Regulation of the erythroid Kruppel-like factor (EKLF) gene promoter ...

T H E JOURNALOF BIOLOGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biolom, Inc.

Vol. 269, No. 22, Issue of June 3, pp. 15440-15444, 1994 Printed in U.S.A.

Regulation of the Erythroid Kruppel-like Factor(EKLF) Gene Promoter by the Erythroid TranscriptionFactor GATA-l* (Received for publication, February 4, 1994)

Merlin CrossleylO, Alice P. Tsangl, James J. Biekefl, and Stuart H. Orkin+O** From the $Division of Hematology-Oncology, Children's Hospital, Dana-Farber Cancer Institute, Department of Pediatrics,

Harvard Medical School, Boston, Massachusetts 02115, the $Howard Hughes Medical Institute, a n d the IBrookdale Center for Molecular Biology a n d Department of Biochemistry, Mount Sinai School of Medicine, New York, New York 10029

Erythroid Kruppel-like factor (EKLF) is an erythroid- either downstream of (or coincident with) GATA-1 in the hierspecific transcription factor that binds aCACCC motif archy of regulators in erythroid cells. found in the human &globin gene promoter. We have MATERIALSANDMETHODS studied the promoter of theEKLF gene and identified binding sites for the transcription factors GATA-1 and Cloning of the EIUF Genomic Sequences and Construction of Proto (8) a CCAAT-bindingProtein 1 (CP1).We show that both typesmoter Plasmids-Radiolabeled EKLF cDNA(1) was used screen of binding sites are required for full activity, and that mouse genomiclibrary prepared in A FIX I1 (Stratagene, La Jolla, CA). the GATA motif at -60 is essential. The EKLF promoter A clone encompassingthe EKLF gene and its5'-flanking sequenceswas can be directly activated in nonerythroid cells in co- recovered as a 12-kilobase Sal1 fragment. Sequence analysis (9) using various primers (including MC173, described below) revealedthat this transfection experiments by forced expression of fragment contained sequences 5' to the start of the EKLF cDNA. GATA-1. These results suggest that EKLF is dependent Oligonucleotidesand Plasmid Constructions-'ho primers MC171, on GATA-1 for its expression andlies downsteam of, or ATTAAGCTTCATACAGGGTTGGT?"IVTCAG, and MC173,GCTCTAcoincident with, GATA-1 in a regulatory hierarchy in GAAGGCCTCCTCTCTCTCTTCTGCCTTATGGGC,were usedtogether erythroid development. with Pfu polymerase (Stratagene) to amplify(10)the region -353 to +34 of the EKLF gene. These primers have sites for HindIII and XbaI, respectively, at their ends to facilitate cloning into the HindIII, XbaI sites immediately upstream of the human growth hormone reporter Erythroid Kruppel-like factor (EKLF),' a zinc finger DNA- gene, in theplasmid pOGH (11).A single G to T mutation at +10 in the 5'-untranslated region was introduced in order to disrupt a potential binding protein related to the Drosophila transcription factor Kruppel, is expressed selectively in erythroid tissues of the upstream initiation codon whichlies between the two majorstart points of transcription, see Fig. 1.Primers MC187, AATAAGCTTGCACACCAmouse and immortal erythroidcell lines, suchas murine erythCACATATCGCA, and MC172, AATAAGCTTGCCTGGGTCTTATCAGG roleukemia (MEL) cells (1).Although the full spectrum of se- were usedtogether with MC173 to similarly construct the -123 and -77 quences to which it bindshas not been defined, EKLF recogdeletion constructs, respectively. Primers MC188,AATAAGCTTGCACACCACACATA'M'GCACA, MC174,ATTAAGCTTGCCTGGGTCmnizes the CACC box motif in the human P-globin promoter. Moreover, mutations in theCACC box motif which are associ- UGGGA, and MC224, AATAAGCTTGCCTGGGTCTTATCAGGated with p-thalassemia prevent EKLF binding (2). Together, GAAGACAGCCAATGAAATGT, were used with primer MC173 to specifically mutate the GATA sites in sites 3, 2, and 1, respectively, while these findings suggest that EKLF participates in globin gene primer MC189, AATAAGCTTGCCTGGGTCTTATCAGGGAAGAC&transcription in erythroid cells, and perhaps in the control of W C A G , was used with MC173 to mutate the CCAAT box in site 1. other red cell genes whose promoters containCACC boxes. Throughout this section relevant (or on antisense the strand) and CCAAT sequences have been underlined and specific muWhether EKLF plays a pivotal role in initiating erythroid differentiation oris a downstream transcription factor primar- tations, designed to disrupt these elements, are shown in bold.All constructs were sequenced to exclude cloningartifacts. ily involved in the regulation of genes, such as globins, which The sequence of double stranded oligonucleotides used as probes in are expressed later in erythroid maturation, is unknown. As an the gel shift analyses, wereMC183,CTGGGTCTTATCAGGGAAGAapproach to addressing these issues, we have studiedthe EKLF CAGC, the site 2 GATA probe (Fig. 4) and MC185, AAGACAGCCAATgene promoter. Herewe show that the promoter contains three CAGATGTGGGCA, the site 1 CCAATbox probe (Fig. 5). Competitor oligonucleotides were as follows,forFig. 4,MC183 and "2185, debinding sites for the essential erythroid transcription factor GATA-1 (3, 4)and one site for t h e CCAAT box binding factor scribedabove, and MC73, GATCTCCGGCAACCCTTTAAGGATTCCP1 (5-7). Of the three GATA motifs one is essential for pro- CCTG, and MC210, ACCACACATATCGCACACACCCCT. For Fig. 5 , the rat albumin CP1 site was MC214, GGGTAGGAACCAATGAAATmoter activity. Fromthese results we infer that the EKLF gene GAAAGGTT, and the other competitors were MC212, AAGACmGpromoter is likely tobe a target of GATA-1. Hence, EKLF lies WCAGATGTGGGCA, and MC216, AAGACAGCCAATGAAATGTGGGCA. Pansient Dansfections and Pansactivation Experiments-All DNA *This workwas supported in part by a grant from the National samples tested were prepared over Qiagen columns (Qiagen Inc., CAI, Institutes of Health (to S. H. 0.). The costs of publication of this article and at least threepreparations of each plasmid were tested. MEL cells were defrayed in part by the payment of page charges. This article must were transfected by electroporation and growth hormone levels assayed therefore be hereby marked "advertisement" in accordance with 18 after 60 h (12). For transactivations, NIH3T3 cells weretransfected by U.S.C. Section 1734 solely to indicate this fact. The nucleotide seguence(s) reportedin this paper has been submitted calcium phosphate precipitation. 10 pg of reporter construct was COto the GenBankmIEMBLData Bank with accessionnumber(s1 UO8199. transfected with 10 pg of GATA-1 expression plasmid pXMGATA-1 (3, ** Investigator of the Howard Hughes Medical Institute. To whom 13) or 10 pg of vector DNA. Growth hormone levels wereassayed after 48 h and thecombined results of three separate experiments is shown. correspondenceshould be addressed: Div.ofHematology-Oncology, Gel Shift Analysis and DNase I Footprinting-Nuclear extract was Children's Hospital, 300 Longwood Ave., Boston, MA 02115. Tel.: 617prepared from MELcells as previously described(14).Binding reactions 735-7910; Fax: 617-735-7262. The abbreviations used are: EKLF, erythroid Kruppel-like factor; were carried out at 4 "C in 20 pl of 10 mM Hepes, pH7.8, 50 mM potassium glutamate, 5 m MgCl,, 1mM EDTA, 1nm dithiothreitol, 5% MEL, murine erythroleukemia.

15440

GATA-1Activates the EKLF Promoter -340

15441

homologous oligonucleotide (lanes 5 and 61, but not with a related oligonucleotide in which the GATA motif has been mutated (lanes 7 and 8).We used antiserum raised against mouse -300 -290 -280 -270 -260 TTCCTTGAAA GTGACACATC ACAACTTTAA TCCCAGCACT CAGGAGACAG GATA-1 to determine whether this species was indeed GATA-1. The antiserum blocked DNA binding (lane 31, whereas preim-250 -240 - 2 3-02 1 0 -220 AGGTAGGCAG ATGTCTGTGA GTTCTAGGCC AGCCTGGTCC ACATAAGCCA mune serum (lane 2 ) had no effect. These data indicate that GATA-1 binds in vitro to site 2. -200 -190 -180 -170 -160 GTCCAGGACA GCCAGGGCTC AGTTACACAG AGAAACCCTG TTTCGAAAAC Site 3 alsocontains a GATA motif, CGATAT (on the antisense strand), which does not conform to thepreferred GATA-1 bind-150 -140 -130 -120 site 3 CTTAAAAAAA AAAAAAAAAA AAAAGACGCA CACCXCACAC ATATCGCACA ing site((T/A)GATA(A/G))(15,17, 18).To test whether GATA-1 recognizes this site, an oligonucleotide containing site 3 was -100 -90 -80 -70 site2 - 6 0 ~ ~ C C C C T C CTTGCCGTTTT T GCTTTG~CTGG G T C G ~G G ~ G A C A G used asa cold competitor in the above assay. As lanes 9 and 10 show, site 3 competes for binding but toa lesser extent than site -50 site 1 -30 -20 -10 CCAATCAGAT G ~ G G C A G A CAGGAGCCCTC CAAGAAACTT TCCTAGCCTC 2. Thus, site 3 also binds GATA-1, albeit somewhat weakly. Site 1 was also used in gel shift assays with MEL extracts +34 + +11 e+21 ATAGCCCATG AGGCAGAAGA GAGAGAGGAG GCCT (Fig. 5 ) . Three complexes, which competed with homologous FIG.1. Sequence of the B’-flankingregion of the EKLF gene. competitor (lanes 2 and 3 ) , were observed. Since site 1 contains The sites protected fromDNase I (see Fig. 3) are boxed above the a perfect consensus sequence for the ubiquitous transcription sequence. GATA sequences (or TATC on the other strand) are under- factor CP1 ((A/G)(A/G)CCAAT), we tested whetherany of these lined and a CCAATbox is ouerlined. The major start points of tranAn oligonucleotide containing complexes was formed by CP1. scription (1) are shown by arrows. Numbering is from the most upthe CP1 sitefrom the rat albuminpromoter ( 5 ) was used as a stream transcriptional start point, defined as + l . cold competitor and competed for the upper complex (compare lanes 6 and 7 with lane I). These data suggest that theupper glycerol, and 1 pg of poly(d1-dC),with 0.5 ng of 32P-labeledoligonuclecomotide probe and 5-10 pg of nuclear extract (151. Competitor oligonucle- complex is formed by CP1. We also demonstrated that the otides were added prior t o the probe at molar excessof 40- or 200- fold. plex can be competed with the CP1 sitefrom the Eagene promoter (6) and that it comigrates with the complex formed on 1p1 of rat serum or anti-GATA-1rat serum, kindly provided by Lin Gu and Doug Engel, was added, where indicated, during the binding reac- this site (data notshown). tion. The resulting complexes were then separated by electrophoresis Although site 1 does not contain a classical GATA motif, it through 6% polyacrylamide gels at 4 “C. does contain thesequences AGATGT (on the sense strand) and Primer MC173 was labeled with 32Pusing polynucleotide kinase, TGATTG (on the antisense strand), both of which have been then used together with primer MC171 to amplify the region of the shown to bind GATA-1 (15, 18). We therefore carried out exEKLF promoter -353to +34.This resulting fragment was then used in periments t o test whether either of the two remaining coma footprinting reaction as previously described (16), except that the binding buffer described above was used for the binding reaction. plexes was formed by GATA-1. As shown in Fig. 5 the lower complex is competed by an oligonucleotide containing the seRESULTS quence GATA (lanes 4 and 51, but not by a related oligonucleThe EKLF Promoter Is Active in Erythroid Cells-As a first otide in which the GATA site has been mutated (lanes 12 and step towards identifying the sequences which control the ex- 13) and binding is blocked by the antiGATA-1 antiserum (lane pression of EKLF, we isolated genomic clones encompassing the 151, but not by preimmune serum (lane 14).We therefore conEKLF gene and its 5”flanking sequences. The DNAsequence of clude that thelower complex is formed by GATA-1.The identity the region immediately upstream of the transcription start site of the middle complex, which did not compete with either the is shown in Fig. 1. We next sought to determine whether the GATA-1 or CP1 oligonucleotide, is unknown. region from -353 to +34 is sufficient to direct transcriptionof a GATA-1 Is the Major Activator ofthe EKLF Promoter-To asreporter gene (human growth hormone) upon introductioninto sess the relative contributions of these transcriptionfactor binderythroid cells. As shown in Fig. 2, sequences from -353 to +34 ing sitesto promoter function,a series of mutant constructs was wereactive in MEL cells. Inclusion of additional upstream tested for promoter activityin MEL cells. First, the relative imsequences (to -607) did not alterpromoter activity (not shown). portance of the three GATA binding siteswas assessed by siteFurther experiments showed that the activity of the EKLF specific mutation. In each instance the G residue of the GATA promoter in transient assays isapproximately half that of the motif was changed to a C or A, as these mutations areknown GATA-1 and herpesvirus thymidine kinasepromoters in MEL to disruptGATA-1 binding. As shown in Fig. 2, mutation of the cells, and that the promoter is inactive in NIH3T3 cells (data weak CGATAT motif in site 3 to CCATAT had only a marginal not shown). effect on promoter activity. Likewise, mutation of the AGATGT The EKLF Promoter contains Binding Sites forGATA-1 and and TGATTG motifs in site 1 to AAATGG and TCATTG, reCPI-To identify binding sites for transcription factors within spectively (mutations which eliminate GATA-1 binding to this the promoter, DNase I footprinting analysis with nuclear ex- site butdo not effect CP1 binding(Fig. 5 , lanes 10 and 11), had tracts of MEL cells was performed. Two clear footprints were only a small effect on promoter activity. However, when the evident (sites 1and 2 in Fig. 3).A third region (site 3) showed TGATAA motif in site2 was mutated toTCATAA, promoter acweaker protection (Fig. 3). Sites 2 and 3 contain consensus is of tivity was reduced t o background, indicating that this site GATA motifs, which are recognized by members of the GATA major importance in the EKLF promoter. family of transcription factors (15, 17, 18).Site 1 contains the To test the contribution of CP1 binding at site 1 to promoter sequence AGCCAAT, which is recognized by CCAAT-binding activity we sought a mutation which would eliminate CP1bindproteins, particularly CP1 (5-7). ing but not disrupt GATA-1 binding to site 1. The sequence To determine which MEL proteins could bind these sites, AGCCAATCAGATGT wasalteredto AGCGTATCAGATGT. oligonucleotides encompassing the footprintedregions were This mutation significantly reduced but did not eliminate CP1 used in gel shift analyses. When site 2, which contains the binding and did not affect GATA-1 binding to this site(Fig. 5 , GATA-1 consensus bindingsite TGATAA (on theantisense lanes 8 and 9). This mutation reduced promoter activity by strand), is used in the assay a single retarded species is ob- %fold, suggesting that CP1does play a positive role at site 1. served (Fig. 4). This species is effectively competed with an Forced GATA-1 Expression Activates the EIUF Promoter in -353 -320 -310 CAT ACAGGGTTGG TTTTCAGGAC CAGTTTGAAA ACAGGCCAGG GTGAGGTCTA

GATA-1 Activates the EKLF Promoter

15442 -353

sitc 3

site 2

site 1

+ +34 100 f 18

I

I

CGATAT

TGATAA

CPI

AGATGT

e +34 e +34

-123

91+ 9

I

-123

76.r 5

I

-17

s7 f 13 +34 I

( ' f .3

+ +34

-11

18+ 6

I

-11

+ +34 I

38+ 3

vector

8 + 4

FIG.2. EKLF promoter activity as assayed by transient transfection in MEL cells. The threeGATA-binding sites areshown by ovals, the critical GATA element at-60 is shaded. The CCAAT protein 1 site is shownby a box. Crosses indicate disruption of specific binding sites, see text of the various constructs and standard errors of three experiments are shown on the right. The activityof the longest for details. The mean activities construct is arbitrarily set a t 100. Vector refers to the growth hormone reporter vector, pOGH (11).

1

2

3

a*" 7-116

J

4

-.

5

6

7 8 a

.

9 1 0 1 1 1 2

. .

0

.

site 3 CGATAT

-98

._ -. r C

F

I"

. .

.-

f-

free probe

FIG.3. DNase I footprinting analysis of the EKLF promoter. The three sites protected by MEL cell nuclear extracts are shown. The protection over site 3 was very weak but reproducible.+ A G represents a Maxam and Gilbert sequencing ladder (30). Footprinting of the restof the region -353to+34revealed no otherprotectedsites, data not shown.

FIG.4. Gel shift analysis of GATA-binding sites in the EKLF promoter. The double stranded oligonucleotide probe used is MC183 and contains the TGATAA site at -60 in site 2. Antisera and double stranded competitoroligonucleotides were addedas follows: lanes 5 and 6,20 and100 ngofMC183; lanes 7 and 8, MC73; lanes 9 and 10, MC210; and tolanes 11 and 12, MC216, see "Materials andMethods" for the full sequence of these oligonucleotides.

Heterologous Cells-We next determined whether forced expression of GATA-1 is able to activate the EKLF promoter in nonerythroid cells in a cotransfection assay. EKLF promoter sequences from -77 t o +34 linked to the growth hormone reporter were introduced into NIH3T3 cells. The promoter had minimal activity in these cells, however, when it was transfected together with an expression vector encoding GATA-1(3) it was activated approximately20-fold (Fig. 6). Similar results were obtained using promoter sequences from -353 to +34 (not shown). Mutation of the critical GATA sequence at site 2 vir-

tually abolished activation (Fig. 6), suggesting that GATA-1 operates largely through site 2 in this assay. These results are consistent with the resultderived from transient transfections in MEL cells, that the GATA motif in site 2 is necessary for EKLF promoter activity, and together the results suggest that GATA-1 acts predominantly through this site. DISCUSSION

Our results indicate that the EKLF promoter contains a functional GATA site at -60 and a CCAAT motif a t -45, and

GATA-1 Activates theEKLF Promoter

15443

GATA family of transcription factors in MEL cells, other members of the family, particularly GATA-2 (19,201, may playa role in EKLF transcription earlier in the erythroid program. Accordingly, wehavealsodemonstratedthat GATA-2, like GATA-1, is able to activate the EKLF promoter in NIH3T3 cells (data not shown). It isprovocative that the EKLF promoter contains GATA and CP1 elements in closeproximity, as thisassociationisalso present inthe human ( (21), human y (22), human P (23), and mouse a-globin gene (24) promoters. Although there is no as yet direct evidence that GATA-1 and CP1 proteins physically interactorotherwisecooperatetoactivatetranscription,this may be the case. Furthermore, although CP1 sites are found in several erythroid-expressed gene promoters, no evidence for an erythroid-specific protein t h a t recognizes this site is available. Therefore, we attributethe erythroid-specific expression of the EKLF gene promoter primarily to theGATA site at -60. Ourevidence is mostreadilyinterpretedasimplicating GATA-1 in the regulationof the EKLF gene. In particular, since the EKLF promoter seems to be dependent on the GATA site it seems unlikely that EKLF would be expressed before GATA the factors are produced during erythroid maturation. Thus we 1 view EKLF as being downstream of GATA-1. This of course free probe does not rule out the possibility of a more complex network in FIG.5. Gel shift analysis of the CCAAT-binding site in the which EKLF reciprocally influences GATA-1 expression. EKLF promoter. The double stranded oligonucleotide probe MC185, GATA-1 expression is believed to increase during cell red matucontains the CCAATbox in site 1. The double strandedcompetitor ration (25, 26) and it is interesting that the GATA-1 gene prooligonucleotides were: lunes 2 and 3, 20 and 100 ng of MC185; lnnes 4 and 5 , MC183; lanes 6 and 7, MC214; lunes 8 and 9, MC212; lunes 10 moter contains two functionalCACCC boxes (12). It is possible and 12, MC216; lanes 12 and 13, MC73, see "Materials andMethods" for that EKLF isinvolved in maintaining or augmentingGATA-1 the full sequence of oligonucleotides. Despite competition with CP1 site expression during red cell maturation. However, we have obcontaining oligonucleotides, there is a residual complex in lanes6, 7, 10, poorly to theCACCC boxes in the and ZI, which co-migrates with CP1. This upper complex is also ob- served that EKLF binds only GATA-1 promoter, and have been unable to demonstrate direct served in Fig. 4,when a TGATAA probe is used a s a probe. activation of the GATA-1 promoter by EKLF in cotransfection assays (data not shown). The available evidence is therefore consistent with the view that EKLFis primarily a target gene of GATA-1. 2s SCL, a basic helixloop helix transcription factor which is also T found in erythroidcells, is similarly dependent ona GATA site in its promoter for expression (27). It is possible that SCL together with EKLF are the first members of a class of transcription factors which are target genes of GATA-1 and are perhaps involved more in the maintenancethan the initiation of erythroid differentiation. The available evidence implicates EKLFas a putative regulator of human P-globin gene transcription, that is, it binds the P-globin CACCC box motifs and mutations which disrupt binding are associated with thalassemia (2). However, the multiplicity of nuclear proteins that recognize CACCC box motifs, including Spl and TEF-1, complicates the interpretation of simple binding and transactivation experiments (28, 29). Re+ ' GATA- 1 cent experiments suggest that EKLF binds with high affinity to a subset of CACCC motifs, generally those with a n extended sequence of the form CCA CAC CCT (data not shown). Further -77 work will be required to identify the putative target genes i34 regulated by EKLF in vivo and to define its precise role in Flo. 6. Transactivation of the EKLF p r o m o t e r by forced exerythroid differentiation.

6

-

I

-

+

T-)lw

pression of GATA-1 in NIH3T3 cells.Activity of the wild type EKLF promoter construct in the absence of GATA-1 was minimal and equivAcknowledgments-We are gratefulto Sabra Goff for her expertise in alent to background activity of the vector control, pOGH (ll),this ac- A cloning and synthesis of oligonucleotides, also to members of the tivity was arbitrarily defined 1. asThe figure showsthe mean and S.E. Orkin lab for many discussions and critical reading of the manuscript. of three experiments. REFERENCES

suggest that GATA-1 and CP1 act from these sites to allow 1. Miller, I. J., and Rieker, J. J. (19931 Mol. Cell. Biol. 13, 277C2786 EKLF expression in MEL cells. The finding that EKLF is de- 2. Feng, W. C., Southwood, C. M., and Hieker, J. J. (1994) J. Riol. Chem. 269, 1492-1500 pendent on a GATAmotif forits expression suggests that it may 3. Tsai, T.F., Martin, D. I. K., Zon, L. I.. DAndrea. A. D., Wong. G . G .and , Orkin, S. H. (1989) Nnture 339,4 4 6 4 5 1 lie below GATA-1 in a hierarchy of regulators which control 4. Evans, T., and Felsenfeld, G. (1989) Cell 58, 877-885 erythroid differentiation. It should be remembered, however, 5. Raymondjean. M.,Cereghini, S., and Yaniv, M. (1988) Proc. Nntl. Acnd. Sci. that although GATA-1 is the most abundant member of the U. S. A. 85, 757-7fil

15444

GATA-1Activates the EKLF Promoter

6. Dorn, A,, Bollekens, J., Staub, A,, Benoist, C., and Mathis, D. (1987) Cell 50, 863-872 7. van Huijsduijnen, R. H., Li, X. Y., Black, D., Matthes, H., Benoist, C., and Mathis, D. (1990) EMBO J . 9, 3119-3127 8. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual, 2nd Ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY 9. Sanger, F., Nicklen, S., and Coulson, A. R. (1977)Proc.Natl. Acad. Sci. U. S. A. 74,5463-5467 10. Saiki, R. K., Scharf, S., Faloona, F., Mullis, K. B., Horn, G. T.,Erlich, H.A., and Arnheim, N. (1985) Science 230,135%1354 11. Selden, R. F., Burke-Howie, K., Rowe, M. E., Goodman, H. M., and Moore, D. D. (1986) Mol. Cell. Bid. 6, 3173-3179 12. Tsai, S.-F., Strauss, E., and Orkin, S. H. (1991) Genes & Deu. 5, 919-931 13. Martin, D. I. K., and Orkin, S. H. (1990) Genes Q Den 4, 18861898 14. Andrews, N. C., and Faller, D. V. (1991) Nucleic Acids Res. 19, 2499 15. Merika, M., and Orkin, S. H. (1993) Mol. Cell. B i d . 13, 39994010 16. Crossley, M., and Brownlee, G. G. (1990) Nature 345, 444-446 17. Orkin, S. H. (1992) Blood 80, 575-581 18. KO, L. J.,and Engel, J. D. (1993) Mol. Cell. Biol. 13, 4011-4022

19. Dorfman, D. M., Wilson, D. B., Bruns, G. A. P., and Orkin, S. H. (1992)J. Bid. Chem. 267, 1279-1285 20. Yamamoto, M., KO,L. J., Leonard, M. W., Beug, H., Orkin, S. H., and Engel, J. D. (1990) Genes & Deu. 4, 1650-1662 21. Watt, P., Lamb, P., Squire, L., and Proudfoot, N. J.(199O)NucleicAcidsRes. 18, 1339-1350 22. Shen, S.-h., Slightom, J. L., and Smithies, 0. (1981) Cell 26, 191-203 23. Macleod, K., and Plumb, M. (1991) Mol. Cell. Biol. 11,43244332 24. Nishioka, Y., and Leder, P. (1979) Cell 18, 875-882 25. Daylot, N., Fibach, E., Ronchi, A., Rachmilewitz, E. A,, Ottolengbi, S., and Oppenheim, A. (1993)Nucleic Acids Res. 21, 4031-4037 26. Sposi, N. M., Zon, L. I., Care, A,, Valtieri, M., Testa, U., Gabbianelli, M., Mariani, G., Bottero, L., Mather, C., Orkin, S. H., and Peschle, C. (1992) Proc. Natl. Acad. Sei. U. S. A. 89, 6353-6357 27. Aplan, P. D., Nakahara, K., Orkin, S. H., and Kirsch,I. R. (1992) EMBO J. 11, 40734081 28. Hartzog, G. H., and Myers, R.M. (1993) Mol. Cell. Biol. 13, 44-56 29. Davidson, I., Xiao, J. H., Rosales, R., Staub, A,, and Chambon, P. (1988) Cell 54,931-942 30. Maxam, A,, and Gilbert, W. (1980) Methods Enzymol. 65,499-560