a serum response element and a binding site for nf-y mediate

THE JOURNAL OF BIOLOGICAL CHEMISTRY Val. 268, No. 7, Issue of March 5, pp. 4989-4996 1993 Q 1993 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in ~. S.A.

A Serum Response Element and a Binding Site for NF-Y Mediate the Serum Response of the Human Thrombospondin 1 Gene*

(Received for publication, September 1, 1992)

Paul Pramson$ and Paul BornsteinStll From the Departments of $Biochemistry and §Medicine, University of Washington, Seattle, Washington 98195

The expression of thrombospondin 1 (TSP l), a member of the TSP gene family, is rapidly induced by growth factors. We tested the ability of human TSP 1- chloramphenicol acetyltransferase constructs to re- spond to serum in stably transfected NIH-3T3 cells. Two transcriptional elements in the TSP 1 promoter, a distal element at -1280 and a proximal element at -65, were required for the response of the human TSP 1 gene to serum. The distal element contains the 5’- CC(A + T)sGG-3’ consensus sequence characteristic of a serum-response element (SRE). Deletions or mutations in this element reduced the serum response of the TSP 1 gene by 80-90%. In gel-shift assays, the -1280 element and the c-fos SRE cross-competed, whereas their functional and binding mutants did not. The proximal element contains the sequence 5”GGCCAATGG G - 3 I , which closely resembles the consensus binding motif for the CCAAT-binding factor NF-Y (CBF, CP1, aCP1). Deletions or mutations in this element also reduced the serum response by 80-90%. Methylation interference analysis of the -65 region identified a pattern of contacts with nuclear factors resembling that for NF-Y, and an NF-Y-binding site and the proximal TSP 1 element cross-competed in gel-shift assays, whereas their binding mutants did not. Finally, an abbreviated TSP 1 promoter/5’-flank, containing the SRE- and NF-Y-binding sites, mediated a serum response that was close in magnitude to that of the parent promoter. We conclude that the serum response of the human TSP 1 gene requires the coordinated function of an SRE- and NF-Y-binding site.

Thrombospondin (TSP)’ is a trimeric secreted protein, originally identified as an abundant constituent of platelet a- granules and now known to be associated with the extracellular matrix and cell surfaces of diverse cell types, including fibroblasts, smooth muscle cells, and glial cells. Functions for TSP have been identified in blood clotting, in cell attachment, migration and growth, in neurite outgrowth, and in control of angiogenesis (for reviews, see Refs. 1 and 2). A role for TSP in growth of smooth muscle cells was first shown by its ability

* This work was supported in part by National Institutes of Health Grants HL 18645 and DE 08229. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accord- ance with 18 U.S.C. Section 1734 solely to indicate this fact.

n Supported by National Institutes of Health Training Grant HL 07312. To whom correspondence should be addressed DeDt. of Bio- chemistry, SJ-70, University of Washington, Seattle, WA 98195. Tel.: 206-543-1789; Fax: 206-685-1792.

The abbreviations used are: TSP, thrombospondin; SRE, serum response element; SRF, serum response factor; CAT, chloramphenicol acetyltransferase; bGH, bovine growth hormone; bp, base pair.

to enhance epidermal growth factor-stimulated DNA synthesis (3,4). Subsequently, anti-TSP monoclonal antibodies were shown to inhibit entry of Go-arrested smooth muscle cells into GI (5). The diversity of functions ascribed to TSP led us to propose that, for some cells, TSP functions as an extracellular morphogen (6). Consistent with this proposal, TSP could play a role in the dramatic changes in cell morphology that coincide with mitogenesis (7). Indeed, the identities of a subset of immediate early genes (see below) suggest a program of mor- phogenetic events that extends from the extracellular matrix to the cytoskeleton. This subset includes (in addition to TSP) the extracellular matrix or cell surface-associated proteins, fibronectin and cyr 61, the integral membrane proteins, tissue factor and the a and p subunits of fibronectin integrin, and the cytoskeletal proteins, @-actin and a-tropomyosin (8, 9).

Recently, TSP has been found to comprise a family of at least three genes: TSP 1, TSP 2, and TSP 3 (1, 10-12). We (13) and others (14) have previously described the TSP 1 promoter. The TSP 1 gene is clearly growth-responsive (15, 16), whereas the TSP 2 gene is either nonresponsive (10) or requires stringent conditions of quiescence to show its responsiveness (17). The response of TSP 3 to growth factors has not yet been reported. This study addresses the basis for the growth responsiveness of the TSP 1 gene.

In accord with its ability to play a role in mitogenesis of some cells, TSP 1 protein (18, 19), mRNA accumulation (15, 16, ZO), mRNA stability (ZO), and transcription (201, are all up-regulated by a number of growth factors. Furthermore, up- regulation of mRNA levels by platelet-derived growth factor (16) and bovine fibroblast growth factor (20) occurs in the presence of cycloheximide, a finding that demonstrates inde- pendence of prior protein synthesis and qualifies TSP 1 as a member of the immediate early gene family. Lau and Nathans (9) have organized the immediate early genes into three groups according to the kinetics of their transcriptional responses to growth factors and subsequent changes in mRNA levels. Accordingly, group I genes show induction of mRNA synthesis within 5 min of growth stimulation. Transcriptional initiation peaks by 10-20 min and mRNA levels peak within an hour. c-fos is a prototypical group I gene. Group I1 gene mRNA levels peak approximately 2 h after stimulation, with peak transcriptional initiation occurring between 1 and 2 h post- stimulation. c-myc is a prototypical group I1 gene. Group I11 genes show changes in transcription and mRNA accumulation that increase more slowly and persist for several hours, beyond the times when groups I and I1 mRNAs have returned to preinduced levels. Fibronectin is prototypical for group 111.

TSP 1 displays characteristics that are intermediate between groups I and 111. Its peak transcriptional initiation, from 10 to 30 min post-stimulation (ZO),’ is characteristic of group I genes. However, its peak mRNA accumulation at 3-4

P. Framson, unpublished data.

4989

4990 Serum Response of Thrombospondin 1

h post-stimulation (15, 16, 20) and its 4-8 h mRNA half-life (20) are characteristic of group I11 genes. The large temporal delay between the peaks of transcriptional initiation and mRNA accumulation for TSP 1 could be a consequence of the long mRNA half-life, of post-stimulation increases in mRNA half-life (ZO), and of the time required for mRNA elongation of an 18-kilobase structural gene (9,21), compared with that for c-fos which is 4 kilobases.

The kinetics-based groupings described by Lau and Na- thans (9) are likely to be augmented in the future by a classification of immediate early genes according to the un- derlying pathways involved in their activation. For example, Lau and Nathans (9) observed that many group I genes, but not all, have either been shown to be transcriptionally regulated through an SRE (22), as are c-fos, @-actin, Krox 24, and Krox 20, or to possess an SRE motif in the sequenced regions of their promoters. Indeed, a sequence closely resembling an SRE has been noted in the promoter for human (13, 14) and mouse (21) TSP 1. In this study we have explored the promoter/5’-flank of human TSP 1 for elements that mediate its response to calf serum. We identified two elements, one of which is, in fact, an SRE, and the other is a binding site for the CCAAT motif-binding factor, NF-Y. These elements in conjunction with the TATA box region, mediate the response of the human TSP 1 promoter to serum.

EXPERIMENTAL PROCEDURES

Plasmids-All plasmid constructions were verified by restriction analysis, and in cases of mutagenesis, by partial DNA sequencing. Numerical designations for TSP 1 sequences are as described in Ref. 13. The parental construct, -2033 CAT, contains the human TSP 1 sequence from -2033 to +750 and has been described (23). The following members of a 5”deletion series were derived by religating vectors derived from -2033 CAT after the following treatment: 1) -1539 CAT, SalI and Bsu36I restriction and Klenow fill-in; 2) -1270 CAT SalI and HindIII restriction and Klenow fill-in; 3) -342 CAT, PstI and Sac11 restriction and T4 polymerase blunting. The SalI site in these constructions originates from the vector polylinker sequence 5’ to the PstI site at -2033.

-1290 CAT was derived from -2033 CAT in a manner similar to -1270 CAT, except that the Klenow step was eliminated and two oligonucleotides spanning the human TSP 1 sequence from -1290 to -1270 and complementary to SalI at the 5’-end and HindIII at the 3’-end, were annealed and included in the ligation. The two oligonucleotides are 5’-TCGACTGAGATCCTTATTTGGTCA-3’ and 5’- AGCTTGACCAAATAAGGATCTCAG-3’. -2033 (mut -1280) CAT was derived from -2033 CAT by oligonucleotide-directed mutagenesis (24) using -2033 CAT (in the phagemid pBS+) as template. The mutagenizing oligonucleotide was 5”GCTGAGATC-TTTATTTGGT- 3’ in which the underlined “T” represents a substitution for a conserved C in the CArG core of the SRE.

-89 CAT was derived from -342 CAT in the following way. A vector was created from -342 CAT by excising the segment between -342 and +166 by AccI restriction followed by Klenow fill-in and NotI restriction. This vector was then ligated to a -89 to +166 fragment derived from a sequencing deletion template (13) by EcoRI restriction followed by Klenow fill-in and NotI restriction. -38 CAT was derived in the following way: -1270/66 bGH (23) was restricted by BsmI (at -38), blunt-ended with T4 polymerase, and digested with BamHI (at the human TSP l/bGH junction) to release the -38 to +66 fragment, which was ligated to pBS+bGH (23) to generate -38/ 66 bGH. This product was then restricted with PstI (in the polylinker) and BssHII to liberate the -38 to -23 fragment. -2033 CAT was treated with PstI and BssHII to eliminate the -2033 to -23 segment and to produce the appropriate vector. Finally, this vector was ligated to the -38 to -23 fragment.

-2033 (A-238/-38) CAT was derived by ligating a vector prepared by SalI restriction followed with Klenow fill-in and SacI restriction of -38 CAT, to a fragment generated from -2033 CAT by BssHII restriction (at -238), Klenow fill-in, and SacI restriction. -2033 (A-238/-89) CAT and -2033 (A-238/-89;mut -65) CAT were derived as was -2033(A-238/-38) CAT, except that their parent vectors were prepared, respectively, from -89 CAT and -89(mut

-65) CAT (see below) by XbaI restriction and Klenow fill-in followed by SacI restriction.

-89(mut -65) CAT was derived in the following way. A polymerase chain reaction product was generated with the M13 universal oligonucleotide, a mutagenizing oligonucleotide, and a sequencing deletion template (13) bearing human TSP 1 sequences from -2033 to approximately +30. The mutagenizing oligonucleotide was 5”GCCTC TAGAGTCGAATTCGCTTCCTGCCCGGCCGCCGAAGCTTm - GTAAGGAATCCCCAGGAATGCG-3‘ bearing polylinkerXba1 and EcoRI sites at its 5’-end (the mutant nucleotides are underlined). The polymerase chain reaction product was blunt-ended with T4 polymerase, kinased, and ligated, and restricted with XbaI and BssHII to generate a mutant -89 to -23 fragment. Finally, this fragment was ligated to a vector prepared by XbaI and BssHII restriction of -89 CAT.

-1307(A-1228/-89) CAT was generated by ligating a fragment whose 5’-end is the HphI site at -1307 blunt-ended with T4 DNA polymerase, and whose 3’-end is the Eco47IIT site at -1228, to a vector generated from -89 CAT by XbaI restriction and Klenow fill- in. T7ARsa CAT was constructed as follows: -38/66 bGH (see above) was treated with AuaI and Klenow followed by HindIII to liberate a -38 to 15 fragment for ligation into a vector prepared from -1270 CAT by treatment first with BamHI and Klenow, and then with HindIII, to liberate its -1270 to 750 segment. The product, -38 CAT, was then treated with PstI and RsaI to liberate a 205-bp fragment bearing TSP 1 intron 1 sequence adjoining 150 bp of the 5’-end of the CAT structural gene. This fragment was lastly ligated to SmaI- and PstI-restricted pBS+. mTS6PP0.5 was created by ligating a 500- bp PstI fragment spanning intron 4 and parts of exons 4 and 5 of mouse TSP 1 (23) into the PstI site of pBS’. pKONeo (25), pUC119 bearing the SRE, and pUC119 bearing the SRE pm12 mutation (26) have all been published.

Cell Culture, Transfection, and Assays of Gene Expression-NIH- 3T3 cells were grown in Dulbecco’s modified Eagle’s medium + 10% calf serum. For the serum stimulation assay, cells were plated at 9 X lo5 cells/lO-cm dish. Approximately 40 h later they were washed once with warm phosphate-buffered saline and then incubated for 24 h in quiescence medium (Dulbecco’s modified Eagle’s medium + 0.4% calf serum). Stimulation was achieved by supplementing the quiescence medium with 15% fetal bovine serum. All media were supplemented with penicillin, streptomycin, and butyl-p-hydroxy-benzoate. All media used for culture of stably transfected cells were supplemented with 400 pg/ml (activity) Geneticin (Gibco). Stable cell populations were multiclonal and were established by electroporating the TSP 1- CAT constructs, together with pKONeo, at a mass ratio of 15:l.

Gene activities in stably transfected cells were assayed by RNase protection (27) of total cellular RNA (28). T7ARsa CAT and mTS6PP0.5 (see above) were used to generate riboprobes to assay the transgene and endogenous gene mRNAs, respectively. T7ARsa CAT protected 150 bases within the coding part of the CAT transgene mRNA, whereas mTS6PP0.5 protected 181 bases and 90 bases of exons 5 and 4, respectively, in mouse TSP 1. Since specific mRNA levels in quiescent cells were very low and, in some cases, very high in serum-stimulated cells, we optimized quantitation by synthesis of riboprobes with [32P]CTP (3000 Ci/mmol) and by quantitation of gels by imaging with a Molecular Dynamics Model 400s Phosphorimager.

Analysis of DNA-Nuclear Factor Binding-Nuclear extracts were prepared from NIH-3T3 cells as described in Ref. 29. For gel shift, 8-10 pg of extract were incubated with specific and nonspecific DNA in 20 p1 of nuclear dialysis buffer (29). Extract was incubated with nonspecific DNA at room temperature for 10 min. Probe and specific competitor (when appropriate) were added and incubated an additional 25 min at room temperature. Gel-shift reactions were electro- phoresed at room temperature on 4% 30:l polyacrylamide-bis gels buffered with 0.045 M Tris borate, 0.001 M EDTA. Methylation interference analysis was performed using the Maxam and Gilbert sequencing protocol as described (25).

Probe-specific binding conditions (in addition to nuclear dialysis buffer composition (29)) are as follows: TSP 1 -1280, c-fos SRE, TSP 1 -65, NF-Y and SP1,2 pg of calf thymus DNA/2O p1; CTF, 50 mM KCI, 5 mM MgCl,, 2 pg of calf thymus DNA/2O pi; and, C/EBP, 5 mM MgCI,, 2 pg of poly(d1-dC), 20 pl. Gel-shift assay with CP2 as probe utilized a HeLa nuclear extract (30) and 2 pg of poly(d1-dC)/ 20 pl. The calf thymus DNA used in these assays was prepared by shearing and boiling.

Cloned DNA gel-shift probes and competitors are as follows. The TSP 1 -1280 fragment extends from -1307 to -1228 and was prepared from -1307(A-1228/-89) CAT (see above) by restriction

Serum Response of Thrombospondin 1 4991

with BamHI and EcoRI. The TSP 1 -1280 mutant fragment also extends from -1307 to -1228 and was prepared from -2033(mut -1280) CAT by HphI and Eco47III restriction. The c-fos SRE fragment and the c-fos SRE mutant fragment were prepared by Hind111 and EcoRI restriction of pUC119 bearing the SRE and pUC119 bearing the SRE pM12 mutation, respectively (26). The TSP 1 -65 fragment and the TSP -65 mutant fragment were prepared by XbaI and BsmI restriction of -89 CAT and -89(mut -65) CAT, respectively. For preparation of gel shift probes the appropriately restricted plasmids were radiolabeled with T4 polymerase or Klenow prior to gel purification.

Oligonucleotide-derived binding elements were generated as follows. NF-Y and NF-Y mutant are 34-mers corresponding to nucleotides -71 to -38 of the Ea gene (31), except that the NF-Y mutant contains a CCAAT to CCAAA mutation. SPl is the high-affinity SP1-binding site in the first intron of a1 type I collagen (30). CTF is the CTF/NFl binding element in Ref. 32 except that it is extended on the 5’-end of each strand with GG dinucleotides. C F B P is as in Ref. 33. CP2 is as in Ref. 34 except that the sequence conforms precisely with the mouse a-globin gene between -95 and -125. These oligonucleotides were all purified on polyacrylamide gels before use. When used as gel-shift probes, oligonucleotides were labeled with T4 polynucleotide kinase or Klenow, and then repurified on polyacrylamide gels.

RESULTS

Two Regions of the TSP 1 Promoter/S’-Flank, between -1290 and -1270 and between -89 and -38, Display Serum- regulated Function-We initially attempted to devise an assay of serum regulation in transient expression, as was first reported for the c-fos promoter (35). However, the 3-6-fold induction by serum of the transfected -2033 CAT construct, obtained with this assay, was too small and variable for identification of &-regulatory elements. The approach that ultimately succeeded relied upon identification of a strain of NIH-3T3 cells in which the endogenous mouse TSP 1 gene was induced roughly 100-fold by fetal calf serum. This strain became the parent cell line for stable expression of -2033 CAT and its derivatives. Fig. lA shows a representative time course of changes in the levels of -2033 CAT and the endogenous mouse TSP 1 mRNAs, determined by ribonuclease protection after serum stimulation of these NIH-3T3 cells. The results are depicted graphically in Fig. 1B. The behavior of the transgene and that of the endogenous gene were similar during the first 2 h of serum stimulation. This close parallel provides the basis of the functional assay for the serum response of the TSP 1 gene. The serum response of the transgene is defined as the fold response of the transgene mRNA, divided by the fold response of the endogenous gene mRNA, a t 2 h. For the -2033 CAT transgene in Fig. 1, the serum response is approximately 1.0.

Fig. 2 presents the serum responses of members of a 5’- deletion series derived from -2033 CAT. These serum responses fell, with one exception, into two categories: approximately 1 and 0.15. Members of the deletion series that contained a t least 1290 bp of 5’-flanking DNA showed serum responses of about 1; those containing 5’-flanking DNA of 1270 bp or less showed serum responses of about 0.15. There- fore, the region between -1290 and -1270 mediates serum- regulated function.

The shortest member of the series, -38 CAT, showed a serum response which appeared to be less than 0.15, a finding that suggested a serum-regulated function for the region between -89 and -38. Construction of a pair of plasmids with internal deletions spanning this region (Fig. 3, lines 2 and 3 ) , enabled us to determine the influence of this region in the context of the full promoter/5’-flank. The A-238/-38 promoter deletion reduced the serum response of the CAT gene to about 0.15; however, the serum response of the CAT gene with the A-2381-89 internal promoter deletion was about

Hours after serum 0 stimulation

pg RNA/lane 35

-2033 CAT

*

Endogenous TSP 1

1

3.3

2

3.3

120, B.

a .- z! a- P

I 2 3 4 5 Hours after serum stimulation

mRNAs from stably integrated human TSP 1 -2033 CAT, FIG. 1. A representative time course of serum induction of

and from the endogenous mouse TSP 1 gene in NIH-3T3 cells. The RNase protection assay in A is depicted graphically in B. The asterisk in A indicates incompletely spliced precursors of endogenous TSP 1 mRNA (10). Approximately 10-fold more RNA was loaded in the 0-h lanes.

Construct

2) -1539 CAT

3) -129oCAT

4) -1270 CAT -1270

5) -342CAT -342

6) -89CAT

7) -38CAT

-89

-38

Response Serum

1.0 +I- 0.3 (IO)

1.2 +I- 0.4 (7)

0.93 +/- 0.44 (6)

0.14 +I- 0.01 (8)

0.13 +/- 0.04 (3)

0.15 +I- 0.07 (7)

0.07 +/- 0.03 (3)

FIG. 2. Serum responses of members of a 5’ deletion series of human TSP 1-CAT transgenes, stably expressed in NIH- 3T3 cells. The serum response of the transgene is defined as the fold response of the transgene mRNA divided by the fold response of the endogenous gene mRNA, at 2 h. Values in parentheses indicate numbers of determinations. Elongated ouak, transcriptional elements identified in this study; filled rectangle, human TSP 1 first exon; open rectangle, CAT structural gene. The entire human TSP 1 first intron separates the 1st exon and CAT structural gene.


equal to that of -2033 CAT. Therefore, the -89 to -38 region is the serum-responsive subregion of the -238 to -38 sequence and has an influence on the serum response of the TSPl gene that is equal in magnitude to that of the region between -1290 and -1270.

-1307(A-1228/-89) CAT (Fig. 3, line 4 ) contains both the -1290 to -1270 and -89 to -38 regions but lacks most of the rest of the promoter/5’-flank. Its serum response of 0.84 was close to that of -2033 CAT; this finding, in conjunction with earlier results, suggests that all essential serum-regulated elements in the promoter/5‘-flank lie in two regions: -1290 to -1228 and -89 to -38. The experiments that follow identify two transcriptional elements within these regions, centered, respectively, at -1280 and -65.

The -1280 Region Displays the Functional and Nuclear Factor-binding Attributes of an SRE--1280 is the center of a sequence that resembles the c-fos SRE, and is conserved between the human and the mouse TSP 1 promoter (Fig. 4). The “CArG” core, a C dinucleotide followed by 6 As or Ts followed by a G dinucleotide, characteristic of an SRE, is conserved among the three genes. Fig. 4 also shows the 5’- sequence end points of -1290 CAT and -1270 CAT. A deletion from -1290 to -1270, which removes most of the SRE motif, reduced the serum response of the resulting construct to approximately 0.15 (Fig. 2, compare lines 3 and 4 ) . Trans- fection of -2033(mut -1280) CAT tested the function of the SRE motif more specifically because the construct contains a T for C substitution in the 5‘ C dinucleotide of the SRE motif (Fig. 4). The G complement of the analogous C of the c-fos SRE is the site of close interaction between the SRE and two of its direct-binding nuclear factors, including the serum

I ) -2033CAT

4) -1307 (A-12281-89) CAT ”””. e 0.84 +/- 0.24 (3)

5) -2033 (mu1 -1280) CAT 1 0.22 +!- 0.11 (7)

~ 1 3 0 7 - 1 2 2 8 4 9

-2033

6) -2033 (A-23W-89; mut -65) CAT

FIG. 3. Serum responses of human TSP1-CAT transgenes bearing mutations or internal deletions. -2033 CAT is duplica- ted from Fig. 2 for reference. Dotted lines indicate deleted DNA segments. Other details are in legend to Fig. 2.

A . - 12x1 I2711

I I TSP 1 ... GTG~TGAGATCCTTATTTGGTCAAGC. .. TSP I ~1280 mutant T

J.

B

c-fus SRE ~ 3 2 0 GGATGTCCATATTAGGACATCT ~ 2 ! a

human TSP I I W TGAGATCCTTATTTGGTCAAGC -1267

m u s e TSP I I223 CAAGATCCTTATTTGGTGATGC ~12(1?

. . . . . . . . . . . . . . . . . . . . . . . . . . .................. ..................

FIG. 4. A, sequence of the human TSP -1280 region, deletion end points, and the -1280 mutant. B, comparison of c-fos, human TSP 1, and mouse TSP 1 sequences. The conserved C and G dinucleotides of the SRE motifs in all sequences are in bold. The dots indicate identical nucleotides. The c-10s SRE shown is sufficient to confer serum responsiveness upon a nonresponsive promoter (26) and conforms closely with the outlines of DNase protection by SRF-containing nuclear complexes (36).

response factor, SRF (see “Discussion”). Analogous one-base mutations of the c-fos SRE impair or abolish binding of SRF to the c-fos SRE (37). Fig. 3 shows that -2033(mut -1280) CAT yielded a serum response that was close to that observed with -1270 CAT (compare Fig. 3, line 5, with Fig. 2, line 4 ) . These results demonstrate a function for the -1280 region, consistent with its identity as an SRE.

We performed a competition gel-shift assay with nuclear extracts from NIH-3T3 cells, utilizing as probe a fragment containing the -1280 region (Fig. 5A) or the c-fos SRE (Fig. 5B). The -1280 fragment, the -1280 mutant, the c-fos SRE, and the c-fos SRE mutant, pm12, were used as competitors. The pm12 mutant is defective for SRF binding and for function in the c-fos serum response (26). As shown in Fig. 5A, the -1280 fragment, when used as a probe, produced a gel- shifted band that was competed by the unlabeled -1280 fragment and by the c-fos SRE, but not by the -1280 mutant, nor by the SRE mutant. As shown in Fig. 5B, the c-fos SRE, when used as a probe, produced a gel-shifted band that comigrated with that produced by the -1280 fragment; this band shift demonstrated the same competition pattern as did the band shift produced with the -1280 fragment. These findings, together with the functional data, establish the -1280 region as an SRE.

The Serum-responsive Function of the -89 to -38 Region Is Attributable to an NF- Y-binding Site at -65-The sequence between -89 and -38 in the TSP 1 promoter (designated the -65 region) and a -65 mutant sequence are shown in Fig. 6A. The -65 mutant incorporates substitutions for 10 closely contacted nucleotides, determined by methylation interference analysis of a gel-shifted band produced by the -65 region (Fig. 7). These interactions are also portrayed schematically in Fig. 6B and are compared with the contacted nucleotides previously established for the NF-Y binding site in the class I1 major histocompatibility complex (31) and for the CP1- binding site in the adenovirus major late promoter (32). NF- Y and CP1 are closely related, if not identical, binding factors (see “Discussion”). The three methylation interference patterns show the same contacted purines on both strands of the DNA within the CCAAT motif itself. Furthermore, as shown in Fig. 6C, the CCAAT motif in TSP 1 exists in a sequence context which shows additional similarities with the consensus binding sites for NF-Y and CP1.

The comparisons summarized in Fig. 6, B and C, led us to test for the ability of an NF-Y-binding site to compete with the -65 region by gel-shift analysis. The analyses in Fig. 8A show a single shifted band produced by the TSP 1 -65 DNA fragment which is competed by an excess of -65 fragment or NF-Y oligonucleotide, but not by mutant -65 fragment or by mutant NF-Y oligonucleotide. The NF-Y oligonucleotide also yielded a shifted band that comigrated with that formed with the -65 probe (Fig. 8A) and the specific gel-shifted band was competed by excess unlabeled NF-Y oligonucleotide or by the -65 DNA fragment but not by their respective mutants (Fig. 8B). These findings suggest that the -65 region contains an NF-Y-binding site; together with the severely compromised serum response observed for the construct -2033 (A-238/- 89; mut -65) CAT (Fig. 3, line 6), these findings demonstrate that the NF-Y-binding site is involved in the serum response of the TSP 1 gene.

The -65 region in TSP 1 also contains a GC box, the core of the SP1 consensus binding site (39), and CCAAT motifs are part of a subset of the binding sites characterized for CP2 (40), C/EBP (41), and CTF (42). We therefore wished to determine whether any of these previously characterized transcription factors interacted with the -65 region of TSP1. The


Probe 4 TSPl -1280-

Competitor - mut. mut.

c-fos TSPl c-fos TSPl -1280 SRE -1280 SRE

Fold excess - I O 100 IO 100 I O 100 IO 1 0 0

A

Probe -c-fo.y SRE"

Competitor -

mut. mut. -1280 SRE -1280 SRE TSPl c-fos TSPl c-fos

Fold excess - I O 100 I O 100 1 0 100 IO 100

B.

FIG. 5. Gel-shift analysis of the TSP 1 -1280 region with nuclear extracts from NIH-3T3 cells utilizing the -1280 DNA fragment as gel-shift probe in A and the c-fos SRE as probe i n B. The arrow in B designates the specific gel-shifted band.

contribution of each of these factors to the binding activity demonstrated by the -65 fragment was therefore tested by gel-shift analysis. Fig. 9 shows that binding sites for these factors failed to compete with the -65 fragment for binding to factors in the NIH-3T3 nuclear extract. However, each of the binding sites was capable of forming saturable complexes with either an NIH-3T3 extract (SP1, C/EBP, and CTF) or a HeLa nuclear extract (CP2) (data not shown). These findings, by exclusion, provide further support for the presence of an NF-Y-binding site in the -65 region of TSP 1.

A

TTCG GAACAT Mutant

B.

VI . v I 1 v v GCCCGGCCGCCGCCCATTGGCCGGAGGAAT CGGGCCGGCGGCGG-GCCTCCTTA

TSP I -65

A A A A A l l Ah AI 1 1 .

AACATTTTTCTGATTGGTTAAAAGTT TTGTAAAIUiGA-TTTTCAA

I I ,

NF-Y b l I 1

AGGTGATTGGTTTATAG v I 1

T C C A ~ T A T C I 1 b A

CPI

NF-Y Consensus C. G C G G W G C C T C C T T TSP I -65

$;NA&~NNNNNN$ C P I Consensus

FIG. 6. The TSP 1 -65 region. A, a CCAAT motif (boxed) and GC box (dotted box), within a region of inverted dyad symmetry centered a t -65 (arrows), are shown. Also shown for the bottom strand are base substitutions incorporated in the -65 mutant. B, comparison of the methylation interference pattern of the -65 region (from Fig. 7) with methylation interference analysis of NF-Y (31) and CP1 (32). Solid box, CCAAT motif; filled and hollow triangles, strong and weak interference, respectively; +, enhanced nuclear factor binding. C, comparison of the -65 sequence with consensus NF-Y (38) and CP1 binding motifs (32).

DISCUSSION

In this study we have begun to map the elements that mediate the rapid induction of the human TSP 1 gene by serum. The arrangement within the TSP 1 promoter of an SRE centered a t -1280 and an NF-Y-binding site at -65 is in keeping with the organization of many promoters that combine the function of a proximal element with that of a more distant enhancer (43). The most relevant comparison may be with the p-actin promoter. Thus, the serum response of &actin also requires both an SRE and an NF-Y-binding site (44, 45). However, the spatial arrangement of these regulatory elements differs from that in TSP 1. In p-actin, the two elements are closely situated within a 50-bp region centered a t -75 relative to the transcription start site, and the NF-Y-binding site is upstream of the SRE. Like TSP 1, p- actin shows growth factor-response kinetics intermediate between those of groups I and I11 genes (9), with transcription and mRNA accumulation peaking 15 min and 4-8 h, respectively, post-induction (46, 47). The comparison between p- actin and TSP 1 is considered again below in relation to the functional synergism between cis-acting elements in transcriptional regulation.

An SRE, a t -1280, Is Involved in the Serum Response of the TSP 1 Gene-The evidence for the identity of the SRE in TSP 1, and for its function in the serum response of the TSP 1 gene, is as follows. (a ) A discrete deletion of the SRE motif reduced the serum response of a TSP-CAT construct by 85% in stably transfected NIH-3T3 cells (compare Fig. 2, lines 3 and 4 ) . ( b ) Mutation of a single nucleotide in the preserved CArG box at the core of the SRE (Fig. 4A), a mutation which has been found to reduce or disrupt nuclear factor binding to the c-fos SRE (37), also reduced the serum response of -2033 CAT by about 80% (compare Fig. 3, lines 1 and 5). (c) The TSP -1280 element and the c-fos SRE yielded gel-shifted bands that comigrated (Fig. 5, A and B ) and are therefore likely to bind the same nuclear factors in NIH-3T3 cell


3\ G + A B F

-45 " - C . (-

c c T-

G + A B F

;-: C+A B F

m r a @ @

c

5'-ATTGG-3' Strand 5'-CCAAT-3' Strand

FIG. 7. Methylation interference analysis of the gel-shifted complex formed with the TSP 1 -65 fragment as shown in Fig. 8A. Nucleotide sequences are shown aligned with gel electro- phoretic patterns for both upper and lower strands. Overexposures permit visualization of interference at the A nucleotides on both strands within the CCAAT motif. G + A, B, and F designate Maxam and Gilbert G + A sequencing of the starting probe, G sequencing of factor-bound probe, and G sequencing of free probe, respectively. Filled and hollow triangles, strong and weak interference, respectively; +, enhanced nuclear factor binding.

extracts. ( d ) A DNA fragment containing the SRE (TSP 1 -1280) competed with the c-fos SRE for nuclear factor binding in a gel-shift assay (Fig. 5B) and, conversely, the c-fos SRE effectively competed with the TSP -1280 element (Fig. 5A). (e) Mutations in the -1280 element that reduced the serum response of TSP-CAT constructs, and in the c-fos SRE that disrupted the binding of SRF, abolished the ability of mutated DNA elements to compete for nuclear factor binding to either the -1280 element or to the c-fos SRE (Fig. 5, A and B) .

In addition to their ability to mediate serum responsiveness, SREs have been shown to be involved in transactivation of genes by platelet-derived growth factor, protein kinase C, cycloheximide, epidermal growth factor, and nerve growth factor (22), v-src (48), Tax (49), and casein kinase I1 (50). The SRE, and more specifically the muscle-specific CArG element, is also capable of tranducing signals directing muscle-specific expression and of mediating transcriptional repression (22). Thus, to the extent that expression of the TSP 1 gene is similarly influenced, it may be productive to examine the role of the SRE in other aspects of the regulation of expression of the TSP 1 gene.

The similar function of the SREs in TSPl and c-fos suggests that studies of the c-fos SRE and the identification of nuclear factors that may function as co-activators and acces- sory proteins in regulation of c-fos expression (43), may also be relevant to the regulation of TSP 1. Although several

Probe - TSP I -65 - 1 NF-Y

A.

Probe t NF-Y *

Competitor - mut.

TSP 1 TSP 1 NF-Y NF-Y mut. -65 -65

Fold excess - IO 1 0 0 1 0 100 IO 100 IO 100

B.

FIG. 8. Gel-shift analysis of the TSP 1 -65 region with nuclear extracts from NIH-3T3 cells utilizing the -65 DNA fragment as gel-shift probe in A and the NF-Y binding element as probe in A and B.

factors have been described that bind directly to the c-fos SRE (22, 51, 52), SRF is most directly implicated in the induction of c-fos by growth factors. SRF activates transcription of c-fos in vitro and SRF-mediated transcriptional activation is correlated with the formation of preinitiation complexes that require only SRF and a highly purified TFIID


3YYUYc)Uh)U

FIG. 9. Gel-shift analysis of the TSP 1 -65 region with nuclear extracts from NIH-3T3 cells. The results show the absence of competition by SP1, CTF, C/EBP, and CP2 binding elements. NF-Y was included as a positive control. The -65 fragment was used as gel-shift probe.

preparation (53). Recombinant human TFIID (TATA-binding protein) does not substitute for highly purified TFIID, a finding that suggests the involvement of one or more co- activators termed TAFs (TATA-binding protein-associated factor) (54).

A Proximal Element, at -65, Binds NF-Y and Contributes to the Serum Response of the TSP 1 Gene-The evidence for the identity of the -65 element as an NF-Y-binding site, and for its role in the serum response of the TSP 1 gene, is as follows. (a) Deletion or mutation of the -65 element reduced the serum response of -2033 CAT by about 85% in stably transfected NIH-3T3 cells (Fig. 3, compare line 1 with lines 2 or 6). (b) The -65 element and the NF-Y consensus sequences are closely matched (Fig. 6C) and the bases contacted by the -65 region, as determined by methylation interference analyses with nuclear extracts from NIH-3T3 cells, correspond closely with those published for NF-Y (31) and CP1 (32) (Fig. 6B). ( c ) The -65 region and the NF-Y oligonucleotide form similar nuclear factor-DNA complexes in a gel-shift assay (Fig. 8A) and are therefore likely to bind the same nuclear factors in NIH-3T3 cell extracts. (d) A -65 region fragment competed with the NF-Y oligonucleotide for nuclear factor binding in a gel-shift assay and, conversely, the NF-Y oligonucleotide effectively competed with the -65 region (Fig. 8, A and B ) . (e) A mutation in the -65 element that reduced the serum response of TSP-CAT constructs and a mutation in the NF-Y-binding site, that had previously been shown to disrupt NF-Y binding, abolished the ability of the respective binding sites to compete for nuclear factor binding in gel- shift assays (Fig. 8, A and B ) . ( f ) Nuclear factor binding to the -65 element was not competed by oligonucleotides bearing consensus binding sites for other candidate transcription factors including CTF, C/EBP, CP2, and SP1 (Fig. 9). On the basis of these findings and the ubiquity of NF-Y (see below) we identify the -65 element as a component in the serum

response of the TSP 1 gene and as a binding site for NF-Y. NF-Y is a member of a group of CCAAT-binding factors,

the most completely characterized of which include mammalian factors a C P l (55), CP1 (32), and CBF (56) and yeast factor, HAP 2,3,4 (57). The three mammalian factors have been characterized, respectively, in lymphoid cells, HeLa cells, and liver. NF-Y binding activity was identified in a number of cells including macrophages and fibroblasts (58), and its mRNA was present in all mouse tissues tested, including mammary and salivary glands, intestine, and colon (59). The following observations demonstrate either the identity or close relatedness of these four factors. Two subunits of NF- Y, cloned from mouse (59), two subunits of CBF (60,61) and HAP 2 and 3 are interspecies homologs of each other. Isolated CP1 and HAP subunits, incapable of binding DNA them- selves, complement each other when reconstituted to form DNA-binding complexes (62). Binding sites for aCPl and CP1 cross-compete in gel-shift assay (32). a C P l is heterotri- meric (55) as is HAP 2,3,4. Recently a third subunit of CBF has also been reported (60), although cloning of this third subunit, or a similar one from NF-Y, has not yet been reported. Last, NF-Y (311, aCP1 (551, CP1 (321, and CBF (63) are all intolerant of mutations within the CCAAT motif.

We have found that an aCP1-binding site cross-competed with the -65 fragment of TSP 1 in a gel-shift assay; furthermore, an aCPl oligonucleotide containing a CCAAT to CA- GAT mutation that is defective for aCP1 binding, failed to cross-compete with the TSP 1 -65 element (data not shown). In addition, the substitution of CCAAA for CCAAT in an NF-Y-binding site resulted in the inability of the mutant oligonucleotide to compete for binding to the TSP 1 -65 element (Fig. 8A). These results support our identification of the TSP -65 element as an NF-Y-binding site and are consistent with the requirement of NF-Y for an intact CCAAT motif.

In contrast to NF-Y, the transcription factors CP2, C/EBP, and CTF can all bind effectively to sequences that lack a CCAAT motif. C/EBP binds with high affinity to the viral enhancer core sequence 5’-TGTGG(A/T)(A/T)(A/T)G-3’; the substitution of GCAAT for CCAAT in a CCAAT motif- containing binding element actually increased binding affinity (64). The palindrome 5’-TGG(A/C)N,GCCAA-3’ that forms the consensus sequence for binding of CTF (65) only occa- sionally includes a CCAAT motif in one of its termini. CP2 also binds with high affinity to a consensus sequence derived from the mouse a-globin gene, 5’-GCN(C/A)NANC(C/A)AG- 3‘, a sequence that clearly lacks a CCAAT motif (34). The consensus sequences listed above are not well preserved within the -65 element of TSP 1. These observations are consistent with the failure of binding sites for CTF, CP2, and C/EBP to cross-compete with the -65 fragment in gel-shift assays (Fig. 9).

The SRE and the NF-Y-binding Site Cooperate in the Serum Response of the TSP 1 Gene-As shown in Fig. 3, the residual effect of either the SRE or NF-Y-binding site alone in mediating the serum response of the TSP 1 gene is small (15-20% of the response with the intact promoter). Thus, these elements and their respective transcription factors prob- ably interact, and this interaction is likely to be functionally synergistic. Cooperativity between an SRE and another element has also been shown in c-fos (with a CRE to mediate the response to epidermal growth factor (66)), in Krox 20 and Krox 24 (with CREs to mediate TAX responsiveness (6711, and in &actin (with an NF-Y-binding site to mediate serum responsiveness (44, 45)). In the cases of Krox 20, Krox 24, and /?-actin, as in the case of TSP1, these interactions were


functionally synergistic. These examples are illustrative of the promiscuity and synergism of interactions between SREs and other regulatory elements and provide indirect evidence for the function of co-activators in the regulation of promoters containing SREs, as suggested by the models for transcriptional regulation presented by Lin et al. (68).

Another potentially interesting example of functional promiscuity of the SRE is provided by the mouse TSP 1 gene. The mouse TSP 1 gene contains an SRE centered at -1210 (Fig. 4) but lacks a consensus NF-Y-binding motif. Instead, at -70, in register with the human NF-Y-binding site, the mouse gene contains a perfect match with the consensus sequence for binding of Egrl/Krox 24/Zif 268. This finding suggests that Egrl may function analogously to NF-Y in mediating the serum response of the mouse TSP 1 gene. It is of interest that an NF-Y-binding site can be replaced functionally by AP1, SP1, or E2-binding sites for cooperation with an HNF1-binding site in driving albumin expression (69). Perhaps, the divergence of the human and mouse TSP 1 genes has performed a related experiment.

Whether serum-regulated elements other than an SRE and an NF-Y-binding site are present in TSP 1 is unknown. However, this study places constraints on their possible lo- cation. The constructs examined in this study all contain an 700-bp region that extends from +66 to +750, and includes the 3' part of the first exon and the entire first intron. While the latter sequences contribute to the high level expression of TSP 1-CAT constructs by a post-transcriptional mechanism (23), preliminary evidence suggests that this region is not required for the serum response of the TSP 1 gene (data not shown). This observation, in conjunction with the behavior of -1307 (A-1228/-89) CAT and -1290 CAT (Fig. 2 and 3), suggests that two regions, -1290 to -1228 and -89 to +66, contain all the elements within the -2033 to +750 sequence that are required for the serum response of the human TSP 1 gene.

Acknowledgments-We thank Michael Gilman for providing a number of reagents, including the cloned c-fos SRE-binding elements, and for advice throughout this project. We also thank DeAnn Liska for providing the HeLa cell nuclear extract and the SP1-binding element and DeAnn Liska and Jim Slack for a critical reading of the manuscript.

REFERENCES

2. Frazier, W. A. (1991) Curr. Opin. Cell Biol. 3 , 792-799 1. Bornstein, P. (1992) FASEB J., 6,3290-3299

3. Ma'ack, R. A,, Cook, S. C. C., and Bornstein, P. (1986) Proc. Natl. Acad. dci U. S. A. 83,9050-9054

4. Good, D. J., Polverini, P. J., Rastinejad, F., Le Beau, M. M., Lemons, R. S., Frazier, W. A., and Bouck, N. P. (1990) Proc. Natl. Acad. Sci. U. S. A.

5. Majack, R. A,, Goodman, L. V., and Dixit, V. M. (1988) J. Cell Biol. 106 ,

6. Sage, E. H., and Bornstein, P. (1991) J. Biol. Chem. 266,14831-14834 7. Mellstrom, K., Hoglund, A.-S., Nister, M., Heldin, C.-H., Westermark, B.,

8. Bravo, R. (199Oj Cell Growth Diff. 1,305-309 9. Lau. L. F.. and Nathans, D. (1991) In The Hormonal Control of Gene

Transcription (Cohen, P., and Foulkes, J. G., e&) Vol. 6, pp. 257-293,

10. Bornstein, P., Devarayalu, S., Li, P., Disteche, C. M., and Framson, P. Elsevier, Amsterdam

(1991) Proc. Natl. Acad. Sci. U. S. A. 88,8636-8640 11. Bornstein, P., O'Rourke, K., Wikstrom, K., Wolf, F. W., Katz, R., Li, P.,

and Dixit, V. M. (1991) J. Btol. Chem. 266,12821-12824 12. VOS, H. L., Devarayalu, S., de Vries, Y., and Bornstein, P. (1992) J. Eiol.

I l b m 287.18198-1819fi

87,6624-6628

415-422

and Lindberg U. (1983) J. Muscle Res. Cell Motil. 4,589-609

13. Donoviel, D. B., Framson, P., Eldridge, C. F., Cooke, M., Kobayashi, S.,

14. Laherty, C. D., Gierman, T. M., and Dixlt, V. M. (1989) J. Biol. Chern.

-. . .. - - . , "_" "" - and Bornstein, P. (1988) J. Biol. Chem. 263, 18590-18593

264.11222-11227

15. Kobayashi, S., Eden-McCutchan, F., Framson, P., and Bornstein, P. (1986)

16. Majack, R. A,, Mildbrandt, J., and Dixit, V. M. (1987) J. Biol. Chem. 262 ,

17. Laherty, C. D., O'Rourke, K., Wolf, F. W., Katz, R., Seldin, M. F., and

18. Majack, R. A., Cook, S. C. C., and Bornstein, P. (1985) J. Cell Biol. 101 ,

Biochemistry 26,8418-8425

8821-8825

Dixit, V. M. (1992) J. Bid. Chem. 267,3274-3281

1059-1070 19. Asch, A. S . , Leung, L. L., Shapiro, J., and Nachman, R. L. (1986) Proc.

20. Donoviel. D. B.. Amacher. S. L.. Judee. K. W.. and Bornstein. P. (1990) J. Natl. Acad. Sci. U. S. A. 83,2904-2908

21.

22. 23.

24. 25.

26. 27.

28. 29.

30. 31.

32.

33.

Cell. Physiol. '146.16-23 ' '

Lawler, J. Duquette, M., Ferro, P., Copeland, N. G., Gilbert, D. J., and

Treisman, R. (1990) Semin. Cancer Biol. 1 , 47-58 Jenkins: N. A. (1991) Genomics 11,587-600

Bomstein, P., Alfi, D., Devarayalu, S., Framson, P., and Li, P. (1990) J.

Zoller, M. J., and Smith, M. (1983) Methods Enzyml. 100,468-500 Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecuhr Cloning,

Gilman, M. Z. (1988) Genes & Deu. 2,394-402 Second Ed.. Cold Spring Harbor Press, Cold Spring Harbor, NY

Melton, D. A., Krieg, P. A,, Rebagliati, T., Maniatis, T., and Green, M. R.

Chomczynski, P., and Sacchi, N. (1987) A d . Biochem. 162 , 156-159 (1984) Nucleic Acids Res. 12,7035-7056

Malik, R. K., Roe, M. W., and Blackshear, P. J. (1991) J. Biol. Chem. 2 6 6 ,

Liska, D., Robinson, V. R., and Bornstein, P. (1992) Gene Ezp. 2,379-389 Dorn, A., Bollekens, J., Staub, A., Benoist, C., and Mathis, D. (1987) Cell

Chodosh, L. A,, Baldwin, A. S., Carthew, R. W., and Sharp, R. A. (1988)

Vinson, C . R., Sigler, P. B., and McKnight, S. L. (1989) Science 246,911-

, . .

Ewl. Chem. 266,16691-16698

8576-8582

60,863-872

Cell 6 3 , 11-24

a1 c 34. Kim, C. G., Swendeman, S. L., Barnhart, K. M., and Sheffery, M. (1990)

35. Treisman, R. (1985) Cell 42,889-902 Mol. Cell. Biol. 10,5958-5966

36. Treisman, R. (1986) Cell 46,567-574 37. Leung, S., and Miyamoto, N. G. (1989) Nucleic Acids Res. 17,1177-1195 38. Li, X.-L., Hooft van Huijsduijnen, R., Mantovani, R., Benoist, C., and

39. Kadonaga, J., Jones, K., and Tjian, R. (1986) Trends Biochem. 11,20-23 40. Lim, L. C., Swendeman, S. L., and Sheffery, M. (1992) Mol. Cell. Biol. 12,

41. Landschulz, W. H., Johnson, P. F., Adashi, E. Y. , Graves, B. J., and

42. Santoro, E., M&nod, N., Andrews, P. C., and Tjian, R. (1988) Nature 334 ,

Y A Y

Mathis, D. (1992) J. Biol. Chern. 267,8984-8990

828-835

McKm ht S L (1988) Genes & Deu. 2 , 786-800

21 R-274 43. Martin, K. J. (1991) Bioessays 13,499-503 44. Danilition, S. L., Frederickson, R. M., Taylor, C. Y., and Miyamoto, N. G.

(1991) Nucleic Acids Res. 19,6913-6922 45. Kawamoto, T., Makino, K., Orita, S., Nakata, A., and Kakunaga, T. (1989)

Nucleic Acids Res. 17,523-537 46. Greenberg, M. E., and Ziff, E. B. (1984) Nature 3 1 1,433-438 47. Leof, E. B., Proper, J. A,, Getz, M. J., and Moses, H. L. (1986) J. Cell.

48. Qureshi, S. A., Cao, X. M., Sukhatme, V. P., and Foster, D. A. (1991) J.

49. Alexandre, C., and Verrier, B. (1991) Oncogene 6,543-551 50. Gauthier-Riviere, C., Basset, M., Blanchard, J.-M., Cavadore, J.-C., Fer-

51. Attar, R. M., and Gilman, M. 2. (1992) Mol. Cell. B~ol. 12 , 2432-2443 52. Metz, F., and Ziff, E. (1991) Oncogene 6,2165-2178 53. Zhu, H., Roy, A. L., Roeder, R. G., and Prywes, R. (1991) New Eiol. 3,455-

"_ " _

Physiol. 127,83-88

Biol. Chem. 266,10802-10806

nandez, A., and Lamb, N. J. C. (1991) EMBO J. IO, 2921-2930

AKA 54. 55. 56.

57. 58.

59.

60.

61.

62.

63.

64.

Pughi B. F., and Tjian, R. (1992) J. Biol. Chem. 267,679-682 Kim, C . G., and Sheffery, M. (1990) J. Biol. Chem. 265,13362-13369 Hatamochi, P. T., Golumbek, S. E., van Schaftingen, E., and de Crom-

Forsberg, S. L., and Guarente, L. (1989) Genes & Deu. 3 , 1166-1178 Hooft van Huijsduijnen, R., Bollekens, J., Dorn, A,, Benoist, C., and Mathis,

Hooft van Huijsduijnen, R., Li, X. Y., Black, D., Matthes, H., Benoist, C.,

Maity, ,S. N., Sinha, S., Ruteshouser, E. C., and de Crombrugghe, B. (1992)

Vuorio, T., Maity, S. N., and de Crombrugghe, B. (1990) J. Bid. Chem.

Chodosh, L. A,, Olesen, J., Hahn, S., Baldwin, A. S., Guarente, L., and

Karsenty, G., Golumbek, P., and de Crombrugghe, B. (1988) J. Biol. Chem.

Johnson, P. F., and McKnight, S. L. (1989) Annu. Reu. Biochem. 68,799-

brugghe, B. (1988) J. Biol. Chern. 263,5940-5947

D. (1987) Nucleic Acids Res. 16,7265-7272

and Mathis, D. (1990) EMBO J. 9,3119-3127

J. Bwl. Chem. 267,16574-16580

266,22480-22486

Sharp, P. A. (1988) Cell 53,25-35

2 6 3 , 13909-13915

829 65. Jones, K. A,, Kadonaga, J. T., Rosenfeld, P. J., Kelly, T. J., and Tjian, R.

66. Fisch, T. M., Prywes, R., and Roeder, R. G. (1987) Mol. Celt. Biol. 7,3490-

67. Alexandre C. Charnay, P., and Verrier, B. (1991) Oncogene 6 , 1851-1857 68. Lin, Y . S.,'C&ey, M., Ptashne, M., and Green, M. R. (1990) Nature 346 ,

69. Tronche, F.! Roilier, A., Sourdive, D., Cereghini, S., and Yaniv, M. (1991)

(1987) Cell 48, 79-89

3502

359-361

J. Mol. E d . 2 2 1 , 31-43

a serum response element and a binding site for nf-y mediate

Documents