1988, 8(11):4736. DOI: 10.1128/MCB.8.11.4736. Mol. Cell. Biol. 

D D Mosser, N G Theodorakis and R I Morimoto transcription rates in human cells.element-binding activity and HSP70 gene Coordinate changes in heat shock

http://mcb.asm.org/content/8/11/4736Updated information and services can be found at:

These include:

CONTENT ALERTS more»cite this article),

Receive: RSS Feeds, eTOCs, free email alerts (when new articles

http://journals.asm.org/site/misc/reprints.xhtmlInformation about commercial reprint orders: http://journals.asm.org/site/subscriptions/To subscribe to to another ASM Journal go to:

on Novem

ber 8, 2012 by GA

LTE

R H

EA

LTH

SC

IEN

CE

S LIB

RA

RY

http://mcb.asm

.org/D

ownloaded from

Vol. 8, No. 11MOLECULAR AND CELLULAR BIOLOGY, Nov. 1988, p. 4736-47440270-7306/88/114736-09$02.00/0Copyright C 1988, American Society for Microbiology

Coordinate Changes in Heat Shock Element-Binding Activity andHSP70 Gene Transcription Rates in Human CellsDICK D. MOSSER, NICHOLAS G. THEODORAKIS, AND RICHARD I. MORIMOTO*

Department ofBiochemistry, Molecular Biology, and Cell Biology, Northwestern University, Evanston, Illinois 60208

Received 15 June 1988/Accepted 16 August 1988

Activation of human heat shock gene transcription by heat shock, heavy metal ions, and amino acid analogsrequired the heat shock element (HSE) in the HSP70 promoter. Both heat shock- and metal ion-induced HSP70gene transcription occurred independently of protein synthesis, whereas induction by amino acid analogsrequired protein synthesis. We identified a HSE-binding activity from control cells which was easilydistinguished by a gel mobility shift assay from the stress-induced HSE-binding activity which appearedfollowing heat shock or chemically induced stress. The kinetics of HSP70 gene transcription paralleled therapid appearance of stress-induced HSE-binding activity. During recovery from heat shock, both the rate ofHSP70 gene transcription and stress-induced HSE-binding activity levels declined and the control HSE-bindingactivity reappeared. The DNA contacts of the control and stress-induced HSE-binding activities deduced bymethylation interference were similar but not identical. While stable complexes with HSE were formed withextracts from both control and stressed cells in vitro at 25°C, only the stress-induced complex was detectedwhen binding reactions were performed at elevated temperatures.

Heat shock genes are expressed in response to a widerange of physiologically and chemically induced stress con-ditions (16). A heat shock element (HSE), which has beenidentified by deletion mapping and site-specific mutagenesis,is required for transcriptional activation of heat shock genes(11, 22). The HSE is able to confer heat shock responsive-ness when fused to a heterologous gene and tested bytransfection or on microinjection into Xenopus oocytes (4,23). In most species the HSE appears as multiple arrays ofthe consensus sequence C--GAA--TTC--G, which is locatedat various distances from the site of transcription initiation(1, 16, 38).

Expression of the human HSP70 gene is also regulatedduring normal conditions of cell growth and differentiation(13, 19, 34), as well as under various stress conditions (31,33). This complexity in transcriptional regulation is due tomultiple cis-acting promoter elements in the 5'-flankingsequences of the human HSP70 gene (10, 20, 33, 35). Aproximal domain which extends to position -68 containssequences that are required for basal transcription, growth-regulated expression, and regulation by adenovirus typeElA (34; G. Williams, T. McClanahan, and R. Morimoto,manuscript in preparation). The promoter sequences neces-sary for heat shock and metal ion induction are located in adistal domain which extends from positions -68 to -107(33). Within this region is a single highly conserved HSE (8of 8 positions match to the consensus sequence) centered atposition -100 which overlaps with a weak HSE (4 of 8positions match to the consensus sequence) centered atposition -90 and flanks adjacent weak HSEs (5 of 8 posi-tions match to the consensus sequence) centered at positions-86 and -110 (12, 33). Although sequences at position -91are related to a metal responsive element in the metallothio-nein II gene (14, 28), the single HSE centered at position-100 has been found to be sufficient for stress-inducedtranscription of the human HSP70 gene (G. Williams and R.Morimoto, manuscript in preparation).A heat shock-activated factor that interacts with the HSE

* Corresponding author.

has been identified in yeasts (26, 27, 32). Drosophila mela-nogaster (21, 30, 36, 37, 39), and human HeLa cells (15, 26).In addition to heat shock, this HSE-binding activity can beinduced by treating Drosophila cells with dinitrophenol andsodium salicylate, both of which induce heat shock proteinsynthesis (39). HSE-binding activity appears rapidly in heat-shocked cells and is not blocked by inhibiting protein syn-thesis (15, 39). This suggests that the factor is present innonstressed cells and gains DNA-binding activity as a resultof a heat-induced modification. A factor which has HSE-binding activity is also present in nonstressed yeasts (26, 27)and Drosophila cells (24). Purified HSE-binding proteinsfrom yeasts (27, 32) and Drosophila cells (32, 37) are able toactivate HSP70 gene transcription in vitro (32) or whenmicroinjected along with a Drosophila HSP70 gene intononshocked Xenopus oocytes (37).

In this study we examined the rate of human HSP70 genetranscription in response to heat shock, metal ions, andamino acid analogs. Coordinate with increased transcription,as measured by nuclei run-on analysis, was the appearanceof the induced form of a HSE-binding activity from heat-shocked, metal ion-treated, and azetidine-treated cells. Pro-tein synthesis was not required for heat shock or metalinduction but was required for the activation ofHSP70 genetranscription and stress-induced HSE-binding activity byamino acid analogs. We identified a HSE-binding activity incontrol cells which was similar but not identical to thebinding activity in heat-shocked or chemically stressed cells.

MATERIALS AND METHODSMeasurement of transcription rates and mRNA levels.

Transcription in isolated nuclei (6) and subsequent purifica-tion and hybridization of nascent transcripts were performedessentially as described previously (2, 29), as were RNApurification and analysis by Si nuclease protection assays(29).

Preparation of cell extracts and gel shift and methylationinterference assays. HeLa cells were grown in culture dishes(diameter, 10 cm) in Dulbecco modified Eagle mediumsupplemented with 5% calf serum. For heat shock the dishes

4736

on Novem

ber 8, 2012 by GA

LTE

R H

EA

LTH

SC

IEN

CE

S LIB

RA

RY

http://mcb.asm

.org/D

ownloaded from

HSE-BINDING ACTIVITY AND HSP70 GENE TRANSCRIPTION 4737

were sealed with Parafilm and immersed in water baths at theindicated temperatures, which were regulated to within±0.1°C. For chemical treatments, cells were incubated withthe metal ion cadmium sulfate (30 ,uM) or the proline analogL-azetidine-2-carboxylic acid (5 mM). Parallel cultures weresubjected to heat shock or incubated with cadmium orazetidine in the presence of 100 p.g of cycloheximide per ml.Whole-cell extracts were prepared from a single dish con-taining 5 x 106 to 10 x 106 cells (39). Cells were harvested,centrifuged, and rapidly frozen at -80°C. The frozen pelletswere suspended in a buffer containing 20 mM HEPES(N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid; pH7.9), 25% (vol/vol) glycerol, 0.42 M NaCl, 1.5 mM MgCI2,0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, and0.5 mM dithiothreitol and centrifuged at 100,000 x g for 5min. The supernatants were frozen in liquid nitrogen andstored at -80°C. The protein concentration was estimatedwith an assay kit (Bio-Rad Laboratories, Richmond, Calif.).For the gel mobility shift assays (7, 8, 25), 10 ,ug of extract

was mixed with 0.1 ng of a [32P]HSE oligonucleotide (seeFig. 5 for the sequence of the double-stranded oligonucleo-tide) and 0.5 ,ug of poly(dI-dC) (Pharmacia Fine Chemicals,Piscataway, N.J.) in 10 mM Tris (pH 7.8)-50 mM NaCI-1mM EDTA-0.5 mM dithiothreitol-5% glycerol in a finalvolume of 25 ,ul. Binding reactions were incubated for 20 minat 25°C, and a dye solution (2.5 ,ul containing 0.2% bromo-phenol blue, 0.2% xylene cyanol, and 50%o glycerol) wasadded and directly loaded onto a 4% polyacrylamide gel in6.7 mM Tris (pH 7.5)-l mM EDTA-3.3 mM sodium acetate.Gels were run at room temperature for 2.5 h at 140 V, dried,and exposed to XAR film (Eastman Kodak Co., Rochester,N.Y.) at -70°C with intensifying screens. The [32P]HSEoligonucleotide was prepared by 5' end labeling one strandwith T4 kinase (Bethesda Research Laboratories, Inc., Gai-thersburg, Md.) and [32P]ATP (ICN Pharmaceuticals Inc.,Irvine, Calif.) by standard procedures (17). The unincorpo-rated nucleotides were removed by Sephadex G-50 chroma-tography. The radiolabeled strand was then annealed to thecomplementary strand.For the competition experiment (see Fig. 1) the binding

reactions contained 0.1 ng of the [32P]HSE oligonucleotideand a 200-fold molar excess of the unlabeled HSE oligonu-cleotide (self-competition) or of an unlabeled heterologousoligonucleotide (non-self-competition). The heterologousoligonucleotide contained the sequences from positions -79to -46 of the human HSP70 promoter, which has a CCAATbox centered at position -65 (12). To measure the stabilityof the HSE-protein complexes, binding reactions with ex-tracts from control or heat-shocked cells were incubated at25°C for 20 min before the addition of a 50-fold excess of thenonlabeled HSE oligonucleotide. The amount of complexremaining after continued incubation in the presence ofcompetitor was assayed by gel electrophoresis, and theresults were quantitated by densitometry. For the experi-ment shown in Fig. 8, the binding reactions were carried outat various temperatures by using buffers adjusted to pH 7.5at 25, 37, 40, or 43°C.

Methylation interference assays were performed as de-scribed by Gilman et al. (9). Binding reactions contained 2 ngof a partially methylated [32P]HSE oligonucleotide, 100 p.g ofextract, and 10 p.g of poly(dI-dC) in a total volume of 50 ,u1.After polyacrylamide gel electrophoresis the bound and freeregions of the gel were excised and cast in a 1% agarose gel.The DNA was recovered by electrophoresis onto NA-45membranes (Schleicher & Scheull, Inc., Keene, N.H.). Therecovered DNA was extracted twice with phenol-chloroform

and once with chloroform-isoamyl alcohol and was precip-itated with ethanol. The pellets were suspended in 100 ,ul of1 M piperidine and incubated at 90°C for 30 min, and thepiperidine was removed by lyophilization. The DNAs weresuspended in 100 ,ul of water, frozen, and lyophilized. Theradioactivity in the dried samples was determined by Ce-renkov counting. Equal counts of samples recovered fromthe free and the bound regions were electrophoreticallyseparated along with a guanine reaction ladder (18) on a 16%polyacrylamide-8 M urea sequencing gel. The gel was run at1,800 V, dried, and exposed to XAR film (Kodak) at -70°Cwith an intensifying screen.

RESULTSIdentification of HSF-HSE complexes in control and heat-

shocked cells. The factor(s) in human cells which interactwith HSEs were identified by the gel mobility shift assay. Asynthetic double-stranded oligonucleotide corresponding tothe two overlapping HSEs located between positions -107and -83 in the human HSP70 promoter was used as a probefor binding to factors from HeLa cell lysates. The DNA-protein complexes formed during incubation at 25°C wereseparated from free radiolabeled HSE oligonucleotide byelectrophoresis on native polyacrylamide gels. The free HSEoligonucleotide migrated to the bottom of the gel, whilespecific and nonspecific complexes migrated more slowly(Fig. 1).HeLa cells maintained at 37°C contained a HSE sequence-

specific binding activity (Fig. 1, lane 2, open arrowhead)which was competed for by excess unlabeled HSE oligonu-cleotide (lane 3) but not by a heterologous oligonucleotidecontaining the HSP70 CCAAT element (lane 4). Whenextracts prepared from HeLa cells incubated at 43°C for 60min were used in the gel mobility shift assay, we found thatthe HSE sequence-specific complex had a mobility on nativeacrylamide gels which was distinct from that seen withextracts from control cells (lane 5, closed arrowhead). TheHSE-binding activities detected in heat-shocked cells areoften detected as two closely migrating bands which werespecifically competed for by excess unlabeled HSE oligonu-cleotide (lane 6) but not by the CCAAT oligonucleotide (lane7), thus demonstrating the sequence specificity of the DNA-protein interaction. It is clear from the results in Fig. 1 thatthe electrophoretic migration of the HSE-binding activityfrom control cell extracts is distinct from that observed fromheat shock extracts. The term heat shock factor (HSF) isused to describe the HSE-binding activity from control cells(cells grown at 37°C) or stressed cells.The dynamic nature of the control and heat-shocked forms

ofHSF was examined in cells that were incubated at 42°C forup to 3 h or in cells that were incubated at 42°C for 30 minand allowed to recover at 37°C (Fig. 2). Stress-induced HSFappeared rapidly during a continuous heat shock, with nearmaximal levels being detected by 15 min at 42°C andgradually declining after 60 min of heat shock. The level ofthe faster-migrating control form of HSF decreased duringheat shock. During recovery from a 30-min heat shock, thelevel of stress-induced HSF disappeared by 60 min at 37°Cand the level of control HSF increased. In other experi-ments, stress-induced HSF declined within 15 min of recov-ery and independent of new protein synthesis (data notshown). These results indicate that the presence of thestress-induced form of HSF reflects the response to heatshock stress.

Kinetics of HSP70 gene transcription: effects of proteinsynthesis inhibitors. To determine whether the appearance of

VOL. 8, 1988

on Novem

ber 8, 2012 by GA

LTE

R H

EA

LTH

SC

IEN

CE

S LIB

RA

RY

http://mcb.asm

.org/D

ownloaded from

4738 MOSSER ET AL.

ns

37 C

U)

(U Z U- ZLU 8 Wj O

43TC (I hi

LL~0 tOZ LL.0 -j Zowc 0

di m oeVan4av--.%

1 2 3 4 5 67

FIG. 1. Detection of a HSE-binding factor in whole-cell extractsprepared from control and heat-shocked cells. The gel mobility shiftassay was performed with a [32P]HSE oligonucleotide and whole-cell extracts prepared from HeLa cells maintained at 37°C orincubated at 43°C for 1 h. Lanes marked self (lanes 3 and 6)contained a 200-fold molar excess of the nonlabeled HSE oligonu-cleotide. Lanes marked non-self (lanes 4 and 7) contained a 200-foldmolar excess of a noncomplementary oligonucleotide. The lanemarked free (lane 1) lacked extract. A specific interaction wasformed with both control (open arrowhead) and heat-shocked(closed arrowhead) extracts. ns, Nonspecific interaction.

the slower-migrating form of HSF correlated with an in-crease in the HSP70 transcription rate, we examined boththe transcription rate of the HSP70 gene and the level ofHSE-binding activity following various forms of stress.HeLa cells were heat shocked at 42°C or incubated with

42- C

.-o-6r_ 'm O, OZ s X

FIG. 2. Levels of control and stress-induced HSF during contin-uous heat shock and recovery. The levels of control of the HSF-HSE complex (open arrowhead) and heat-shock-induced HSF-HSEcomplex (closed arrowhead) were analyzed by gel mobility shiftassays with extracts prepared from HeLa cells that were incubatedat 42°C for periods of up to 180 min and from cells that wererecovering at 37°C from a 30-min heat shock at 42°C. ns, Nonspecificband.

cadmium or azetidine for various times; nuclei were isolatedand used in an in vitro transcription run-on assay. Thenascent 32P-labeled RNA was isolated and hybridized tofilter-bound plasmid DNA containing the human HSP70gene. The filters were washed and exposed to X-ray film, andthe hybridization intensities were quantified by scanningdensitometry (Fig. 3A). The rate of HSP70 gene transcrip-tion increased more than 20-fold by 30 min and declined tonear background levels between 2 and 4 h of continuous heatshock. In contrast to the rapid activation and repression ofheat shock-induced HSP70 gene transcription, cadmium orazetidine treatment resulted in a delayed response, requiringapproximately 1 h before increased transcription rates weredetected and 2 h to attain a 10-fold increase. Elevated levelsof cadmium- and azetidine-induced transcription were de-tected for 4 to 6 h.To extend the results obtained by transcription analysis,

we analyzed the level of HSP70 mRNA in stressed cells.Cytoplasmic RNA was isolated from cells that were treatedas described above; the level of HSP70 mRNA was deter-mined by Si nuclease protection assay (Fig. 3B). HSP70mRNA levels increased in response to stress and reflectedthe HSP70 gene transcription rate.

In experiments performed in parallel with the transcriptionstudies described above, we examined the effect of theprotein synthesis inhibitor cycloheximide at a level (100 ,ug/ml) which was sufficient to block greater than 98% of proteinsynthesis. Protein synthesis was not required for the activa-tion of HSP70 transcription by heat shock or cadmium, norfor its subsequent repression during continuous stress (Fig.3A). Similar results were obtained with the protein synthesisinhibitors anisomycin and pactamycin (data not shown). Weconsistently observed subtle effects of cycloheximide on thetranscription of the HSP70 gene. For example, the rate ofHSP70 gene transcription induced by heat shock was ini-tially higher in cells treated with cycloheximide and then fellmore rapidly. In cadmium-treated cells, cycloheximidecaused a delay in the induction of HSP70 transcription. Incontrast, the transcriptional induction ofHSP70 by azetidinewas abolished in the presence of cycloheximide, presumablybecause azetidine must be incorporated into newly synthe-sized proteins to have an effect. These results reveal that theactivation and repression ofHSP70 gene transcription whichoccurs during heat shock or cadmium treatment is notdependent on protein synthesis, in particular, on newlysynthesized HSP70, whereas cycloheximide pretreatmentblocks the ability of azetidine to induce HSP70 transcription.

Kinetics of HSF-binding activity following heat shock, metalion, and amino acid analog stress. HSE-binding activity wasstudied in experiments performed in parallel with thosedescribed in Fig. 3. Nearly maximal levels of the stress-induced form ofHSF appeared after 15 min at 42°C (Fig. 4A,lane 2). The levels ofHSF remained nearly maximal until 120min (lane 5) and then declined to weakly detectable levels by240 min (lane 6). In the presence of cycloheximide, thestress-induced form of HSF was activated to nearly maximallevels at 15 min but declined rapidly between 30 and 60 min(lanes 8 and 9). Incubation with cycloheximide alone forperiods of up to 6 h did not cause the appearance ofstress-induced HSF (Fig. 4A and B, lanes 12; and Fig. 4C,lane 10).Comparison of the results presented in Fig. 3 and 4 reveal

that the kinetics of HSP70 gene transcription during heatshock coincide with the levels of stress-induced HSF. Themore rapid decline in HSP70 gene transcription duringcontinuous heat shock in the absence of protein synthesis

, ,

MOL. CELL. BIOL.

.1- ..:,.;-- y

--:. Z.;., -- CO:4 :...... -

m ti ob

on Novem

ber 8, 2012 by GA

LTE

R H

EA

LTH

SC

IEN

CE

S LIB

RA

RY

http://mcb.asm

.org/D

ownloaded from

HSE-BINDING ACTIVITY AND HSP70 GENE TRANSCRIPTION 4739

A. Transcripton Rate

HEAT SHOCK CADMIUM

-CCx+ Cx

0 2 4 6 0 2 4 6

AZETIDINE

0 2 4

B. mRNA Levels

0 2 4 6 0 2 4 6 0 2 4 6

Time (hours)FIG. 3. HSP70 gene transcription and mRNA levels in heat-shocked and cadmium- and azetidine-treated cells. HeLa cells were incubated

at 42°C or treated with 30 ,uM cadmium sulfate or 5 mM azetidine, as indicated, in the presence (+Cx) or absence (-Cx) of cycloheximide.(A) In vitro transcription rates were measured by run-on analysis in isolated nuclei from cells that were treated as described above. Theradiolabeled transcripts were isolated and hybridized to filter-bound plasmid DNA encoding the human HSP70 gene. The levels ofhybridization were determined by laser densitometry of autoradiograms. (B) HSP70 mRNA levels. Cytoplasmic RNA was isolated from cellstreated as described above; HSP70 mRNA levels were analyzed by S1 nuclease protection assays. The probe protected 150 nucleotides ofthe human HSP70 gene 5'-noncoding region. The protected fragments were visualized by gel electrophoresis and autoradiography. ThemRNA levels were quantified by laser densitometry.

paralleled the rapid decrease in levels of stress-inducedHSF. These results also reveal that protein synthesis is notrequired either for the activation of HSP70 gene transcrip-tion or for the appearance of the heat shock-induced form ofHSF. However, a more rapid decrease in levels of stress-induced HSF and HSP70 gene transcription rates occurredduring continuous heat shock in the presence of cyclohexi-mide.

In cadmium-treated cells the level of stress-induced HSFincreased by 30 to 60 min (Fig. 4B, lanes 2 and 3), attainedmaximal levels between 120 and 240 min (lanes 4 and 5), anddeclined by 360 min (lane 6). The mobility of metal-inducedHSF was indistinguishable from that of the heat-inducedHSF. As was observed for heat shock, the relative levels ofcadmium-induced HSF (Fig. 4B) and HSP70 gene transcrip-tion (Fig. 3) coincided. In the presence of cycloheximide

(Fig. 4B, lanes 7 to 11), appearance of the stress-inducedHSF was delayed approximately 60 min, attained maximallevels between 120 and 240 min (lanes 9 and 10), anddeclined by 360 min (lane 11).

Azetidine treatment resulted in the appearance of stress-induced HSF (Fig. 4C, lanes 2 to 5). Maximal levels did notappear until 120 min, which was also the period at which theHSP70 gene transcription rate was maximal (Fig. 3A).Stress-induced HSF did not appear when azetidine-treatedcells were also treated with cycloheximide (Fig. 4C, lanes 6to 9). These results suggest that the de novo synthesis ofazetidine-substituted proteins provides the signal for theappearance of stress-induced HSF.

Further support for multiple mechanisms of stress induc-tion of HSF comes from experiments in which HeLa cellswere heat shocked in the presence of cycloheximide for 120

+ Cx40

2

a

N

aI-

40'

30-

20 -

10'

-Cx

20

- Cx

40

la

20'iZ 10'E

VOL. 8, 1988

on Novem

ber 8, 2012 by GA

LTE

R H

EA

LTH

SC

IEN

CE

S LIB

RA

RY

http://mcb.asm

.org/D

ownloaded from

4740 MOSSER ET AL.

A

42'C - 3422 cycioheximdeO~~~~~~~~~ Co

0 0O O tO 0 0O O v} 1 1 N )- CvO - CN 0

a

ftb%b ft% A ff ^

mu 6 7 8 9 I1

1 2 3 4 5 6 7 5 9 10 1112 13

Cadim+ 1Cadrrium (30PM) cycloheximideu ~ ~~~''6'' '' -

° CD

fs,)6 6 cv' o

a_

N v wCs

v co o fCO- ei co mtSxN 3

B

ns

2 3 4 5 6 8 9 10 11 12 13

Azetidirle:;5rm ,

c

AzetldineCyclOhexinide

CJC

1 2 3 4 5 6 7 8 9 10

FIG. 4. Kinetics of HSF-binding activity following heat shock(42°C), cadmium, or azetidine treatment. Whole-cell extracts were

prepared from HeLa cells that were heated at 42°C (A) or treatedwith 30 ,uM cadmium sulfate (B) or 5 mM azetidine (C) either in thepresence or absence of cycloheximide (Cx; 100 ,Ig/ml) and used forgel shift assay, as described in the text. Extracts were also preparedfrom control cells (37°C) and cells that were incubated with cyclo-heximide for 4 or 6 h. ns, Nonspecific band.

min and then treated with cadmium for 120 min. Stress-induced HSF was no longer detected in cycloheximide-treated cells that were heat shocked for 120 min (Fig. 4A,lane 10). The requirement for continued protein synthesiscould be necessary to maintain HSF levels or the factor(s)involved in HSF activation. To distinguish between thesepossibilities, we asked whether cadmium treatment couldreactivate the stress-induced form of HSF in heat-shockedand cycloheximide-treated cells. The addition of cadmiumresulted in the reappearance of the stress-induced form ofHSF (Fig. 4A, lane 13) and a corresponding reactivation ofHSP70 gene transcription (data not shown). However, heat

shock treatment of cadmium- and cycloheximide-treatedcells did not further increase the levels of stress-inducedHSF (Fig. 4B, lane 13).These results suggest that heat shock and metal ions might

activate HSF by distinct mechanisms. This is consistent withour observations that only heat shock induction of HSFrequires protein synthesis for the continued maintenance ofthe induced form of HSF. Although the factor or factorsnecessary for the heat shock pathway are not active duringprolonged incubation with cycloheximide, our results sug-gest that a metal ion-responsive pathway for HSF activationremains available.

Specificity of the HSF-HSE complex from control andheat-shocked cells. We identified the specific nucleotides inHSE which interacted with control or stress-induced HSF ina sequence-specific manner using methylation interference.The [32P]HSE oligonucleotide was partially methylated withdimethyl sulfate and used as a substrate for the gel shiftassay. The bands corresponding to the control and heat-shocked forms of HSF were eluted, treated with piperidine,and analyzed by urea-acrylamide gel electrophoresis. Thoseguanine residues which interfered with factor binding whenmethylated are underrepresented in the bound fraction, ascompared with the free DNA. The methylation interferencereactions shown in Fig. 5 correspond to the upper and lowerstrands when extracts from HeLa cells maintained at 37°C,incubated at 42°C, or maintained in the presence of cadmiumwere used. Overall the pattern of direct protein contacts withguanine residues was very similar, yet a few differencesbetween the control and heat shock patterns were apparent.Contacts with three guanine residues at positions -105,-104, and -94 in the upper strand were identical for HSFfrom control, heat-shocked, and cadmium-treated cells.However, on the lower strand the HSF from control cellswas in contact with an additional guanine residue at position-107 and the heat-shock- and cadmium-induced HSF was intighter contact with residues at positions -% and -97. Thenonspecific band (Fig. 1) was also examined and was notfound to have any specific DNA-protein interactions (datanot shown).The results of methylation interference revealed five com-

mon guanine nucleotides in the HSE which contacted con-trol or stress-induced HSF. To further establish the impor-tance of these DNA-protein contacts, we used a HSE mutantoligonucleotide (HSE-) in which the guanine nucleotides atpositions -105, -104, and -94 on the upper strand and atpositions -97 and -96 on the lower strand were changed(Fig. 6). While the control and heat-shocked forms of HSFwere specifically competed for by the wild-type HSE oligo-nucleotide (lanes 2 and 6), the mutant HSE- oligonucleotidedid not compete for binding with either form of HSF (lanes4 and 8). These results further demonstrate that the coreHSE promoter element (C--GAA--TTC--G) specifically in-teracts with HSF from control and stress-induced cells.

Dissociation rate of HSF-HSE complex from control andstressed cells. We examined the stability of DNA-proteincomplexes that were formed with extracts from control andheat-shocked cells. This was determined by measuring thelevel of preformed HSF-HSE complex that remained afterprolonged incubation in the presence of a 50-fold molarexcess of unlabeled HSE oligonucleotide. The dissociationhalf-time for the HSF-HSE interaction formed with extractsfrom heat-shocked cells was 39 min (Fig. 7). In contrast, theHSF-HSE interaction formed with extracts from controlcells had a dissociation half-time of 150 min (Fig. 7). Nodecrease in binding was seen when a 50-fold molar excess of

idmu4, -40

AALAesEb

Ir

MOL. CELL. BIOL.

llp.

imo "60" bins

on Novem

ber 8, 2012 by GA

LTE

R H

EA

LTH

SC

IEN

CE

S LIB

RA

RY

http://mcb.asm

.org/D

ownloaded from

HSE-BINDING ACTIVITY AND HSP70 GENE TRANSCRIPTION 4741

CONTROL

upper lower

F B L L B F

U.

4m

HEAT SHOCK CADMIUUM

upper lower upper lower

F B L L B F F B L L B F

a* m a - -__5 _

410-O

M- .- _--_

__ do

j - -- IC

1-G-GAr GCCALC CTIGGAA7A - _C^,-ACC ^ GG-C GCz A ACG

HSE

11 \\ \HSE :;zz TC : GT:CQGC-GT-C:C

3AG C A4 G( h)

B 3I C 43-Ct1 h)

2 3 4 5 6 8FIG. 6. Gel mobility shift competition studies with wild-type and

mutant HSEs. (A) Nucleotide sequences of wild-type and mutant(HSE-) HSEs. The guanine residues altered in HSE- are indicatedin boldface type with the nucleotide position in the HSP70 pro-moter. (B) Gel mobility shift assays with the wild-type HSE bindingto control or heat-shocked extract and solution competition (100-fold excess) by self-HSE (lanes 2 and 6), non-self-HSE (CCAAToligonucleotide) (lanes 3 and 7), and HSE- (lanes 4 and 8).

0. 0

A.3C 33T~GAG.Z7.A! 7C A7C CA iC--A

B.

FIG. 5. Methylation interference of HSF-HSE complexes byusing extracts from control, heat-shocked, and cadmium-treatedcells. Methylation interference analysis was performed with a[32P]HSE oligonucleotide that was 5' end labeled on either the upperor the lower strand. Binding reactions contained whole-cell extractsprepared from control (370C), heat-shocked (420C for 1 h), orcadmium-treated (30 ,uM cadmium for 2 h) cells. Free (F) and bound(B) DNA (as noted in Fig. 1) was separated by electrophoresis on a4% polyacrylamide gel. The DNAs were recovered from the gel andafter purification were cleaved with piperidine. The cleavage prod-ucts were then separated on a 16% polyacrylamide-urea gel togetherwith a guanine reaction ladder (L). The sequence of the humanHSP70 promoter from positions -70 to -120 is shown at the bottomof the figure. The HSE consensus sequence is underlined with solidboxes. The results of the methylation interference experiment withextracts from heat-shocked or cadmium-treated cells (A) and withcontrol cell extracts (B) are shown. The guanine residues whichinterfered with binding when they were methylated are indicated byclosed circles. The open circle indicates a weaker interaction.

a nonspecific oligonucleotide containing the CCAAT boxwas used as a competitor of the HSF-HSE preformedcomplexes (data not shown). Results of these studies offerfurther support that the control and stress-induced forms ofHSF interact with the same guanine nucleotides in the HSEoligonucleotide, yet with different affinities.HSF-HSE binding studies at different temperatures. We

have shown in previous experiments that the control andstress-induced forms ofHSF interact with the synthetic HSEin a sequence-specific manner. We next considered thepossibility that the heat shock-induced form of HSF couldform more stable protein-DNA complexes than the control

form at elevated temperatures. Binding reactions with ex-tracts from control, heat-shocked, or cadmium-treated cellswere incubated at 25, 37, 40, and 43°C and immediatelyloaded onto acrylamide gels. The results shown in Fig. 8reveal that the stability of the control HSF-HSE complexwas reduced as the temperature of the binding reaction wasincreased above 25°C. While the control form of HSFformed stable interactions at 25°C (Fig. 8, lane 1), a shift to37°C or above (lanes 2 to 4) caused rapid dissociation of theHSF-HSE complex. The control HSF-HSE complex reap-peared when the complex was allowed to reform at 25°C

100

70

50

ED

D)

30

10

Timr (hours)

FIG. 7. Dissociation rates for the HSF-HSE complex from con-trol and heat-shocked cells. A competition binding assay wasperformed by adding a 50-fold molar excess of the HSE oligonucle-otide to preformed HSF-HSE complexes. Competition reactionswere performed with extracts from control and heat-shocked (43°C)cells for 1 h. Portions were removed at various times, and sampleswere analyzed by gel electrophoresis. The results were quantitatedby scanning densitometry and normalized.

A " lb...1WiW WWbeS

VOL. 8, 1988

on Novem

ber 8, 2012 by GA

LTE

R H

EA

LTH

SC

IEN

CE

S LIB

RA

RY

http://mcb.asm

.org/D

ownloaded from

4742 MOSSER ET AL.

Control

Ln0

o

41 2- a 4

N co~ 111-

123

ns

LI)N

t*oLO) LO~ CD 0N CM Itt

C HS C C

5 6 7 8

ns

Heat Shock

9V 0 O

c\j n IRr

9 10 11 12FIG. 8. HSF-HSE interactions in binding reactions performed at different temperatures. Gel shift assays were performed with extracts

prepared from control cells (lanes 1 to 4), cells heated at 42°C for 2 h (lanes 9 to 12), and cells treated with cadmium sulfate for 2 h (lanes 13to 16). Binding reactions were performed for 20 min at 25°C (lanes 1, 9, and 13), 37°C (lanes 2, 10, and 14), 40°C (lanes 3, 11, and 15), or 43°C(lanes 4, 12, and 16). Also, binding reactions were performed with extracts from control (C) and heat-shocked (HS) cells at 25°C (lanes 5 and6), with extracts from control cells at 40°C (lane 7), and with extracts from control cells at 40°C for 20 min followed by an additional 20 minof binding at 25°C (lane 8). ns, Nonspecific band.

after the initial incubation at 40°C (lane 8). This observationrules out the possibility that HSF or the HSE oligonucleotideprobe was degraded during incubation at the elevated tem-peratures. When binding reactions were performed withextracts from heat-shocked (lanes 9 to 12) or cadmium-treated (lanes 13 to 16) cells, the HSF-HSE complex wasfound to be stable up to 40°C and some binding activity wasdetected at 43°C. The increased stability of the stress-induced HSF observed with extracts from both heat-shocked and cadmium-treated cells suggests that heat shockor chemical stress alters HSF, thus allowing a more stableDNA-protein complex to form.

DISCUSSION

Heat shock gene transcription can be induced by a varietyof changes in the cellular state, including temperature eleva-tion and exposure to metal ions and amino acid analogs. Asshown by the results of this study, the kinetics of transcrip-tional activation differed greatly between heat shock andeither cadmium or azetidine treatment, with heat shockhaving immediate effects on transcription rates and cadmiumor azetidine requiring approximately 60 to 120 min formaximal levels to be attained. Coincident with the rate ofHSP70 gene transcription was the appearance of the stress-induced form of HSF. Although our studies offer evidencethat the in vivo rate of HSP70 gene transcription corre-sponds with the level of stress-induced HSF, we have notyet examined its transcriptional activity in vitro. We inter-pret our data to suggest that the control form of HSFrepresents a transcriptionally inactive state, while stress-induced HSF corresponds to the transcriptionally activefactor. This interpretation is consistent with the evidencethat basal expression of the human HSP70 gene does notrequire the HSEs at position -100 and is primarily depen-dent on the proximal promoter elements extending to theregion at position -68 (10, 20, 33, 35).We identified a factor in control HeLa cells that bound to

an oligonucleotide containing the HSE and had a mobility onnondenaturing gels that was distinct from that of a HSE-binding protein from heat-shocked cells. Similar HSE-binding activities have been detected in control and heat-

shocked mouse erythroleukemia cells (MEL), humanerythroleukemia cells (K562), human adenovirus-trans-formed 293 cells, and human T lymphoblastoid cells (CEM)(P. Kotzbauer, D. Mosser, and R. Morimoto, unpublisheddata). The specificity of the HSF interactions from controland stressed HeLa cells was established by solution compe-tition experiments, identification of the contact points be-tween HSF and consensus guanine residues in the HSE, andnucleotide substitution of the essential guanine nucleotidesthat contact HSF. Although the apparent mobility of theDNA-protein complex from control or stressed cells wasdistinct on native gels, the patterns of DNA-protein contactsbased on methylation interference studies were very similarbut not identical. An additional difference between thecontrol and stress-induced HSFs was revealed by the disso-ciation rate of the respective HSF-HSE complexes. How-ever, binding ability, as detected by the gel shift assay, onlyrevealed the presence of HSE-binding proteins and did notindicate whether the factor was bound in vivo. In controlcells the factor might not be accessible to chromatin, andfollowing heat shock, HSF could be released from a seques-tered state and bind to the HSE. From results presented hereand elsewhere (15, 39), it is clear that HSF is not synthesizedde novo and, therefore, must be in a transcriptionally inertstate within the cell.The induction and repression ofHSP70 gene transcription

and stress-induced HSF levels are independent of proteinsynthesis and, therefore, are independent of de novo HSP70synthesis. This conclusion contrasts with the proposed self-regulation of HSP70 expression in D. melanogaster (5) andmay reflect species differences. Although it is clear thatactivation of HSF and increased transcription rates do notrequire continued protein synthesis, our results reveal thatthe rate of transcription repression is more rapid in cyclo-heximide-treated cells. The coincidence between HSP70gene transcription rates and stress-induced HSF levels in thepresence or absence of continued protein synthesis suggeststhat the factor or factors that regulate stress-induced HSFalso mediate regulation of HSP70 transcription.

During exposure of cells to elevated temperatures, heavymetal ions, or amino acid analogs, the abundance of the

ns

Cadrmum

cli n st00 d o

hi

13 14 15 16

ns

MOL. CELL. BIOL.

on Novem

ber 8, 2012 by GA

LTE

R H

EA

LTH

SC

IEN

CE

S LIB

RA

RY

http://mcb.asm

.org/D

ownloaded from

HSE-BINDING ACTIVITY AND HSP7O GENE TRANSCRIPTION 4743

control form of the HSF diminishes while the stress-inducedform accumulates. When heat-shocked cells are allowed torecover at 37°C, this pattern of binding is reversed and thecontrol form of HSF reappears. This reciprocity is alsoreflected in the rate of HSP70 gene transcription and maycorrespond to the interconversion of a constitutive factorfrom an inactive to an active state. The active state could bea posttranslationally modified form of the inactive factor, ashas been suggested for yeast HSF (26). In yeasts, theHSE-protein complex formed with extracts from heat-shocked cells has a slower mobility than the complex formedwith extracts from nonstressed cells. Since the difference ingel shift mobility could be reduced by treating the heat-shocked cell extract with phosphatase, it has been suggestedthat transcriptional activation is the consequence of the heatshock-induced phosphorylation of a constitutively boundfactor (26). We favor a similar explanation for the differencein the gel shift mobility of the control and stress-inducedforms of human HSF. Alternatively, the DNA-protein com-plexes in HeLa cells might contain different HSE-bindingproteins or reflect monomeric and multimeric interactions.This latter suggestion seems less likely, since a single HSFhas been purified from yeasts and D. melanogaster (27, 32,37). It remains to be shown whether the control and stress-induced forms of human HSF correspond to posttransla-tional modifications of a common protein.Our evidence that cadmium induces the level of stress-

induced HSF could suggest either that the HSE-bindingfactor is itself a heavy metal-responsive protein, perhapsdistinct from metallothionein regulatory proteins that act onthe human metallothionein I and II genes, or that a metal-sensing receptor acts on HSF. There are at least threevertebrate genes (HSP70, the glucose-responsive geneGRP78, and the metallothionein genes) which are transcrip-tionally induced following incubation by the metal ionscadmium, copper, and zinc (14, 28, 31, 33). Comparison ofthe upstream promoter elements of these genes does notsuggest that there is a common mechanism; for example,HSE is only found immediately upstream of the HSP70 genebut not in the metallothionein (14) or the GRP78 (3) genes.Likewise, the metallothionein genes are not heat shockresponsive. Yet transcription of all three genes responds tothe same metal ions, suggesting that metal ion-sensitiveintermediates in the pathway of transcription activation arelikely to be present, transmitting this information to tran-scription factors such as HSF.

If the HeLa factor present in control cells binds constitu-tively in a stable complex with the HSE, how does heatshock activate transcription? We showed that the controlform of the complex is very unstable when binding reactionsare incubated at elevated temperatures in vitro (Fig. 7).During heat shock, the HSE could become available forbinding by the stress-induced HSF if the control form bindspoorly at elevated temperatures in vivo. This suggestioncould explain why factor binding and transcriptional activa-tion are slower when cells are stressed by means other thanheat shock. In these situations, binding of the stress-inducedHSF would be limited by the slow dissociation rate of theconstitutively bound control form of the factor. However, ifthe factor found in control HeLa cells does not interact withthe chromatin in vivo simply because the two are in separatecompartments, then the dissociation rate of this complexmay be irrelevant to transcriptional control.Although the mechanism by which HSF activity is in-

duced in vivo is not known, our studies offer some insightinto the pathway of HSF activation. Depending on the

environmental or chemical stress, there are differences in thekinetics of HSP70 gene transcription and the levels ofstress-induced HSF. The slower kinetics seen in metal ion-or amino acid analog-treated cells relative to that seen inheat-shocked cells could also reflect different pathways forthe stress-induced activation of HSF. The cellular perturba-tions associated with heat shock are likely to be rapid,whereas chemical inducers would act slowly on the target forHSF activation. Our results with azetidine show that incu-bation of HeLa cells with this proline analog required 30 to60 min in order to accumulate critical levels of de novo-synthesized, azetidine-substituted proteins before activationof HSF could occur. These results reveal the sensitivity ofthe intercellular trigger for HSF activation. It appears thatheat shock and cadmium activate HSF through independentpathways, since cells that were heat shocked in the presenceof cycloheximide for 2 h no longer contained the stress-induced HSF, yet the addition of cadmium to these culturescaused the reappearance of stress-induced HSF. It is ofinterest to establish the biochemical pathways by whichamino acid analogs, metal ions, and heat shock activate asingle transcription factor, HSF. Such studies will revealwhether multiple pathways for stress-induced gene activa-tion are available to the cell.

ACKNOWLEDGMENTS

This study was supported by a Public Health Service grant fromthe National Institutes of Health and grants from the March ofDimes Foundation, Amersham Corp. (Arlington Heights, Ill.), aNatural Sciences and Engineering Research Council of Canadafellowship to D.D.M., and Faculty Research Award FRA 313 fromthe American Cancer Society to R.I.M.We thank Carl Wu and Enzo Zimarino for suggestions; Kim

Milarski, Terri McClanahan, Stephanie Watowich, and Gregg Wil-liams for comments on the manuscript; and the NorthwesternUniversity Biotechnology Facility for oligonucleotide synthesis.

LITERATURE CITED1. Amin, J., R. Mestril, P. Schiller, M. Dreano, and R. Voellmy.

1987. Organization of the Drosophila melanogaster hsp7O heatshock regulation unit. Mol. Cell. Biol. 7:1055-1062.

2. Banerji, S. S., N. G. Theodorakis, and R. I. Morimoto. 1984.Heat shock-induced translational control of hsp7O and globinsynthesis in chicken reticulocytes. Mol. Cell. Biol. 4:2437-2448.

3. Chang, S. C., S. K. Wooden, T. Nakaki, Y. K. Kim, A. Y. Lin,L. Kung, J. W. Attenello, and A. S. Lee. 1987. Rat gene encod-ing the 78-kDa glucose-regulated protein GRP78: its regulatorysequences and the effect of protein glycosylation on its expres-sion. Proc. Natl. Acad. Sci. USA 84:680-684.

4. Corces, V., A. Pellicer, R. Axel, and M. Meselson. 1981. Inte-gration, transcription and control of a Drosophila heat shockgene in mouse cells. Proc. Natl. Acad. Sci. USA 78:7038-7042.

5. DiDomenico, B. J., G. E. Bugaisky, and S. Lindquist. 1982. Theheat shock response is self-regulated at both the transcriptionaland posttranscriptional levels. Cell 31:593-603.

6. Fodor, E. J. B., and P. Doty. 1977. Highly specific transcriptionof globin sequences in isolated reticulocyte nuclei. Biochem.Biophys. Res. Commun. 77:1478-1485.

7. Fried, M., and D. M. Crothers. 1981. Equilibria and kinetics oflac repressor-operator interaction by polyacrylamide gel elec-trophoresis. Nucleic Acids Res. 9:6505-6525.

8. Garner, M. M., and A. Revzin. 1981. A gel electrophoresismethod for quantifying the binding of proteins to specific DNAregions. Nucleic Acids Res. 9:3047-3060.

9. Gilman, M. Z., R. N. Wilson, and R. A. Weinberg. 1986.Multiple protein-binding sites in the 5'-flanking region regulatec-fos expression. Mol. Cell. Biol. 6:4305-4316.

10. Greene, J. M., Z. Larin, I. C. A. Taylor, H. Prentice, K. A.Gwinn, and R. E. Kingston. 1987. Multiple basal elements of a

VOL. 8, 1988

on Novem

ber 8, 2012 by GA

LTE

R H

EA

LTH

SC

IEN

CE

S LIB

RA

RY

http://mcb.asm

.org/D

ownloaded from

4744 MOSSER ET AL.

human hsp70 promoter function differently in human and rodentcell lines. Mol. Cell. Biol. 7:3646-3655.

11. Holmgren, R., V. Corces, R. Morimoto, R. Blackman, and M.Meselson. 1981. Sequence homologies in the 5' regions of fourDrosophila heat-shock genes. Proc. Natl. Acad. Sci. USA 78:3775-3778.

12. Hunt, C., and R. I. Morimoto. 1985. Conserved features ofeukaryotic HSP70 genes revealed by comparison with thenucleotide sequence of human HSP70. Proc. Natl. Acad. Sci.USA 82:6455-6459.

13. Kao, H.-T., 0. Capasso, N. Heintz, and J. R. Nevins. 1985. Cellcycle control of the human HSP70 gene: implications for therole of a cellular ElA-like function. Mol. Cell. Biol. 5:628-633.

14. Karin, M., A. Haslinger, H. Holtgreve, R. I. Richards, P.Krauter, H. M. Westphal, and M. Beato. 1984. Characterizationof DNA sequences through which cadmium and glucocorticoidhormones induce human metallothioniein-IIa gene. Nature(London) 308:513-519.

15. Kingston, R. E., T. J. Schuetz, and Z. Larin. 1987. Heat-inducible human factor that binds to a human hsp7O promoter.Mol. Cell. Biol. 7:1530-1534.

16. Lindquist, S. 1986. The heat shock response. Annu. Rev.Biochem. 55:1151-1191.

17. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1985. Molecularcloning: a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

18. Maxam, A. W., and W. Gilbert. 1980. Sequencing end-labeledDNA with base-specific chemical cleavages. Methods Enzymol.65:499-560.

19. Milarski, K. L., and R. I. Morimoto. 1986. Expression of humanHSP70 during the synthetic phase of the cell cycle. Proc. Natl.Acad. Sci. USA 83:9517-9521.

20. Morgan, W. D., G. T. Williams, R. I. Morimoto, J. Greene,R. E. Kingston, and R. Tjian. 1987. Two transcriptional activa-tors, CCAAT-box binding transcription factor and heat shocktranscription factor, interact with a human hsp7O gene pro-moter. Mol. Cell. Biol. 7:1129-1138.

21. Parker, C. S., and J. Topol. 1984. A Drosophila RNA polymer-ase II transcription factor binds to the regulatory site of anHSP70 gene. Cell 37:273-283.

22. Pelham, H. R. B. 1982. A regulatory upstream promoter elementin the Drosophila HSP70 heat-shock gene. Cell 30:517-528.

23. Pelham, H. R. B., and M. Bienz. 1982. A synthetic heat-shockpromoter element confers heat-inducibility on the herpes sim-plex virus thymidine kinase gene. EMBO J. 1:1473-1477.

24. Shuey, D. J., and C. S. Parker. 1986. Binding of Drosophilaheat-shock gene transcription factor to the HSP70 promoter. J.Biol. Chem. 261:7934-7940.

25. Singh, H., R. Sen, D. Baltimore, and P. A. Sharp. 1986. Anuclear factor that binds to a conserved sequence motif in

transcriptional control elements of immunoglobulin genes. Na-ture (London) 319:154-158.

26. Sorger, P. K., M. J. Lewis, and H. R. B. Pelham. 1987. Heatshock factor is regulated differently in yeast and HeLa cells.Nature (London) 329:81-84.

27. Sorger, P. K., and H. R. B. Pelham. 1987. Purification andcharacterization of a heat-shock element binding protein fromyeast. EMBO J. 6:3035-3041.

28. Stuart, G. W., P. F. Searle, H. Y. Chen, R. L. Brinster, andR. D. Palmiter. 1984. A 12-base-pair DNA motif that is repeatedseveral times in metallothionein gene promoters confers metalregulation to a heterologous gene. Proc. Natl. Acad. Sci. USA81:7318-7322.

29. Theodorakis, N. G., and R. I. Morimoto. 1987. Posttranscrip-tional regulation of hsp7O expression in human cells: effects ofheat shock, inhibition of protein synthesis, and adenovirusinfection on translation and mRNA stability. Mol. Cell. Biol. 7:4357-4368.

30. Topol, J., D. M. Ruden, and C. S. Parker. 1985. Sequencesrequired for in vitro transcriptional activation of a DrosophilaHSP70 gene. Cell 42:527-537.

31. Watowich, S. S., and R. I. Morimoto. 1988. Complex regulationof heat shock- and glucose-responsive genes in human cells.Mol. Cell. Biol. 8:393-405.

32. Wiederrecht, G., D. J. Shuey, W. A. Kibbe, and C. S. Parker.1987. The Saccharomyces and Drosophila heat shock transcrip-tion factors are identical in size and DNA binding properties.Cell 48:507-515.

33. Wu, B. J., R. E. Kingston, and R. I. Morimoto. 1986. Humanhsp70 promoter contains at least two distinct regulatory do-mains. Proc. Natl. Acad. Sci. USA 83:629-633.

34. Wu, B. J., and R. I. Morimoto. 1985. Transcription of the humanhsp70 gene is induced by serum stimulation. Proc. Natl. Acad.Sci. USA 82:6070-6074.

35. Wu, B. J., G. T. Williams, and R. I. Morimoto. 1987. Detectionof three protein binding sites in the serum-regulated promoter ofthe human gene encoding the 70-kDa heat shock protein. Proc.Natl. Acad. Sci. USA 84:2203-2207.

36. Wu, C. 1984. Activating protein factor binds in vitro to upstreamcontrol sequences in heat shock gene chromatin. Nature(London) 311:81-84.

37. Wu, C., S. Wilson, B. Walker, I. Dawid, T. Paisley, V. Zimarino,and H. Ueda. 1987. Purification and properties of Drosophilaheat shock activator protein. Science 238:1247-1253.

38. Xiao, H., and J. T. Lis. 1988. Germline transformation used todefine key features of heat-shock response elements. Science239:1139-1142.

39. Zimarino, V., and C. Wu. 1987. Induction of sequence-specificbinding of Drosophila heat shock activator protein withoutprotein synthesis. Nature (London) 327:727-730.

MOL. CELL. BIOL.

on Novem

ber 8, 2012 by GA

LTE

R H

EA

LTH

SC

IEN

CE

S LIB

RA

RY

http://mcb.asm

.org/D

ownloaded from