.
In sum, we have identified a powerful enhancer in the WT1 promoter that accounts for the majority of its transcriptional activity.
A Prominent Protein-DNA Complex Forms Only in Association with the Functional Enhancer SequencesThe simplest model to explain the
linker scanning transfection data is that a transcriptional activator
is capable of binding the wild-type and functional mutated enhancers,
but not the non-functional mutated enhancers. To test this hypothesis,
EMSAs were performed using the 36-bp linker scanning oligonucleotides
as probes. In all EMSAs comparing the wild-type and mutated enhancer
oligonucleotide probes, the probes were labeled to comparable specific
activities and incubated with equivalent amounts of protein. Nuclear
extracts were prepared from WT1-expressing human and pig
fetal kidney and the above cell lines. As predicted by the model, a
prominent protein-DNA complex (upper band) forms in vitro
with the wild-type enhancer sequence, but not with the B or C mutated
enhancers, which lacked enhancer activity (Fig. 7). In
contrast, the A, D, and E mutated enhancers that retained enhancer
activity were able to form the same upper band complex as the wild-type
(Fig. 8). Interestingly, the D mutated enhancer, which
had only partial activating potential (Figs. 4, 5, 6), exhibited reduced
affinity for the upper band factor (Fig. 8). An additional
double-stranded oligonucleotide, designated F, with bases 50 to
45
changed to adenines, also readily formed the upper band complex (data
not shown). These in vitro findings correlate well with the
cell culture transfection results and suggest that activity of the
WT1 enhancer depends on its ability to form this protein-DNA
complex, and that the complex likely contains the relevant
transcriptional activator.
To further refine the DNA contacts required for DNA binding of this
factor, six additional double-stranded linker scanning oligonucleotides
were made containing 2-bp tandem mutations that span the 12 bp of the B
and C regions. They are designated B1-3 and C4-6 and have the
following 2 bp changed from the wild-type sequence to 2 adenines: B1
(74,
73), B2 (
72,
71), B3 (
70,
69), C4 (
68,
67), C5
(
66,
65), and C6 (
64,
63) (see Fig. 1B). In EMSA
assays, formation of the prominent upper band complex is nearly lost
with the B3 mutation, and completely lost with the three C mutations
(Fig. 9). Therefore, the wild-type sequence of
nucleotides that is essential for protein-DNA complex formation is
CTCCTCCT. It is also likely that the first C of the D region contributes to binding as well. The (CTC)3 sequence and
position are completely conserved in the mouse WT1 gene
(Fig. 1A).
Additionally, as a means of further correlating protein-DNA complex
formation with enhancer activity, another set of 81/+191 bp
WT1-CAT reporter constructs were made using the B1-3 and C4 oligonucleotides and tested in transiently transfected HeLa cells. These 2-bp mutated enhancers conferred transcriptional activation that
again correlated with their respective abilities to form an in
vitro protein-DNA complex. The B1 and B2 mutated enhancers conferred 8.9- and 7.1-fold activation, respectively, the B3 enhancer (that exhibited much reduced binding) 2-fold activation, while the C4
mutation that showed no binding had no enhancer activity (Fig.
10). These observations underscore how much the
conditions of the in vitro binding reactions reflect the
activity of the enhancer in vivo and, consequently, provide
additional support that the protein-DNA complex formed in
vitro may contain the relevant transcriptional activator.
The DNA-binding Protein Has Characteristics of Sp1
Formation of the protein-DNA complex requires zinc. When EMSA reaction buffer (containing 0.1 mM EDTA) lacked zinc, no upper band complex formed (data not shown). Addition of excess zinc in the presence of the EDTA or the zinc chelator orthophenanthroline allowed complex formation (data not shown). Sp1, which contains three C2H2 zinc fingers, requires zinc for DNA binding.
Protein-DNA complex formation could be prevented by preincubation of extracts with the lectin wheat germ agglutinin (WGA). This inhibition could be overcome by including in the EMSA reaction N-acetylglucosamine, the ligand recognized by WGA (data not shown). This suggested that the reaction was specific, and that the DNA-binding protein was likely O-glycosylated, as has been described for several transcription factors, most notably Sp1 (52) and HNF1 (53). The protein could also be partially purified using a WGA-agarose affinity resin, as has been shown for both Sp1 (45, 52) and HNF1 (53).
Binding of the factor to DNA was optimized in EMSA studies (data not shown). A KCl concentration of 100 mM was more than 2-fold better than 50 or 150 mM, above or below which binding fell off considerably. Titration of pH showed that pH 8.4 (or even 9.0) was better than pH 7.9, 7.4, 7.1, or 6.5, below which binding was greatly reduced. EDTA concentration was kept at 0.1 mM as zinc concentration was optimized. Binding improved up to 0.25 mM ZnSO4, but was completely abolished with 0.5 mM or greater ZnSO4. Sp1's exquisite sensitivity to zinc concentration for DNA binding is known (54, 55). Addition of magnesium interfered with protein-DNA complex formation. With the exception of magnesium, our optimized conditions are similar to those reported for Sp1 protein-DNA interactions with the GC box: 100 mM KCl, 3-12.5 mM MgCl2, 0-1 mM EDTA, 1 mM dithiothreitol, 0-0.1% Nonidet P-40, HEPES or Tris, pH 7.5 or 7.9 (48, 56-58).
The factor was found in all cell lines tested and in differing amounts, as assessed by EMSA (data not shown). The amount of the factor present in nuclear extracts correlated with the activity of the enhancer in that cell line. The amount of the protein was determined to be 1/5,000 of total nuclear extract protein in HeLa S3 cells, which is comparable to reported Sp1 content (45). In HeLa S3 cells, we found about one-third of the binding activity in the cytoplasmic S100 fraction, but at one-fifth the concentration as in nuclear extract.
The WT1 Enhancer Binding Factor Is Sp1EMSA studies with Sp1
antibody and competition with a consensus Sp1 oligonucleotide strongly
support that the DNA-binding protein is Sp1. An affinity-purified mouse
monoclonal antibody directed against Sp1 amino acids 520-538 (Santa
Cruz Biotechnology) or an affinity-purified rabbit polyclonal antibody
made against Sp1 amino acids 436-454 (Santa Cruz Biotechnology) were
added to EMSA reactions containing the wild-type WT1
enhancer sequence or an Sp1 consensus oligonucleotide. These antibodies
are directed against sequences that are amino-terminal to the
DNA-binding zinc finger domain and would therefore be expected to
supershift, not prevent formation of, an Sp1-containing protein-DNA
complex. Both antibodies are Sp1-specific, as they do not cross-react
with Sp1 family members Sp2, -3, or -4 (Santa Cruz Biotechnology). As
shown in Fig. 11, both antibodies were capable of
supershifting the prominent upper band complex formed with the
WT1 or consensus GC box oligonucleotides, strongly
suggesting that the DNA-binding material contains Sp1. Furthermore, the
similar appearance of the supershifts from the two different
oligonucleotides suggests they are composed of identical material. With
the polyclonal antibody, nearly all material in the protein-DNA complex
is supershifted, indicating that the major protein component is Sp1, as
might be expected for a gel shift with the GC box.
Oligonucleotide competition studies in EMSAs revealed that the
WT1 oligonucleotide could effectively compete for Sp1
binding with a consensus Sp1 site. The WT1 wild-type
enhancer oligonucleotide was labeled to the same specific activity as
the Sp1 consensus oligonucleotide, and the molar excess of unlabeled
competitor is shown (Fig. 12). A 10-fold excess of
either unlabeled oligonucleotide (lanes 3 and 8)
sustantially reduced binding of Sp1-immunoreactive material to the
other binding site (Fig. 12), indicating each site binds Sp1 with
comparable affinity. In fact, the WT1 enhancer appears to be
the better competitor and hence the higher affinity site (see below)
under these assay conditions. The opposite strand of the CTCCTCCTC
WT1 sequence (consensus in uppercase: GaGGaGGAG) is
consistent in 7 of 9 bases with the Sp1 consensus sequence determined
by Kadonaga et al. (T/G, GGGCGG, A/G, G/A, C/T) (59) and
later confirmed (60). However, it is still surprising that the 2 nucleotide differences are in the 6-base Sp1 core, GGCGGG, and that the
WT1 site was not detected in the DNA target assay (60).
Nevertheless, this work indicates that the WT1
(CTC)3 enhancer is a high affinity Sp1 binding site (see
below).
To determine the molecular weight of the WT1
enhancer-binding factor and to confirm it as Sp1, we performed
Southwestern analysis (51, 61) of HeLa cell crude nuclear extract using
the 31-bp WT1 wild-type, C mutated, and 26-bp Sp1 consensus
oligonucleotides. We had also noted a subtle difference in
electrophoretic mobility by EMSA of the Sp1-DNA complex using the
WT1 versus the Sp1 consensus oligonucleotide and considered
that the post-translational modification of the WT1
enhancer-binding material might be different from that which bound the
Sp1 consensus site. We therefore attempted to size separate well the
p105 and p95 forms of Sp1 using a 6.5% denaturing acrylamide gel
before hybridizing with the radiolabeled oligonucleotides. As shown in
Fig. 13, the p105 form of Sp1 bound the WT1
wild-type and Sp1 consensus oligonucleotides, while, as expected from
the EMSA studies, binding to the C mutated sequence was considerably
reduced. The reduced binding to the C mutated enhancer also supports
the specificity of the Sp1 protein-WT1 enhancer interaction.
In addition, that the assay worked after size separation of proteins
and shows a strong interaction between Sp1 and the WT1
enhancer indicates that no cofactor is required for their interaction.
These results further support that the WT1 enhancer binds
Sp1, and the same form of Sp1 that binds the consensus Sp1
oligonucleotide.
The WT1 Enhancer Is a High Affinity Sp1 Binding Site
The
dissociation constants (KD) of the 31-bp
WT1 enhancer and 26-bp Sp1 consensus oligonucleotides for
Sp1 protein were determined in EMSAs by titrating an increasing amount
of probe against a constant amount of Sp1 in WGA partially purified nuclear extract preparations (Fig. 14A). The
amount of bound and free probe were determined by Cerenkov counting of
bands excised from the dried, autoradiographed gel. A 140-fold
concentration range of probe was tested, including amounts that
exceeded by more than 10-fold the observed KD
values. Binding was saturable, reaching a plateau of ~0.28
nM with 0.78 nM total probe and beyond (Fig.
14, A and B), and the EC50 was
~0.35 nM. The KD was calculated in
Rosenthal (modified Scatchard) plots as the negative reciprocal of the
slope (47). The KD of Sp1 for the wild-type
WT1 enhancer sequence was 3.7 × 1010
M (Fig. 14B, inset), and for the Sp1
consensus oligonucleotide was 8.1 × 10
10
M (data not shown). These experiments were performed
multiple times, with the WT1 site revealing a consistent
2-fold higher Sp1 affinity than the GC box. Moreover, two different GC
box oligonucleotides from different suppliers gave identical results.
This observation is also supported by the competition studies (Fig.
12), in which the WT1 enhancer was the better competitor,
and Southwestern analysis (Fig. 13), in which equivalent amounts of
probe of similar specific activity were used, but in which binding of
the labeled WT1 oligonucleotide appears qualitatively better
than the consensus oligonucleotide. The published Sp1
KD values for a GC box consensus site (identical to
our Sp1 consensus sites) range from 4.1 to 5.3 × 10
10 M using DNase I footprinting and EMSAs
(48), which are perhaps no different from what we have observed for the
WT1 enhancer site, but which are better than what we have
observed for the GC box. It may well be that the relative affinities of
Sp1 for different binding sites change under different conditions and,
more importantly, that these differences may be biologically relevant.
Divalent cation concentrations may be important in this regard, as our studies were performed in the absence of magnesium, which was included
in the GC box-Sp1 interaction studies (48, 56), but which we found to
interfere with complex formation with the CTC repeat. The correlation
we have observed between the function of the wild-type and mutated
WT1 enhancers and their ability to bind Sp1 in
vitro suggests the conditions we are using for EMSAs provide a
good model of the in vivo setting. Other non-consensus sites
with Sp1 affinities 10-20-fold lower than the GC box have also been
reported (59). The high affinity of Sp1 for the single WT1
enhancer site may help explain its potency, and again underscores the
importance of Sp1 to WT1 transcription.
We conclude from these experiments that Sp1 binds the WT1 enhancer in vitro and that the enhancer contains a single, near-consensus, high affinity Sp1 site that binds Sp1 at least as well as a consensus GC box. Moreover, in conjunction with the data above, the binding of Sp1 to the wild-type and mutated enhancers in vitro parallels the activation potential of the enhancers, suggesting that the activator is indeed Sp1.
Sp1 Transactivates a Reporter Gene through the WT1 EnhancerTo ascertain the effect of Sp1 on the WT1
enhancer in an Sp1-null setting, we transiently transfected into
Drosophila Schneider S2 cells the wild-type and mutated
enhancers in the 81/+191 bp context and the
48/+191 bp construct,
with and without an Sp1 expression vector. Without Sp1, all the
constructs had comparably low absolute reporter activity levels (Fig.
15, black bars). With cotransfection of 100 ng of Sp1 expression vector, reporter activity of the mutated enhancer
and enhancerless constructs increased comparably, from ~0.43 to
~2.4%, whereas activity increased from 0.37 to 14.6% (~6-fold
more) with the wild-type
81/+191 bp construct (Fig. 15). Therefore,
the WT1 enhancer was not functional in the absence of Sp1,
and was only active when able to bind Sp1. This result correlates well
with the observations in human cells and in EMSAs with human protein,
and again supports that the transcriptional activator that exerts its
effect on the critical WT1 enhancer is Sp1. Although Sp1 did
activate the constructs with mutated enhancers or lacking the enhancer,
their activated levels were still only ~16% of the level achieved
with the intact
81/+191 bp construct. Therefore, despite that each
mutated construct contains 273 bp of 70+% GC-rich WT1
sequence with many potential Sp1 binding sites, more than 80% of Sp1
transactivation is mediated by the intact CTC repeat. This experiment
confirms that Sp1 is capable of strongly transactivating a reporter
gene linked to the small WT1 enhancer containing a single,
alternative Sp1 site.
CTC Repeats in Other Growth-related Genes Also Bind Sp1
Through homology searches, we have identified a number of genes that also contain CTC repeats in their promoters, several at a distance from an identified TSS comparable to that found in WT1. As a group, these genes are transcribed from TATA-less promoters, are tissue-restricted in their expression, and are growth-related, which suggests that, as an Sp1 site, the CTC repeat may serve a unique regulatory role. Genes in this group are egfr (49), pdgf-A chain (62, 63), egr-1 (64), c-myc (65), c-myb (66), and vav (50). Thy-1 has a similar opposite strand site downstream of its TSS (67). To determine whether other CTC repeats were high affinity Sp1 sites, we tested ~40-bp oligonucleotide promoter fragments containing CTC repeats from egfr and the vav proto-oncogene in EMSA and EMSA competition studies. As expected, these other CTC elements were readily capable of forming the same protein-DNA complex as the WT1 element and could compete effectively for Sp1 binding to the WT1 element (data not shown), indicating they too were high affinity Sp1 sites.
Sp1 Expression during NephrogenesisSaffer et al. showed that Sp1 tissue levels vary up to 100-fold during murine development, as assessed by immunohistochemistry and RNA dot blotting (68). At postnatal day 32, high Sp1 protein levels were detected in spleen, early hematopoietic cells, testes, and uterine decidua and moderate levels in kidney glomerular cells (68); WT1 is expressed in all these tissues (2, 29, 36, 69-71). High levels of Sp1 were also found in tissues that do not express WT1, such as gastric and lung epithelia and spermatids (68), but no WT1-expressing tissue had little or no Sp1 expression, with the possible exception of testis Sertoli cells. Consequently, we hypothesized that Sp1 might be spatially and temporally regulated in nephrogenesis and thereby account, to some extent, for the WT1 expression pattern.
To determine if a correlation exists between Sp1 and WT1, we
examined Sp1 protein expression by immunohistochemistry in day 18 fetal
(Fig. 16, A-C) and adult rat kidney (Fig.
16, D-F). At embryonic day 18, all stages of nephrogenesis
can be observed (Fig. 16A). Near-contiguous sections were
examined for Sp1 immunoreactivity or hematoxylin and eosin staining
(data not shown) to allow identification of structures. Control
sections stained with the secondary antibody alone were essentially
negative (Fig. 16, B and E). In Fig.
16A, more Sp1 is found in the outer, undifferentiated part
of the section (o) than in the inner, differentiated region
(i). Sp1 staining is most prominent in nuclei of uninduced
mesenchyme cells (Fig. 16A, closed arrow), which
do exhibit weak WT1 expression (30-32), and early-induced
mesenchyme cells (open arrow), which have increasing amounts
of expressed WT1 (29). Sp1 staining remains dense in early
tubule structures and is reduced in more differentiated cortical
tubules (Fig. 16A), becoming barely detectable in mature proximal tubules (Fig. 16, D and F, open
arrows). In contrast, Sp1 expression is maintained at moderately
high levels in developing (Fig. 16C, center) and
mature glomeruli (Fig. 16, D and F,
center). The only renal cell type that expresses
WT1 in adulthood is the podocyte (30, 31). To determine
whether the Sp1-staining glomerular cells included podocytes, we
counterstained the Sp1-stained adult rat kidney sections with periodic
acid-Schiff reagent (Fig. 16F). Periodic acid-Schiff reagent
stains the carbohydrate-rich basement membranes, which allows
distinction of glomerular endothelial cells, which reside inside
glomerular capillary loops, from podocytes, which are outside the
loops. In Fig. 16F, closed arrows denote podocytes staining with moderate Sp1 levels. Glomerular mesangial cells
(Fig. 16F, arrowhead), which do not express
WT1, had high Sp1 levels. Thus, the
WT1-expressing structures have at least moderate levels of
Sp1, but not all high Sp1-expressing cells express WT1.
Because Sp1 expression completely overlaps that of WT1, Sp1
might therefore be a prerequisite for WT1 expression. In
addition, falling Sp1 levels may account for loss of WT1
expression, as observed in developing proximal convoluted tubules (Fig.
16, D and F). These findings support the
importance of Sp1 to WT1 expression.
We are reporting several novel findings. First, using an unbiased
approach, we have identified a critical Sp1 site in the WT1
promoter that accounts for the majority of its transcriptional activity, and which is the most important enhancer over 24.4 kb of
WT1 5-flanking and first intron sequence, based on these
studies in stably transfected cultured cells. Second, although previous work has suggested the CTC repeat may be an Sp1 site (see below), we
believe we have provided conclusive evidence that the CTC repeat is an
Sp1 site, that it binds the same form of Sp1 as the consensus GC box,
and that it does so with an affinity at least as high as the consensus
GC box. Third, we have found that similar CTC repeats reside in the
immediate 5
-flanking regions of a subset of tissue-restricted,
growth-related genes, suggesting this site may be a critical,
operationally distinct Sp1 site. Finally, we have shown for the first
time that Sp1 expression is temporally and spatially regulated during
nephrogenesis in a pattern that not only supports a role in
WT1 regulation, but also suggests Sp1 itself may be
important in kidney development. These studies indicate that the CTC
repeat element and the associated ubiquitous transcription factor Sp1
are critical to expression of the tissue-restricted WT1
gene.
Although ubiquitous, Sp1 may participate in tissue-specific
WT1 expression in several ways. Foremost, our
immunocytochemistry data indicate that Sp1 is spatially and temporally
regulated in nephrogenesis and further support that Sp1 may be
necessary for WT1 expression. Similarly, Sp1 is expressed at
different levels in gastric (68) and hematopoietic development (72) and
influences differential gene expression in these cells (72, 73).
Monocyte-specific expression of the CD11b (74) and CD14 promoters (75)
also depends on Sp1. Change in Sp1 levels, therefore, is one likely way
Sp1 could contribute to tissue-specific WT1 expression.
There is also growing evidence of other mechanisms through which Sp1 may influence differential gene expression. A great number of Sp1-interacting proteins have been identified, many of which are tissue-restricted transcription factors, such as GATA-1 (72, 76),
GATA-2, and GATA-3 (76), HNF4 (77), the retinoic acid (78),
triiodothyronine (79), and estrogen receptors (80), Ets-1 (81), and
MyoD (82). Such factors, perhaps bound to elements beyond the 24-kb
region we examined, may interact with Sp1 at the critical promoter site
we have identified. For example, a GATA-1-responsive enhancer was found
in the WT1 3 region (38, 83), outside the sequences we
tested. GATA-1 that binds the WT1 enhancer in hematopoietic
cells, or GATA-3 in the kidney (84), might interact with promoter-bound
Sp1. We have also observed that the CTC repeat confers a striking
20-fold activation in 786-O cells,3 a renal
cell carcinoma line lacking a functional copy of the von Hippel
Lindau (VHL) tumor suppressor gene (86). This
activation is ~3-fold higher than that in any other transiently
transfected cell line3 and suggests the VHL gene
product might normally inhibit the CTC enhancer or Sp1 itself. A
functional relationship between these two transcription factors, which
we are examining, would suggest that differential gene expression
involving Sp1 may in addition include interactions with other
ubiquitous factors with differing tissue levels. Sp1 might also affect
WT1 expression in other ways. Sp1 levels and/or activity can
be increased by oncogene expression (87-89), growth factors (90-92),
or cytokines (93, 94). Whether any of these stimuli affect
WT1 expression remains unexplored.
Somewhat surprisingly, we and others (37, 38) have found that the
WT1 promoter-reporter constructs are transcriptionally active when transfected in either non-WT1-expressing or
WT1-expressing cell lines. In contrast, we have used a
similar construct (1.9/+0.2 kb WT1-lacZ) to
generate four transgenic mouse lines and were unable to detect
transgene expression in any mouse tissue, despite having observed
expression in transfected cultured cells.3 What are
possible explanations for this discrepancy in transfected versus transgenic WT1-reporter expression? In
cell lines, it is well recognized that stably integrated DNA exists in
an "active" conformation (95-98), which would favor ubiquitous
expression. Selection of neo-resistant clones requires that
integration occur in a transcriptionally active site, and
co-transfected DNA should integrate there as well (99, 100).
WT1 may also be regulated by DNA methylation, as the gene
contains CpG islands (17, 18). Lack of methylation of transfected DNA
may have contributed to its ubiquitous expression (98, 101). A
WT1 intron 3 silencer (102) was not included in our
constructs and therefore would not have accounted for the difference in
expression between transgenic and transfected cells. Some additional
repressive chromatin feature may form in transgenic cells subjected to
differentiation forces that cannot form on transfected DNA in a
monomorphic cell line. The cell lines' endogenous WT1 genes
do appear to be subject to a higher level of regulation, however. We
mapped the DNase I-hypersensitive sites of the endogenous
WT1 locus in these lines and found differential hypersensitivity of the WT1 promoter.2 In sum,
although we do not have WT1 sequence that confers
tissue-specific expression, we have identified the key
non-tissue-specific cis element that resides in a promoter region
accessible only in WT1-expressing cells.
Based on its high GC content, the WT1 promoter was
previously considered as a target for regulation by zinc finger
transcription factors (35, 103, 104). Rupprecht et al. (104)
used large amounts of CMV-driven WT1 expression vectors to show that
WT1 negatively autoregulates murine WT1 promoter-reporter
constructs. The WT1-mediated repression depended predominantly on
sequence from +31 to +201 bp, although DNase I footprinting over the
513 to +201 bp region using purified WT1 protein revealed seven
protected regions spanning 25-70 bp each, including the homologous CTC
repeat we identified ((104), Fig. 1A). Similarly, Malik
et al. (103) examined negative autoregulation of the human
WT1 promoter using an inducible WT1 system and found that
high WT1 protein levels reduced reporter activity ~50% and that
repression also depended on WT1 promoter sequence from +38
to +195 bp. Therefore, WT1 may well autoregulate its own promoter, but
high WT1 protein levels are required. Finally, in their initial
description of the GC-rich WT1 promoter, Hofmann et
al. (35) tested whether Sp1 might be important for its activity.
Over the
453 to +191 bp region, purified Sp1 protein protected nine
elements, 15-27 bp in size. The essential CTC repeat we identified was
mostly protected in this assay, but there was no indication of its
importance. Cotransfection of 10 µg of CMV-driven Sp1 with a similar
453/+191 bp construct increased reporter expression about 3-fold
(35). Our work, in contrast, identifies as the major enhancer of the
WT1 promoter a single high affinity Sp1 site. Moreover,
mutation of this site (Fig. 6) indicates that endogenous amounts of Sp1
acting through this CTC repeat account for ~80% of WT1
promoter activity.
CTC repeats similar to the WT1 element are found in the
proximate promoters of other growth-related genes (see above). The egfr immediate 5-flank contains four closely spaced CTC
repeats that are S1 nuclease-sensitive (105). These elements are
important to overall egfr promoter transcription, as their
deletion reduced egfr promoter-reporter activity 3-5 fold
(105). These sites were considered to bind Sp1, as well as an
unidentified factor, because purified Sp1 protein footprinted this
region (105). The c-myc gene promoters were subjected to
DNase I footprinting, and a number of the footprints identified with
crude nuclear extract could be reproduced with purified Sp1 protein,
including a similar CTC repeat (65). However, these authors did not
demonstrate, as we have, that the crude nuclear extract protein that
binds the CTC repeat is Sp1.
CTC repeats may even be targets for transcriptional repression by WT1
protein, but large amounts of WT1 expression vector are
required (24, 103, 104, 106) (see above). In contrast, we have not
observed WT1, but only Sp1 binding to the CTC repeat, even in crude
nuclear extracts from high WT1-expressing Wilms' tumor (gift of A. J. Garvin, Medical University of South Carolina), LP9 mesothelial cells
(2) (gift of J. Rheinwald), or fetal kidney, supporting that Sp1
preferentially binds the CTC repeat.3 It is still
intriguing, however, that the CTC repeat is a potential site for
Sp1-WT1 competition, as has been reported elsewhere for Sp1 with Egr-1
(107) and other factors (79, 108-110). Although sequence other than
the CTC-repeat has been shown necessary for WT1-mediated repression
(103, 104, 106, 111), two WT1 sites, one 5 and one 3
of the TSS, may
be required for repression, as proposed by Wang et al.
(112). Because WT1 homodimerizes (113), the presence of adjacent WT1
binding sites may increase WT1's affinity for the sites. Most
convincingly, transfected WT1 represses the endogenous egfr
gene, and this effect is mediated by CTC repeats (24). The ability to
localize either Sp1 or WT1 near a TSS may therefore be a critical role
of the CTC repeat.
Might the CTC repeat element itself serve a regulatory function? CTC repeats and similar polypyrimidine tracts can exist in non-B DNA structural conformations (85, 114-116), which have long been proposed to act as transcriptional regulatory elements. Because these structures permit some single-strandedness, polypyrimidine tracts may exhibit S1 nuclease sensitivity (28, 85, 106, 114-116) and therefore be particularly accessible to transcription factor binding. However, the WT1 CTC repeat, its surrounding sequence (Fig. 1A), and even the adjacent plasmid DNA do not contain features typical of polypyrimidine tracts with higher order structure. For example, no direct repeat of the (CTC)3 sequence is found nearby that might serve as a target for DNA slippage, as proposed for several promoters (85, 115), including that of egfr (106). No inverted repeat described for cruciform structures (28) is found either. In addition, unlike the WT1 region, the continuous polypyrimidine tracts that have been shown to be S1 nuclease sensitive are long simple repeats, i.e. 45 bp of d(TC)n·d(GA) (116) and 33 bp of d(GA)n·d(TC)n (114). For these reasons, the (CTC)3 element function is most likely attributable to interacting transfactors.
In conclusion, we have identified a critical enhancer in the WT1 gene as a single, alternative Sp1 site located immediately upstream of the start of WT1 transcription. This site, a CTC repeat, exhibits at least as high an Sp1 affinity as the consensus GC box and is found in other TATA-less, tissue-restricted genes. The Sp1 expression pattern during nephrogenesis lends support to Sp1 serving a key role in WT1 transcription and perhaps in regulation of other genes critical in nephrogenesis.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U77682[GenBank].
We thank H. Rennke for review of immunohistochemistry results; K. Call, D. Haber, S. Patwardan, J. Rheinwald, M. Segal, and R. Tjian for reagents; and S. Jackson, J. Saffer, and D. Tenen for valuable advice.