A Nonerythroid GATA-Binding Protein Is Required for ... - NCBI - NIH

Vol. 10, No. 9

MOLECULAR AND CELLULAR BIOLOGY, Sept. 1990, p. 4854-4862 0270-7306/90/094854-09$02.00/0 Copyright C) 1990, American Society for Microbiology

A Nonerythroid GATA-Binding Protein Is Required for Function of the Human Preproendothelin-1 Promoter in Endothelial Cells DAVID B.

WILSON,123 DAVID M. DORFMAN,'2'3'4 AND STUART H. ORKIN'.23.5*

Division of Hematology-Oncology, The Children's Hospital,' Dana-Farber Cancer Institute,2 Department of Pediatrics, Harvard Medical School,3 Department of Pathology, Brigham and Women's Hospital,4 and Howard Hughes Medical Institute,5 Boston, Massachusetts 02115 Received April 1990/Accepted 22 June 1990

Endothelin-1 (ET-1) is a 21-amino-acid peptide synthesized by endothelial cells that has potent vasoconstrica 212-amino-acid prepropeptide, termed preproendothelin-1 (PPET-1). To identify cis-acting sequences essential for PPET-1 gene transcription, bovine aortic endothelial (BAE) cells were transfected with plasmids containing 5'-flanking sequences of the human PPET-1 gene fused to the human growth hormone gene as a reporter. Deletional analysis of these fusion plasmids showed that the sequence spanning positions -141 to -127 of the human PPET-1 promoter is required for full transcription activity. Introduction of clustered point mutations into this region of the promoter reduced transcription activity. Gel shift analysis, methylation interference, protein-DNA cross-linking, and oligonucleotide competition studies revealed that BAE cell nuclear extract contains a 47-kilodalton DNA-binding protein recognizing the core motif TATC (GATA) located at positions -135 to -132 of the PPET-1 promoter. The size and specificity of this DNA-binding protein resemble GF-1, a previously described transcription factor of erythroid cells that binds to the same core motif. Gel shift analysis indicated that GF-1 and the DNA-binding protein interacting with the PPET-1 promoter have different tissue distributions; the former is restricted to a subset of hematopoietic cells, and the latter is found in various cell types, including BAE, NIH 3T3, and HeLa cells. By using an antiserum to the C-terminal region of GF-1, the two proteins were also found to be antigenically distinct. When a growth hormone fusion plasmid containing the proximal 141 nucleotides of the PPET-1 promoter was transfected into a variety of cell types, there was preferential expression in cells of endothelial origin. We conclude that a nuclear factor with binding specificity for a GATA motif similar to that of the transcriptional activator GF-1 is necessary for the efficient and cell-specific expression of the human PPET-1 tor activity. Human ET-1 is derived from

gene.

the human PPET-1 promoter to identify cis elements and trans-acting factors controlling its expression. In the experiments reported here, we have delineated a short cis element containing the core motif TATC (or GATA) which is required for efficient and cell-specific expression of the PPET-1 gene. This element is recognized by a DNA-binding protein with sequence specificity which is remarkably similar to that of the erythroid cell transcription factor GF-1 (23, 34, 39) (also referred to as Eryfl [10, 11, 33], NF-E1 [35], EF-1 [37], and EF-ya [14]), yet distinguishable by its cellular distribution and by use of an antiserum to GF-1. Our findings, therefore, implicate a distinct GATA-binding protein in the regulation of PPET-J gene expression.

Endothelin-1 (ET-1), a potent vasoconstrictor peptide originally isolated from cultured porcine aortic endothelial cells (36), is the first-discovered member of a family of related peptides. The three endothelins, ET-1, ET-2, and ET-3, each consist of 21 amino acids and two intrachain disulfide bridges (2, 16). Human ET-1 is generated by endopeptidase cleavage of a 212-amino-acid prepropeptide termed preproendothelin-1 (PPET-1) (8, 18, 30). The 2.0kilobase mRNA for PPET-1 is expressed in cultured endothelial cells (36) and aortic endothelial cells in vivo (21). PPET-1 mRNA expression is enhanced by thrombin, calcium ionophore A23187, epinephrine (36), transforming growth factor ,B (19), interleukin 1B (37), phorbol ester (17), and increased shear stress (38). The complete nucleotide sequence of the human PPET-J gene (17) and its chromosomal location (1) have been determined. The gene consists of five exons distributed over 6.8 kilobases of genomic DNA (17). By inspection, the 5'-flanking region of the human PPET-J gene has been found to contain nucleotide sequences for several potential cis-regulatory elements, including (i) the phorbol ester-responsive element, also known as the APl/c-jun-binding element; (ii) the binding motif for nuclear factor 1 (NF-1); and (iii) hexanucleotide sequences for the acute-phase regulatory elements (17). To date, however, there is no direct evidence that these regulatory elements participate in regulation of the gene. In this study, we have undertaken a functional analysis of

*

MATERIALS AND METHODS Oligonucleotide preparation and labeling. Oligonucleotides used as gel shift probes or primers for polymerase chain reaction (PCR) were synthesized on a 380B DNA synthesizer (Applied Biosystems, Inc.) and desalted on NAP-10 columns (Pharmacia). Oligonucleotides were end labeled with [-y32P]ATP (Amersham Corp.) by using T4 polynucleotide kinase (Amersham) (22) and then purified by polyacrylamide gel electrophoresis (PAGE). DNA fragments used as hybridization probes in Northern (RNA) analysis were labeled by using the Klenow fragment (Amersham), [a-32P]dCTP (Amersham), and random hexanucleotide priming (22). Isolation and sequencing of a human PPET-1 genomic clone. A full-length human PPET-J cDNA was isolated from a gtll human umbilical vein endothelial (HUVE) cell library (13)

Corresponding author. 4854

VOL. 10, 1990

by plaque hybridization with a labeled 63-mer probe complementary to the published sequence of human PPET-J (17). The cDNA clone was used to screen a human genomic library in EMBL3A. A genomic clone containing 1.8 kilobases of PPET-J 5'-flanking sequence was isolated. The sequence of this genomic clone, determined by the dideoxynucleotide method (29) with Sequenase (United States Biochemical Corp.), was found to agree with published data (17). Plasmid constructions. Sall linkers were added to a 1.8kilobase EcoRI restriction fragment containing the 5'-flanking region of the PPET-J gene. The fragment was digested with SalI and BglII and then cloned into p4GH (Nichols Institute) via 5' Sall and 3' BamHI sites. The resultant fusion plasmid, termed pPPET-1-GH, contained a portion of the PPET-J promoter (extending from positions - 1410 to + 145) fused to the human growth hormone gene. Deletion constructs (containing portions of the PPET-1 promoter fused to the growth hormone gene) were generated by PCR with pPPET-1-GH as a template and synthetic oligonucleotide primers (9). These fusion constructs were cloned into p+GH via 5' Sall and 3' BamHI sites. To monitor for nucleotide incorporation errors during PCR all fusion plasmids were sequenced before use in transfection assays. A plasmid containing the thymidine kinase promoter fused to the growth hormone gene (pTK-GH) was obtained from Nichols Institute. Cell cultures, DNA transfections, and growth hormone assays. Bovine aortic endothelial (BAE) and HUVE cells were generously provided by T. Collins. BAE, NIH 3T3, and HeLa cells were grown in Dulbecco modified Eagle medium supplemented with 10% calf serum, 10 mM HEPES (N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid), 50 U of penicillin per ml, 50 ,ug of streptomycin per ml, and 2 mM glutamine. HepG2 and murine erythroleukemia (MEL) cells were maintained in 10% fetal calf serum supplemented with 10 mM HEPES, 50 U of penicillin per ml, 50 ,ug of streptomycin per ml, and 2 mM glutamine. EA.hy926 cells (7), provided by C. S. Edgell, were grown in Dulbecco modified Eagle medium supplemented with 10% fetal calf serum, 10 mM HEPES, 50 U of penicillin per ml, 50 ,ug of streptomycin per ml, 2 mM glutamine, and lx hypoxanthine-aminopterin-thymidine supplement (GIBCO). K562 cells were maintained in RPMI 1640 supplemented with 10% fetal calf serum, 10 mM HEPES, 50 U of penicillin per ml, 50 ,ug of streptomycin per ml, and 2 mM glutamine. HUVE cells were grown on collagen-coated plates in medium 199 supplemented with 20% fetal calf serum, 25 ,ug of endothelial cell growth factor (Sigma Chemical Co.) per ml, 0.1 U of heparin per ml, 25 mM HEPES, 50 U of penicillin per ml, and 50 p,g of streptomycin per ml. All cells were plated in 6-cm-diameter dishes at least 24 h before transfection. DNA transfections of BAE, HeLa, HepG2, NIH 3T3, EA.hy926, and HUVE cells were performed by the calcium phosphate precipitation technique (3). The cells (approximately 80% confluent) were transfected with 10 ,ug of plasmid DNA. The cells were washed 16 h later, and fresh medium was applied. Samples of the medium were removed 48 to 64 h later and assayed for growth hormone by using a commercially available radioimmunoassay kit (Nichols Institute). K562 cells were transfected by electroporation as previously described (23). Samples of the K562 cell medium were assayed for growth hormone 72 h after transfection. Nuclear extracts. Extracts from BAE, NIH 3T3, HepG2, EA.hy926, and HUVE cells were prepared by the method of Schreiber et al. (32). Nuclear extracts from K562, MEL, and

NONERYTHROID GATA-BINDING PROTEIN

4855

HeLa cells were prepared by the method of Dignam et al. (6). Protein assays on the purified nuclear extracts were performed with Bradford reagent (Bio-Rad Laboratories). Gel shift assays. Gel shift assays were performed as previously described (23). Incubation mixtures contained 2 ,ug of poly(dI-dC), 5,000 to 10,000 cpm of end-labeled DNA fragment (approximately 3 x 105 cpm/ng), and 2 to 6 ,ug of nuclear extract in a total volume of 20 ,ul. For competition experiments, a 200- to 300-fold molar excess of unlabeled double-stranded oligonucleotide was added. DNA-protein complexes were resolved from unbound DNA on 6% polyacrylamide gels run in 0.5x TBE (90 mM Tris-borate [pH 8.2], 2.5 mM EDTA) buffer. Methylation interference assays. End-labeled DNA probe was partially methylated with dimethyl sulfate (25) and then used in preparative-scale gel shift assays. Following PAGE, DNA-protein complexes were identified by autoradiography and eluted from the gel. The labeled DNA was purified with Elutip-d columns (Schleicher & Schuell, Inc.) and then subjected to Maxam-Gilbert chemical cleavage (25). Protein-DNA cross-linking. A DNA fragment (positions -141 to -127) of the PPET-J promoter was cloned into the HindlIl site of Bluescript pKS+II (Stratagene Inc.) in both the forward and reverse orientations. To generate a crosslinking probe labeled on both strands, a 1:1 mixture of plasmids containing the forward and reverse orientations was denatured, labeled with [a-32P]dCTP plus bromouridine 5'-triphosphate (BUdR) by using the Klenow fragment and an M13 -40 primer (22), and then digested with HindIII. The resultant probe was isolated by PAGE and incubated for 15 min with BAE cell nuclear extract under the conditions described for gel shift assays. The incubation mixture was then exposed to a UV light on a transilluminator to cross-link the DNA and protein. Following the addition of MgCl2 and CaCl2, the samples were digested with DNase I (22). Labeled proteins were identified by sodium dodecyl sulfate (SDS)-PAGE (20) and autoradiography. RNA isolation, Northern blotting, and hybridization. Total cellular RNA was isolated by the guanidinium thiocyanatephenol-chloroform method (4) and then size fractionated by electrophoresis (20 ,ug per lane) through 1% agarose gels containing 0.66 M formaldehyde, 20 mM MOPS (morpholinepropanesulfonic acid), 5 mM sodium acetate, and 1 mM EDTA (pH 7). The RNA was then transferred to a Magnagraph hybridization membrane (Micron Separations, Inc.) in lOx SSC (1.5 M NaCl, 015 M sodium citrate). The membrane was baked at 80°C for 2 h, prehybridized in 50% deionized formamide-5x Denhardt solution (0.04% Ficoll, 0.04% polyvinylpyrrolidone, 0.4% bovine serum albumin)-5x SSPE (0.9%o NaCl, 60 mM sodium phosphate [pH 7.4], 6 mM EDTA)-0.1% SDS-25 ,ug of boiled sonicated salmon sperm DNA per ml at 42°C for >4 h, and then hybridized in 50% formamide-2 x Denhardt solution-5 x SSC-0.1% SDS-25 p.g of boiled salmon sperm DNA per ml-32P-labeled cDNA (50 x 106 cpm) at 42°C for 24 h. After hybridization, the membrane was washed once in 2x SSC0.5% SDS at room temperature, twice in 1 x SSC-0.5% SDS at 37°C, and then once in 0.1x SSC-0.5% SDS at 37°C. After autoradiography, the membrane was stripped and reprobed with 32P-labeled cDNA for chicken a-tubulin to confirm the presence of intact mRNA. RNase protection assays. A DNA fragment containing PPET-J-flanking sequence (positions -54 to +145) linked to a portion of the first exon of the growth hormone gene was synthesized by PCR with pPPET-1-GH as a template and specific primers. This fragment was cloned into the Sall site

4856

WILSON ET AL.

MOL. CELL. BIOL.

(J)

49o... M0 -377-If

GH

-45?

141

OH

-12? -4

'4 'OO

-2-EM.*.3

GH

0 +145 -400 -300 -200 - 400 bp. -4500 -1400" -500 FIG. 1. 5' deletion analysis of the human PPET-J promoter. BAE cells were transfected with plasmids containing PPET-1 5'-flanking sequence fused to the human growth hormone gene as a reporter. Horizontal bars indicate the segment of the PPET-J promoter present in each construct. Growth hormone (GH) production is graphed as a percentage, with 100% equal to 25 ng of growth hormone per ml of culture medium. Results are reported as the means ± standard deviations of four to six independent transfections.

of Bluescript pKS+II. The Bluescript vector was linearized with XhoI and used to synthesize a 32P-labeled riboprobe with T7 polymerase and a commercially available kit (Stratagene). RNase protection assays were performed by a modification of a previously described method (26). The uniformly labeled RNA probe (250,000 cpm) was incubated overnight at 37°C with 25 ,ug of total cellular RNA in 30 mM PIPES [piperazine-N,N'-bis(2-ethanesulfonic acid)] (pH 6.4)-300 mM NaCl-1 mM EDTA-80% formamide in a total volume of 85 ,ul. This was followed by the addition of 350 ,ud of 10 mM Tris (pH 7.5)-5 mM EDTA-300 mM NaCl-40 ,ug of RNase A per ml-2 ,ug of RNase T1 per ml and incubation at 37°C for 15 min. RNase was inactivated by the addition of 10 ,ll of 20% SDS and 5 ,ul of proteinase K and incubation at 37°C for 15 min. After phenol-chloroform extraction and ethanol precipitation, the samples were electrophoresed on denaturing 5% polyacrylamide gels. Antibody preparation. Antiserum against GF-1 was prepared by immunizing rabbits with a synthetic peptide corresponding to the C-terminal region of murine GF-1 chemically cross-linked to keyhole limpet hemocyanin. Immunoblotting and gel shift analysis confirmed that this antiserum specifically recognizes murine GF-1 (S. Tsai and S. H. Orkin, unpublished observations). Details of the antiserum preparation and characterization will be presented elsewhere. RESULTS Expression of PPET-1 in various cultured ceUs and cell lines. Previous studies demonstrated that PPET-J is expressed in cultured porcine aortic endothelial cells (36), BAE cells (8), and HUVE cells (8). In preliminary experiments, we surveyed PPET-1 expression in a variety of cell lines by Northern blot analysis. mRNA hybridizing to a human PPET-1 cDNA probe was present in BAE, HUVE, and

EA.hy926 (a hybrid of HUVE cells and human lung carcinoma cell line A549) cells (data not shown). No hybridization was seen to RNA from HeLa, HepG2, or NIH 3T3 cells. Thus, PPET-1 is preferentially expressed in cells of endo-

thelial origin. Deletional analysis of the human PPET-1 promoter. To identify cis-acting sequences essential for PPET-J expression, plasmids containing 5'-flanking sequence from the human PPET-J gene fused to the human growth hormone gene as a reporter were transiently expressed in BAE cells. BAE cells were the preferred type of endothelial cell for the transfection experiments because these cells appeared to transfect with greater efficiency than HUVE or EA.hy926 cells. 5' deletion analysis of the fusion plasmids is shown in Fig. 1. Constructs containing the proximal 157 base pairs of the PPET-1 promoter exhibited nearly maximal expression of the reporter gene. In contrast, those containing the proximal 114 base pairs of the promoter had markedly reduced expression. Transfection of HUVE or EA.hy926 cells with this series of fusion plasmids produced a similar pattern; expression decreased dramatically between positions -157 and -114 (data not shown). RNase protection studies demonstrated that RNA expressed during the transient transfection of BAE cells was correctly initiated at position +1 of the PPET-J gene in the growth hormone fusion plasmids (Fig. 2). To further define cis-acting sequences essential for expression, internal deletions in the human PPET-J promoter were made (Fig. 3). Deletions distal to position -141 were found to have little effect on expression, but deletions between positions -141 and -127 reduced expression. While the region between -141 and -127 appeared necessary for full expression, it was not sufficient; a fusion plasmid spanning -157 to +145 but containing an internal deletion between

VOL. 10, 1990

-54

r

10-X.i IGH

LIJ

w 1

4 AAAAA

mRNA

275 nt

probe

164 nt

protected fragment

+145

PPET-1

4857

NONERYTHROID GATA-BINDING PROTEIN

2 3

4275

w. * _

'.F X

ig_. :''

fS

-4164

!

-;

'.

B s

r:

x

',,. -J-

X i:t v

. b1'>,

:& 9

FIG. 2. RNase protection assay. RNA was harvested from BAE cells transfected with a plasmid containing a portion of the PPET-1 promoter (positions -141 to + 145) fused to the human growth hormone (GH) gene. RNase protection assays were performed as described in Materials and Methods. The sizes (in nucleotides [nt]) of the riboprobe and protected fragment are illustrated. Lane 1, Intact riboprobe; lane 2, control experiment in which no RNA was added to the protection assay; lane 3, fragment protected by RNA from the transfected cells.

-114 and -88 exhibited decreased expression in the transfection assay. This suggests that a cis element or elements present in the region between -141 and -127 act in concert with nmore proximal elements to produce full transcription activity. Clustered point mutations between positions -141 and -127 of the PPET-1 promoter were introduced. When transfected into BAE cells, these mutated promoters directed reduced expression of the reporter (Fig. 4). We conclude that DNA sequences from -141 to -127 of the PPET-J promoter are critical for its activity. A trans-acting factor binds to a (T/A)TATC(T/A) motif located in the PPET-1 promoter. Gel shift assays were performed to identify nuclear proteins specifically recognizing sequences from positions -141 to -127 of the promoter. In initial experiments, an end-labeled DNA fragment spanning the region from -157 to -114 was incubated with BAE cell nuclear extract in the presence of specific and nonspecific competitor DNA. A DNA-protein complex was obGH Expression (ng/mIl GM -D--7 24 -

Q7, Lb.i7

22

served in the absence of specific competitors (Fig. 5, lane 1). Addition of excess unlabeled DNA spanning -157 to -114 blocked formation of the labeled DNA-protein complex (lane 2). An unlabeled DNA fragment extending from -141 to -127 was then prepared and found to be an equally effective competitor (lane 3). Formation of the complex was not blocked by a heterologous competitor (lane 4). When end labeled and used as a probe, the DNA fragment spanning -141 to -127 also yielded a DNA-protein complex on gel shift analysis (lanes 5 to 8), confirming that this segment of the promoter is recognized in a sequence-specific manner. To define those residues critical for protein-DNA interaction, methylation interference analysis was performed. The DNA-binding protein present in BAE cell nuclear extract specifically contacted the motif TTATC (GATAA) located at positions -136 to -132 (Fig. 6). The pattern of DNA-protein

GH Expression (ng/ml)

GATTTCCTTTCGGGCCTGGCCTTATCTCCGG ...............

ACCG ...

.GAAT ....................

AO

I

i

1[.I i O;D-.--: --

FIG. 3. Internal deletion analysis of the human PPET-1

pro-

moter. BAE cells were transfected with plasmids containing PPET-1 5'-flanking sequence fused to the human growth hormone (GH) gene as a reporter. The horizontal bars indicate the positions of the internal deletion in each construct. Results are reported as the means of three to six independent transfections. Statistical analysis

revealed standard deviations of '20%.

...

Wild Type

20

Mutation

A

3

Mutation

B

3

Mutation

C

3

AGAGG ..... ........................

FIG. 4. Effect of clustered point mutations on activity of the human PPET-1 promoter. Growth hormone (GH) fusion plasmids containing clustered point mutations in the PPET-1 5'-flanking region were prepared by PCR using pPPET-1-GH as a template and synthetic oligonucleotide primers containing the desired nucleotide substitutions. Each construct extended from positions -157 to + 145 of the PPET-l gene. The locations of the clustered point mutations and the (T/C)TATC(T/A) motif (see legend to Fig. 6) are indicated. The fusion plasmids were transfected into BAE cells, and growth hormone expression was measured. Results represent the means of two independent transfections.

4858

MOL. CELL. BIOL.

WILSON ET AL. 1

2

3

4

5

6

7

8

2

1

1 2 A_

B

urn

n0 I -136

-135 -133

134

-132

FIG. 5. Gel shift analysis of the human PPET-1 promoter using two overlapping DNA probes. BAE cell nuclear extract was incubated with an end-labeled DNA fragment spanning either positions -157 to -114 of the PPET-J promoter (lanes 1 to 4) or positions -141 to -127 of the PPET-I promoter (lanes 5 to 8) in the presence of the following unlabeled specific competitor DNA fragments: lanes 1 and 5, no competitor; lanes 2 and 6, a DNA fragment spanning positions -157 to -114 of the PPET-1 promoter; lanes 3 and 7, a DNA fragment spanning positions -141 to -129 of the PPET-I promoter; and lanes 4 and 8, a 40-mer from the 5'-flanking region of the human thrombomodulin gene containing no homologous sequence. The position of the major protein-DNA complex is indicated with an arrowhead. Smaller protein-DNA complexes that compete in a parallel fashion are presumed to be proteolysis fragments of the major protein-DNA complex.

_

~~~~

GGCCTTAT.CTCCuG C X~~~~~

r.CCGGAAluTA5AGCC

B

~l aw

C GGA AG AG G~.S. A; CCAeT C'

contacts is remarkably similar to that obtained with the erythroid cell transcription factor GF-1 (also called NF-E1, Eryfl, EF-1, and EF-ya), which recognizes a core motif of TATC (GATA) within a broader consensus sequence of the form (T/C)TATC(T/A) (11, 28). Comparison of the clusteredmutagenesis data (Fig. 4) with the methylation interference data suggests that the sequence flanking this core motif may contribute to binding of the factor in BAE cell nuclear extract; specifically, the fusion plasmid containing mutations from -141 to -138 (mutation A in Fig. 4) exhibited decreased expression even though the TTATC motif was intact. This observation parallels a finding with the erythroid factor in that not all DNA sequences conforming to the minimal consensus element constitute high-affinity binding sites (11). That the DNA-binding proteins in BAE cell nuclear extract and GF-1 have quite similar binding specificities was further established through gel shift experiments. Several double-stranded oligonucleotides previously assessed for their binding to GF-1 (23, 34) were used as competitors in gel shift assays. The competitor oligonucleotides were added in a large molar excess (200-fold). A gel shift experiment using these competitors and a labeled PPET-J promoter probe (-157 to -114) is illustrated in Fig. 7. A fragment from the Ay-globin gene promoter effectively inhibited binding of the PPET-J promoter to the BAE cell nuclear protein (lane 4). A mutated version of the A-y-globin promoter fragment with reduced binding to GF-1 did not inhibit binding (lane 5). Similarly, a fragment of the A-y-globin 3' enhancer was an effective competitor (lane 6), whereas a mutated version of this sequence was not (lane 7). Sequences from the human 3' ,B enhancer, the chicken 3' fi enhancer, and the murine a 5' promoter, all of which contain a (T/C)TATC(T/A) motif recognized by GF-1, also were effective competitors (data not shown). Additional gel shift experiments using GF-1

~

O

FIG. 6. Methylation interference analysis of the PPET-J factor. A DNA fragment spanning positions -154 to -114 of the PPET-1 promoter was selectively 5'-end labeled on the top (A) or bottom (B) strand. The fragment was then partially methylated and used as a probe in a gel shift assay with BAE cell nuclear extract. Bound DNA was eluded from the protein complex and subjected to MaxamGilbert chemical cleavage. (A) Lane 1, A>G cleavage of DNA recovered from the gel shift complex; lane 2, A>G cleavage of unbound DNA. (B) Lane 1, G>A cleavage of DNA recovered from the gel shift complex; lane 2, G>A cleavage of unbound DNA. (C) Summary of the methylation interference pattern for the PPET-1 promoter. (D) Previously published (23) methylation interference pattern for GF-1 and the Ay-globin promoter. Sites of methylation interference are indicated with dots. The (T/C)TATC(T/A) motifs are enclosed in boxes.

(from crude extracts of erythroleukemia cell nuclei or COS cells expressing cloned murine GF-1 cDNA) and the labeled PPET-J promoter probe produced similar results; a gel shift complex was observed between GF-1 and the PPET-1 promoter DNA, and formation of this complex was blocked by competitors containing an intact (T/C)TATC(T/A) motif (D. I. K. Martin and S. H. Orkin, unpublished observations). In summary, these competition experiments confirm that the protein in BAE cell nuclear extract which binds to the PPET-J promoter shares DNA-binding specificity with the erythroid factor GF-1 and its homologs. Photoaffinity cross-linking of the DNA-binding protein. The Mr of the DNA-binding protein recognizing the PPET-J promoter was determined by BUdR photocross-linking. In the absence of competitor DNA, two bands were observed: a major species with an Mr of 47,000 and a less prominent

NONERYTHROID GATA-BINDING PROTEIN

VOL. 10, 1990

1

2

3 4

5

6

7

1

2

4859

3

-200 493

'69 Mr

-46

-.30

FIG. 7. Oligonucleotides containing the (T/C)TATC(T/C) motif compete for formation of a gel shift complex. A group of oligonu-

cleotides with previously characterized GF-1-binding activity were used as competitors in the gel shift assays with BAE cell nuclear extract and an end-labeled DNA fragment spanning positions -157 to -114 of the human PPET-1 promoter. The competitors were present in a 200-fold molar excess. The position of the DNA-protein complex is indicated with an arrowhead. The following competitors were used: lane 1, no competitor; lane 2, unlabeled PPET-1 promoter fragment spanning positions -155 to -114; lane 3, heterologous 40-mer from the 5'-flanking region of the human thrombomodulin gene; lane 4, Ay-globin promoter fragment (-193 to -168) containing the sequence ACACTATICTCAATGCAAATATCTGT; lane 5, mutated A-y-globin promoter fragment (-193 to -168) containing the sequence ACACTATATCAATGCAAATATATGT; lane 6, A-y-globin enhancer fragment containing the sequence CCTAAGAAAGACTATCTCAGTCGAATCTG; and lane 7, mutated Ay-globin enhancer fragment, containing the sequence

CCTAAGAAAGACTATATCAGTCGAATCTG.

band with an Mr of 30,000, possibly a proteolytic fragment of the larger band (Fig. 8, lane 2). In the presence of specific competitor DNA, neither band was seen (lane 1). To confirm that the 47,000-Mr band corresponds to the DNA-binding protein responsible for the gel shift complex, the BUdR probe was incubated with BAE cell nuclear extract and cross-linked with UV light, and the mixture was then subjected to native PAGE under the conditions used for gel shift assays. The DNA-protein complex comigrating with the gel shift complex was eluted, digested with DNase I, and then subjected to SDS-PAGE. A 47,000-Mr labeled protein was again seen (lane 3). The size of this labeled protein is similar to the size of GF-1 (approximately 48 kilodaltons [kDa]), as determined by protein purification (34), expression of recombinant protein (34), and Western immunoblot analysis of MEL cell nuclear extracts with antiserum to GF-1 (Tsai and Orkin, unpublished observations). The DNA-binding protein recognizing the PPET-1 promoter is present in NIH 3T3 and HeLa cells. To determine the tissue distribution of the protein binding to the PPET-J promoter, gel shift assays were performed on extracts from different cell lines (Fig. 9). A protein-DNA complex with similar mobility was observed with nuclear extracts from BAE, human cervical carcinoma (HeLa), and mouse fibroblast (NIH 3T3) cells. In each instance, formation of the proteinDNA complex was blocked by addition of specific competitor DNA. Nuclear extracts from HUVE and EA.hy926 cells yielded protein-DNA complexes with faster mobility, possibly because of proteolysis of the DNA-binding proteins in

FIG. 8. BudR cross-linking of the DNA-binding protein in BAE cell nuclear extract. A BUdR- and 32P-labeled DNA fragment from positions -141 to -127 of the human PPET-1 promoter was incubated with BAE cell nuclear extract in the presence (lane 1) or absence (lane 2) of a 300-fold molar excess of specific competitor (fragment consisting of positions -157 to -114 of the PPET-J promoter), cross-linked with UV light, and then digested with DNase I. The resultant mixture was subjected to SDS-PAGE. Arrowheads indicate the labeled proteins migrating at 47 and 30 kDa. To confirm that the 47-kDa protein corresponds to the DNAbinding protein responsible for the gel shift complex, the BUdRlabeled probe was incubated with BAE cell nuclear extract, crosslinked with UV light, and then electrophoresed on nondenaturing polyacrylamide gels. The band corresponding to the protein-DNA complex was eluted, treated with DNase I, and then subjected to SDS-PAGE (lane 3). Molecular weight markers (in thousands) are shown at the right.

the crude nuclear extract preparations of these cells (data not shown). We were unable to use gel shift analysis to detect the presence of this DNA-binding protein in erythroleukemia cells (K562 and MEL) because GF-1, which is present in abundance in nuclear extracts of these cells, binas 1

2

3

4

5

6

7

8

.-L

_tt!

FIG. 9. The DNA-binding protein recognizing the PPET-J promoter fragment spanning positions -141 to -127 is present in BAE, NIH 3T3, and HeLa cells. Gel shift assays with end-labeled PPET-J fragment were performed with nuclear extracts from BAE (lanes 1 to 3), HeLa (lanes 4 to 6), and NIH 3T3 (lanes 7 and 8) cells. The incubation mixtures contained the following competitor DNAs in a 200-fold molar excess: lanes 1, 4, and 7, no competitor; lanes 2, 5, and 8, unlabeled PPET-1 promoter fragment spanning positions -157 to -114; and lanes 3 and 6, a 40-mer from the 5'-flanking region of the human thrombomodulin gene, containing no homologous sequence. The position of the major protein-DNA complex is indicated with an arrowhead.

4860

WILSON ET AL. 1

2 3 4

MOL. CELL. BIOL.

TABLE 1. Cell-specific expression of the human PPET-J gene

5 6 7 8 9 1011