Proc. Nati. Acad. Sci. USAVol. 84, pp. 2203-2207, April 1987Biochemistry

Detection of three protein binding sites in the serum-regulatedpromoter of the human gene encoding the 70-kDa heatshock proteinBARBARA J. Wu, GREGG T. WILLIAMS, AND RICHARD I. MORIMOTO*Department of Biochemistry, Molecular Biology and Cell Biology, Northwestern University, Evanston, IL 60201

Communicated by Robert L. Letsinger, December 18, 1986 (received for review October 7, 1986)

ABSTRACT The basal promoter of the human gene en-coding 70-kDa heat shock protein (HSP70) controls maximaltranscriptional activity and serum-regulated expression. Wedemonstrate that the three promoter elements defined by invivo studies-CCAAT, serum-regulated element (SRE), and"TATA"-correspond to protein-binding sites in vitro. Thepromoter interactions with protein factors in HeLa cell crudenuclear extracts were detected by an exonuclease m digestionassay. The sequence specificity was demonstrated with pro-moter probes containing wild-type sequences or unique linker-scanner mutations that alter each of the elements. We suggestthat the protein factor binding to the SRE is involved in theserum-regulated expression of the human gene for HSP70.

The expression of the human gene for 70-kDa heat shockprotein (HSP70) is induced by a wide range of stimuli,including physiological stress such as heat shock and heavymetals or during cell growth after serum stimulation. In eithercase, expression is regulated at the primary level of tran-scription. The human HSP70 gene promoter contains at leasttwo regulatory domains: one that is required for heat shockand metal induction and a separate element that is requiredfor serum stimulation (1).The basal promoter of the human gene for HSP70 resides

within 70 base pairs (bp) ofthe transcription initiation site andis essential for wild-type levels of transcriptional activity incontrol cells and is required for serum inducibility. The basalpromoter contains three elements: an inverted "CCAAT" at-68, a serum-regulated element (SRE) at -58, and a "TATA"at -28. Two of these elements, CCAAT and TATA, arecommon to the promoters ofmany eukaryotic class II genes (2).The third element, the SRE, was proposed because a 5' deletionmutation to position -58 remained serum-inducible, whereas amutation to position -47 was unresponsive (1).We present evidence that these promoter elements, iden-

tified by in vivo analysis, correspond to protein binding sitesin vitro. Exonuclease III assays were conducted with HeLacell crude nuclear extracts and promoter probes containingeither wild-type sequences or unique "linker-scanner" de-letion mutations used to show the sequence specificity of thepromoter-protein interactions.

MATERIALS AND METHODSConstruction of Linker-Scanner Mutations. Linker-scanner

mutations were generated by the method ofHaltiner et al. (3).The parental plasmid for the 5' deletions, pA5, was obtainedby cloning the 1974-bp HindIII-BamHI (partial) fragmentcontaining 188-bp of promoter sequences into pBRN/B withSph I linkers (Bethesda Research Laboratories). The parentalplasmid for the 3' deletions, pA3, was obtained by cloning the

184-bp HindIII-Hpa II fragment containing promoter se-quences from -188 to -4 into pBRN/B in the oppositeorientation with Sph I linkers. These plasmids were treatedwith BAL-31 (International Biotechnologies, New Haven,CT) at the Sal I site. Nru I digestion and intramolecularligation positioned the deletion endpoint adjacent to a Bgl IIsite. Deletion endpoints were confirmed by dideoxy sequenc-ing on supercoiled templates (3). Selected pairs of 5' and 3'deletion mutants were recombined by cloning the BamHI-Bgl II fragment from a 3' deletion plasmid into the Bgl II siteof the appropriate 5' deletion plasmid, regenerating 188 bp ofpromoter sequences containing clustered point mutations atthe Bgl II site.DNA Probes. The Hae III-Hpa II fragment, spanning the

promoter region -150 to -4, was 5'-end-labeled at the HaeIII site and used to detect proximal boundaries of proteinbinding to CCAAT and SRE. The Ssp I-Hinfl fragment,spanning the promoter region -102 to +52, was 5'-end-labeled at the Hinfl site and used to detect distal boundariesof protein binding to CCAAT, SRE, and TATA elements.The Ssp I-Hinfl fragment 5'-end-labeled at the Ssp I site wasused to detect the proximal boundary ofprotein binding to theTATA element. End-labeling reactions with [y-32P]ATP(Amersham) and polynucleotide kinase (New England Bio-labs) have been described (4). Only preparations of wild-typeand linker-scanner-labeled fragments of equivalent specificactivities were used as probes.Nuclear Extracts. HeLa cells were grown in Dulbecco's

modified Eagle's medium supplemented with 10% fetal calfserum. The cells were plated at 20% confluence and harvest-ed at 80% confluence. Nuclear extracts were prepared asdescribed by Dignam et al. (5). We routinely obtained 1 mgof protein per 108 cells.

Binding and Exonuclease Reactions. The binding reactionmixture in 50 A.l consisted of 0.2 ng of probe (5000 cpm) in 10mM Tris, pH 7.8/50 mM NaCl/1 mM EDTA/1 mMdithiothreitol/5% glycerol containing 10 4g of HeLa cellnuclear extract and S pug of poly(dI-dC) (Pharmacia) per ml.After incubation at 250C for 20 min, the concentration ofMgCl2 was brought to 10 mM, and exonuclease III(Boehringer Mannheim) was added to 4 x 103 units per ml.Digestion proceeded at 30'C for 20 min, was stopped with theaddition of an equal volume of 20 mM EDTA/40 /xg oftRNAper ml/1% NaDodSO4, and extracted twice with phenol/chloroform, 1:1 (vol/vol). Nucleic acids were precipitatedwith ethanol and applied to 8% polyacrylamide/urea se-quencing gel (6). After electrophoresis, the gel was dried andexposed to x-ray film. Molecular length markers were pro-vided by end-labeled Hinfl fragments ofpM45 (7) and Hpa IIfragments of pA3.

Abbreviations: HSE, heat shock element; SRE, serum-regulatedelement; HSP70, 70-kDa heat shock protein; Ad5, adenovirus type 5.*To whom reprint requests should be addressed at: Department ofBiochemistry, Molecular Biology and Cell Biology, NorthwesternUniversity, 2153 Sheridan Road, Evanston, IL 60201.

2203

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad. Sci. USA 84 (1987)

RESULTS

Location of Linker-Scanner Mutations. Three linker-scan-ner mutations, depicted in Fig. 1, were used to demonstratethe sequence specificity of the promoter-protein interac-tions. The linker-scanner mutation LS 73/63 alters theCCAAT element by a 10-bp substitution from -73 to -63. LS57/47 alters the SRE by a 10-bp substitution from -57 to -47.LS 40/26 alters the TATA element by a 14-bp substitution from-40 to -26. Save for the indicated substitution, wild-type andlinker-scanner promoters are identical in sequence.

Detection of Sequence-Specific Binding to the CCAAT andSerum-Regulated Elements. The method of in vitro binding ofspecific end-labeled promoter probes with protein factors inHeLa cell crude nuclear extracts and subsequent exonu-clease III digestion (8) was used to resolve multiple pro-moter-protein interactions. DNA fragments that are posi-tioned asymmetrically over the putative binding sites werechosen as probes. This avoids the potential difficulty ofquantitatively detecting binding sites that lie beyond theintrinsic exonuclease III resistance at one-half fragmentlength that is generated by nuclease digestion proceeding onboth strands. The Hae III-Hpa II fragment, spanning -150to -4 and 5'-end-labeled at the Hae III site, served as theprobe to detect proximal boundaries of exonuclease IIIresistance resulting from promoter-protein interactions. TheSsp I-Hinfl fragment, spanning -102 to +52 and 5'-end-labeled at the Hinfl site, served as the probe for distalboundaries.

Incubation of wild-type 32P-labeled Hae III-Hpa II probewith HeLa cell crude nuclear extract resulted in four prom-inent exonuclease III-resistant fragments (Fig. 2A, lane 1)that were not detected in the absence of extract (Fig. 2A, lane4) and were not detected in proteinase K-treated extract (datanot shown). The boundaries of these exonuclease III-resist-ant bands correspond to positions -18, -32, -34 and -52 ofthe HSP70 gene promoter. The -52 band was not detectedwith the CCAAT-mutant LS 73/63 probe (Fig. 2A, lane 2).The -34 band was not detected, while the -32 band wasreduced in relative intensity with the SRE-mutant LS 57/47probe (Fig. 2A, lane 3). Therefore, the -52 band is associatedwith protein binding to the SRE. The -18 band is not signifi-cantly affected by either of the mutant probes. However, wenoticed that the -32/-34 band associated with binding to theSRE increased in relative intensity with the CCAAT-mutant LS73/63 probe (Fig. 2A, lane 2), and the -52 band associated withbinding to CCAAT increased in relative intensity with theSRE-mutant LS 57/47 probe (Fig. 2A, lane 3).The distal boundaries detected with the wild-type Ssp

I-Hinfl 32P-labeled probe correspond to positions -76, -68,-66 and -36 (Fig. 2B, lane 1). The -76 band was notdetected with the CCAAT-mutant LS 73/63 probe (Fig. 2B,

lane 2) and, therefore, corresponds to the distal boundary ofprotein binding to the CCAAT element. By the same logic,the -68/-66 doublet that was not detected with the SRE-mutant LS 57/47 probe (Fig. 2B, lane 3) corresponds to thedistal boundaries of protein binding to the SRE. The majorexonuclease III-resistant band at -36 flanking the TATAelement increased in relative intensity with the SRE-mutantLS 57/47 probe (Fig. 2B, lane 3). We also observed that therelative intensity of the -68/-66 doublet associated withbinding to the SRE increased with the CCAAT-mutant LS73/63 probe (Fig. 2B, lane 2).

Binding to the SRE resulted in two exonuclease III-resistant bands for both distal and proximal boundaries (Fig.3). The distal -68 and -66 bands detected with the SspI-Hinfl 32P-labeled probe were abolished by the SRE-mutantLS 57/47. However, of the proximal -32 and -34 bandsdetected with the 32P-labeled Hae III-Hpa II probe, only the-34 band was completely abolished by LS 57/47. The -32band was merely reduced. Perhaps the residual -32 band(Fig. 2, lane 3) is an intrinsic exonuclease III pause and notthe result of protein binding. Different preparations of theseprobes display an intrinsic -32 band of varying intensities inthe absence of extract (Fig. 2A, lanes 4-6).We conclude that protein-binding to the CCAAT element

protects the region -76 to -52 from exonuclease III diges-tion. Protein binding to the SRE protects the region -68/-66to -34/-32. These boundaries are summarized in Fig. 3.

Detection of Sequence-Specific Binding to the TATA Ele-ment. The boundaries of protein binding to the TATAelement were identified with the mutant LS 40/26 (Fig. 1).This linker-scanner mutant did not completely alter theTATA element, converting the wild-type sequence TATA-AAAGC to CGTAAAAGC. The distal boundary was exam-ined with Ssp I-Hinfl 32P-labeled probes. The -36 banddetected with the wild-type probe (Fig. 4A, lane 1) was notdetected in the absence of nuclear extract (Fig. 4A, lane 3)and was reduced substantially in relative intensity with theLS 40/26 probe (Fig. 4A, lane 2). Therefore, the -36 bandcorresponds to the distal boundary of protein binding to theTATA element. We note that the LS 40/26 probe alsoreduced the relative intensity ofthe -76 band associated withCCAAT and increased the relative intensity of the -68/-66doublet associated with SRE (Fig. 4A, lane 2). The proximalboundary of protein-binding to the TATA element wasidentified with 32P-labeled Ssp I-Hinfl probes (Fig. 4B). The-22 band was substantially reduced with the TATA-mutantLS 40/26 probe (Fig. 4B, lane 2) and was not detected in theabsence of extract (Fig. 4B, lane 3). Therefore, the -22 bandcorresponds to the proximal boundary of protein binding tothe TATA element. Both the -36 (Fig. 2B) and the -22 (Fig.

-110 -100 -90 -80 -70 -60 -50 -40 -30 -20 -10 +1 HSP 70WILD

AACCCCTGGAATATTCCCCACCTGCAGCCTCATCCAGCTCGGTGATTGCCTCAGAACGGAAAAGGCGGGTCTCCCTCACCACTTATAAAACCCCAGCCCCAACCCCTCCCCA TYPE

cgaGatcTcG LS 73/63

cgaGatctcG LS 57/47

cagGatcgatCTcg LS 40/26

FIG. 1. The positions of the HSP70 gene promoter elements upstream of the initiation site (+ 1). The basal promoter resides within -70 andis composed ofCCAAT, SRE, and TATA. The position of the heat shock element (HSE) is also indicated. The nucleotide sequence of wild-typeand linker-scanner mutant promoters and the positions of linker substitutions are given. The solid line represents identity to wild-type sequences,which extends from -188 to +150. Within the substituted linker, residues retaining identity to wild-type sequences are capitalized.

2204 Biochemistry: Wu et al.

CCA SREHSE -Ei-

Biochemistry: Wu et al.

A

IEXTRACT |EXTRACT|M1M21 2 3 4 5 6

probe-(- 1 8 ) + -__M_

( -32 ) Wm *(-34) _

(-52)*

w 160-= 154

* -147w -131u -122w -116w -1110_ - 89

Proc. Natl. Acad. Sci. USA 84 (1987) 2205

BIEXTRACIEXTRACTI Ml M2

1 2 34 5 6

p robe

(-76 )-( 68 )4(-66)_"Oon

(-36) *

MA

U-

U-

U-U-

4o

loo

,M

-160-154147

-131-122-116-110

lo -89

__ _m

40{0

FIG. 2. Detection of protein binding to CCAAT and SRE. (A) Binding and exonuclease III reactions were conducted with the 32P-labeledHae III-Hpa II wild-type (lanes 1 and 4), the CCAAT-mutant LS 73/63 (lanes 2 and 5), or the SRE-mutant LS 57/47 (lanes 3 and 6) probesin the absence (-) or presence (+) of HeLa cell extracts. Other lanes: Ml, labeled Hpa II fragments of pA3; M2, labeled Hinfl fragments ofpM45. (B) Binding and exonuclease III reactions were conducted with the Ssp I-Hinf 32P-labeled wild-type (lanes 1 and 4), the CCAAT-mutantLS 73/63 (lanes 2 and 5), or the SRE-mutant LS 57/47 (lanes 3 and 6) probes. Other lanes: Ml, labeled Hinfl fragments of pM45; M2, labeledHpa II fragments ofpA3. Arrows point to undigested probe and exonuclease III-resistant fragments, whose corresponding position on the HSP70promoter is given in Fig. 3.

4C) bands associated with binding to TATA increased inrelative intensity with the SRE-mutant LS 57/47 probes.We conclude that protein-binding to the TATA element

protects a region from -36 to -22 from exonuclease IIIdigestion. A summary ofthese boundaries is shown in Fig. 3.

DISCUSSIONWe have demonstrated that the basal promoter of the humangene for HSP70 is composed of at least three protein-binding

sites in vitro. The sequence specificity of each binding sitewas confirmed by unique linker-scanner mutations. Theboundaries of protein-promoter interaction at these siteswere defined by resistance to exonuclease III digestion: a14-bp region (-36 to -22) at the TATA element, a 24-bpregion (-76 to -52) at the CCAAT element, and a 36-bpregion (-68 to -32) at the SRE. A rather large protectedregion is positioned asymmetrically over the SRE unlike theprotected regions centered over the CCAAT or TATA

-130 -110 4 -90l

-70I

ok-50 -30 -10 + +10( I

-36i I 1 ;22

TATAAA-68/-66~1I-1 -3j/;-32GAAGGGAAAA

-76

HI1 -52

ATTGG

-0

FIG. 3. Schematic summary locating the distal and proximal boundaries ofprotein-binding to the HSP70 gene promoter derived by resistanceto exonuclease III. e, The 5'-end-labeled site of each probe. The location of the proximal and distal boundaries of protein factors are noted,and the positions of the CCAAT, SRE, and TATA elements are shown.

6+oz-150 +30

Proc. Natl. Acad. Sci. USA 84 (1987)

A1 23

Probe --

(-76) - -

(-68,-66) - w0

(-36)-- a

B1 2 3

I (-22)

C

1 2

(-22)- O

FIG. 4. Detection of protein binding to the TATA element. (A)Binding and exonuclease III reactions were conducted with the SspI-Hinfl 32P-labeled probes. Lanes: 1, wild-type probe with HeLanuclear extract; 2, TATA-mutant LS 40/26 probe with HeLa nuclearextract; 3, TATA-mutant LS 40/26 probe in the absence of extract.(B) Binding and exonuclease III reactions were conducted with the32P-labeled Ssp I-Hinfl probes. Lanes: 1, wild-type probe with HeLanuclear extract; 2, TATA-mutant LS 40/26 probe with HeLa nuclearextract; 3, TATA-mutant LS 40/26 probe in the absence of extract.(C) Binding and exonuclease III reactions were conducted withwild-type (lane 1) and the SRE-mutant LS 57/47 (lane 2) 32P-labeledSsp I-Hinfl probes in the presence of HeLa cell extract. Arrowspoint to exonuclease III-resistant fragments whose correspondingposition on the HSP70 gene promoter is shown in Fig. 3 and toundigested probe.

elements. Although we have not systematically analyzedother sequences within the 36-bp protected region, our initialinquiry using the linker-scanner LS 45/35 showed thatbinding to the SRE remained unaltered (data not shown).We have not proven formally that distinct proteins are

interacting at each site; however, we infer from the results ofother laboratories that at least two factors are distinct. Twoof the binding sites, the CCAAT and TATA elements, areshared among many eukaryotic promoters (2). Protein factorsthat bind to these sites have been partially purified and appearto be distinct. The CCAAT element of the herpes simplexvirus thymidine kinase gene binds to a factor called CTF(CCAAT-transcription factor) purified from HeLa cell nuclei(9) and to a similar factor called CBP (CAT binding protein)purified from rat liver nuclei (10). CTF also binds to theCCAAT element of the human HSP70 gene basal promoterand results in a 26-bp region of resistance to DNase I cleavage(11). TFIID (transcription factor IID), a factor purified fromHeLa cell nuclei, binds to the TATA element and protects a10-bp region on the adenovirus type 5 (Ad5) major latepromoter (12). A 20-bp region of the Drosophila HSP70 genepromoter containing the TATA element is protected by TBF(TATA binding factor), a factor isolated from Drosophilanuclei (8).By exonuclease III digestion, the boundaries of protein

binding to the SRE overlap the boundaries of protein bindingto the CCAAT and TATA elements (Fig. 3). Therefore, wemay expect to see an effect of a mutant site on the utilizationof the adjacent unaltered sites. For this inspection, it isimportant that the loss of binding to the mutant site is notartificially enhancing the detection of binding to the adjacentsites by simply relieving a "roadblock" to exonuclease IIIdigestion. Therefore, the mutant site must be "down-stream," relative to the direction of exonuclease III diges-tion, from the unaltered site under examination. TheCCAAT-mutant LS 73/63 32P-labeled Hae III-Hpa II probedisplays increased relative intensity of bands correspondingto binding at SRE (Fig. 2A, lane 2). This suggests that proteinbinding to CCAAT may affect the accessibility of binding toSRE. Another example of interaction among the binding sitesis the increased SRE and decreased CCAAT binding detectedwith the TATA-mutant LS 40/26 Ssp I-Hinfl 32P-labeledprobe (Fig. 4A, lanes 1 and 2). The interaction between

Table 1. Sequence homology of human HSP70 gene promoter topromoters of other human genes and viral enhancers

Sequence

HSP70 gene -57 AAGGGAAAAGIFNB -66 GTGGGAAATT

-74 AAGTGAAAGTFOS* -280 CGTGGAAACC

-269 CAGGGAAAGGIL2 (distal) GGAGGAAAAA

(proximal) -GAGGAAAA-SV40 enhancer TGTGGAAAGTAd5 enhancer -300 AAGTGAAATC

The SRE ofthe human HSP70 gene promoter is homologous to thatof promoters ofIFNB (16), IL2 (18), and FOS (17) as well as to viralenhancers (13-15). The 6-bp "core" sequence is underlined. SV40,simian virus 40; IFN-,B, interferon 8; IL-2, interleukin 2.*In c-fos, the homology is found in the inverted orientation.

TATA and CCAAT may be explained by the stabilization ofbinding to the CCAAT offered by binding to TATA. Ananalogous interaction between USF (upstream stimulatoryfactor) and TFIID has been documented for the Ad5 majorlate promoter (12). The interaction between TATA and SREmay be direct or indirect-i.e., the increased detection ofbinding to SRE resulting from the destabilization of bindingto CCAAT. However, the increased binding to TATA de-tected with the SRE-mutant LS 57/47 32P-labeled Ssp I-Hinflprobe (Fig. 4C) argues for a direct interaction betweenbinding to TATA and SRE. We suggest that the interactionamong factors binding to these promoter elements willcontribute to the transcriptional activation of the humanHSP70 gene's basal promoter.The SRE of the human HSP70 gene promoter lies within

-57 to -47 as defined by in vivo expression studies and invitro binding assays. Within this purine-rich region is a 6-bpcore sequence GGGAAA that is common to simian virus 40and Ad5 viral enhancers (13-15) and to human genes forinterferon p (INFB) (16), c-fos (FOS) (17), and interleukin 2(IL2) (18) promoters (the designation of a human gene is inparentheses; Table 1). A point worth noting is that thesecellular promoters are growth-regulated, as is the HSP70gene promoter. We propose that the factor(s) binding to theSRE is involved in growth-regulated expression of the HSP70gene. The negatively regulated IFNB promoter consists of anegatively acting element positioned in between an upstreampositively acting element and the TATA element (15). Theorganization of the human HSP70 basal promoter resemblesthis motif-the SRE positioned in between an upstreamCCAAT and the TATA elements. Considering that in vitrobinding to the SRE overlaps with binding to both CCAAT andTATA and that binding of protein factors to the SRE appearsto affect the efficiency of promoter-protein interactions atCCAAT and TATA, the in vivo binding to the SRE underconditions of serum starvation may possibly repress HSP70expression.

We are grateful to Nic Jones, Dan Linzer, and Bob Holmgren forcritical discussion; to Carl Wu and Vincenzo Zimarino for advice onexonuclease III assays; and to the members of our laboratory-Sheila Banerji, Nick Theodorakis, Kim Milarski, and StephanieWatowich-for patient encouragement. B.J.W. is a Leukemia So-ciety of America Special Fellow. This work was supported by grantsfrom the Leukemia Research Foundation and the National Institutesof Health.

1. Wu, B., Kingston, R. & Morimoto, R. (1986) Proc. Nati.Acad. Sci. USA 83, 629-633.

2. Dynan, W. & Tjian, R. (1985) Nature (London) 316, 774-778.3. Haltiner, M., Kempe, T. & Tjian, R. (1985) Nucleic Acids Res.

13, 1015-1025.

2206 Biochemistry: Wu et al.

Biochemistry: Wu et al.

4. Wu, B., Hunt, C. & Morimoto, R. (1985) Mol. Cell. Biol. 5,330-341.

5. Dignam, J., Leibowitz, R. & Roeder, R. (1983) Nucleic AcidsRes. 11, 1475-1489.

6. Maxam, A. & Gilbert, W. (1980) Methods Enzymol. 65,499-560.

7. Lamb, R. A., Lai, C.-J. & Choppin, P. W. (1981) Proc. Natl.Acad. Sci. USA 78, 4170-4174.

8. Wu, C. (1985) Nature (London) 317, 84-87.9. Jones, K., Yamamoto, K. & Tjian, R. (1985) Cell 42, 559-572.

10. Graves, B., Johnson, P. & McKnight, S. (1986) Cell 44,565-576.

Proc. Nati. Acad. Sci. USA 84 (1987) 2207

11. Morgan, W., Williams, G., Morinoto, R., Greene, J.,Kingston, R. & Tjian, R. (1987) Mol. Cell. Biol. 7, 1129-1138.

12. Sawadogo, M. & Roeder, R. (1985) Cell 43, 165-175.13. Weiher, H., Konig, M. & Gruss, P. (1983) Science 219,

626-631.14. Hearing, P. & Shenk, T. (1983) Cell 33, 695-703.15. Hearing, P. & Shenk, T. (1986) Cell 45, 229-236.16. Goodbourn, S., Burstein, H. & Maniatis, T. (1986) Cell 45,

601-610.17. Treisman, R. (1985) Cell 42, 889-902.18. Fujita, T., Shibuya, H., Ohashi, T., Yamanishi, K. &

Taniguchi, T. (1986) Cell 46, 401-407.