the c-terminal hydrophobic repeat ofschizosaccharomyces pombe heat shock factor is not required for...

. 14: 733–746 (1998)

The C-terminal Hydrophobic Repeat of
Schizosaccharomyces pombe Heat Shock Factor is notRequired for Heat-induced DNA-binding
KIRSTIE A. SALTSMAN†, HOLLY L. PRENTICE‡ AND ROBERT E. KINGSTON*

Department of Molecular Biology, Massachusetts General Hospital, Boston, MA 02114 and Department ofGenetics, Harvard Medical School, Boston, MA, U.S.A.

Received 30 September 1997; Accepted 10 January 1998

The C-terminal hydrophobic repeat (CTR) of heat shock transcription factor (HSF) has been proposed to regulateDNA binding by intramolecular interactions with the leucine zipper motifs present in the HSF trimerization domain.Schizosaccharomyces pombe provides a useful model organism for the study of the regulation of HSF DNA bindingbecause, unlike Saccharomyces cerevisiae, S. pombe hsf is highly heat shock inducible for DNA binding and containsa clear homology to the CTR. We examined the role that the CTR plays in the regulation of S. pombe hsf byconstructing isogenic strains bearing deletion and point mutations in the chromosomal copy of hsf. Surprisingly, wefound that point mutation of key hydrophobic amino acids within the CTR, as well as full deletion of it, yieldedfactors that show normal binding at normal growth temperatures and full levels of heat-induced binding. Deletionof the CTR did, however, slightly lower the temperature required for maximal activation. In contrast, a largedeletion of the C-terminus, which removes close to a third of the coding sequence, was deregulated and bound DNAat control temperature. Several of the deletion mutants were significantly reduced in their level of expression, yet theyshowed wild-type levels of DNA binding activity following heat shock. These experiments demonstrate thatappropriate regulation of the DNA binding activity of S. pombe hsf is not solely dependent upon the CTR, andimply that a feedback mechanism exists that establishes proper levels of DNA binding following heat shock despitemutations that significantly alter levels of total hsf. ? 1998 John Wiley & Sons, Ltd.

Yeast 14: 733–746, 1998.

— heat shock; protein-DNA interactions; transcriptional regulation

INTRODUCTION activation of a pre-existing, ubiquitous transcrip-

The heat shock response is induced by a variety ofenvironmental stresses including heat shock, treat-ment with heavy metals or ethanol, and incorpor-ation of amino acid analogs, and it results in thesynthesis of a set of proteins called the heat shockproteins (HSPs); Lindquist and Craig, 1988;Hightower, 1991). Transcriptional induction of theheat shock protein genes is accomplished by the

*Correspondence to: R. E. Kingston, Department of Molecular
Biology, Wellman 10, Massachusetts General Hospital, Boston,
? 1998 John Wiley & Sons, Ltd.

tion factor called heat shock factor (HSF). Uponinduction in higher eukaryotes, HSF undergoes amonomer to trimer transition, becomes hyper-phosphorylated, and binds to the promoters ofheat shock protein genes. The HSF binding site,the heat shock element (HSE), has been highlyconserved throughout evolution and consists of aseries of NGAAN motifs arranged in alternatingorientation. Induction of binding to the HSEappears to be one of the primary mechanisms foractivation of HSF, and hence induction of thestress response, in most eukaryotic cells.

HSF was first cloned from Saccharomycescerevisiae and was subsequently cloned from anumber of organisms including human, mouse,tomato, Drosophila melanogaster, Kluyveromyces

MA 02114, USA. Tel.: (+1) 617 726 5990; fax: (+1) 617 7265949; email: [email protected]†Present address: Department of Biological Sciences, StanfordUniversity, Stanford, CA 94305-5020, U.S.A.‡Present address: Biogen, 14 Cambridge Center, Cambridge,MA 02142, U.S.A.

CCC 0749–503X/98/060733–14 $17.50

lactis and Schizosaccharomyces pombe (Sorger and HSF2 (hHSF2) to be constitutively localized to the

734 . . .

Pelham, 1988; Wiederrecht et al., 1988; Closeet al., 1990; Scharf et al., 1990; Jakobsen andPelham, 1991; Rabindran et al., 1991; Sarge et al.,1991; Schuetz et al., 1991; Gallo et al., 1993).Comparison of the amino acid sequences of theseHSFs revealed two highly conserved domains inthe amino-terminal part of the protein. The mostamino terminal of these was defined as the DNAbinding domain by deletion analysis in S. cerevi-siae (Wiederrecht et al., 1988). It was subsequentlycrystallized from the yeast K. lactis and found tobe of the helix-turn-helix class of DNA bindingmotifs (Harrison et al., 1994). The second con-served domain lies immediately carboxy terminalto the DNA binding domain and has been foundto mediate the trimerization of HSF. This domainconsists of a series of hydrophobic heptad repeatssimilar to the leucine zipper, and these are thoughtto interact with one another through intermolecu-lar interactions in the formation of the HSF trimer(Sorger and Nelson, 1989; Peteranderl and Nelson,1992). Leucine zippers have alpha-helical struc-tures, and interactions between them occur viacontacts between hydrophobic amino acids occu-pying the ‘a’ and ‘d’ positions of a heptad sequence‘abcdefg’. A third conserved domain lies close tothe C-terminus of the protein. It too consists of ahydrophobic heptad repeat, but this repeat is shortand may not be able to maintain the stable proteininteractions typical of leucine zippers.

A general model has emerged for the regulationof HSF trimerization and DNA binding whichproposes that under non-stressful conditions theC-terminal hydrophobic repeat (CTR) interactswith the leucine zippers located in the trimerizationdomain, thus shielding it and keeping the factor inan inactive, monomeric form (Lis and Wu, 1993).Upon heat shock this interaction would be dis-rupted, freeing the trimerization domain such thatit is able to interact with other monomers and formthe active, trimeric form of HSF. Evidence tosupport this has come from mutagenesis studies ofhuman HSF1 (hHSF1) and D. melanogaster HSF.Mutation of the CTR in both of these has yieldedfactors that bind DNA and trimerize constitu-tively. These mutations have been proposed toalter association of the CTR with the trimerizationdomain because mutations in the trimerizationdomain of hHSF1 have had the same effect(Rabindran et al., 1993; Zuo et al., 1994). Inaddition, point mutation of the trimerizationdomain or deletion of the CTR cause human


nucleus, whereas wild-type hHSF2 only enters thenucleus following heat shock induction (Sheldonand Kingston, 1993). This suggests that inter-actions between the trimerization domain and theCTR might also modulate nuclear localizationunder certain circumstances.

Some recent experiments are not consistent withthis model for HSF activation (Zandi et al., 1997).In this work, where D. melanogaster HSF mutantswere inducibly expressed from a metallothioneinpromoter, deletion of the CTR did not signifi-cantly alter DNA binding induction. Instead, asequence nearer the N terminus was proposed toplay a key role in the regulation of HSF bindingactivity and nuclear localization. While studieswith hHSF1 support a role for the CTR in regula-tion of DNA binding, they also imply that othersequences are important, as a deletion of theextreme C-terminus, which leaves the trimerizationdomain and CTR intact, has been found to cause asignificant increase in DNA binding and trimeriza-tion at control temperature (Rabindran et al.,1993).

The fission yeast S. pombe provides a goodmodel system for understanding induction of HSFDNA binding. The interpretation of the effects ofHSF mutants in metazoans is problematic becauseof the presence of endogenous HSF in the cell. Theability to perform homologous recombination inS. pombe circumvents this problem and allows oneto study the effects of mutations in the absence ofwild-type hsf. S. cerevisiae would also provide thisadvantage, but because S. cerevisiae HSF can bindconstitutively to certain heat shock elements, char-acterization of induced DNA binding is difficult(Giardina and Lis, 1995). In contrast, the DNAbinding activity of S. pombe hsf increases dramati-cally upon heat shock (Gallo et al., 1991). We havemutated the CTR of S. pombe hsf and analysed theeffects of these mutations on hsf activity followingintegration of the mutants into their normalgenomic location. These studies demonstrate thatin this setting the CTR is not required for theregulation of hsf DNA binding activity.

MATERIALS AND METHODS

Plasmids and fusion constructs

Construction of the epitope tagged hsf A triplepoint mutation was made at the N-terminus of thehsf coding sequence in a plasmid which contained

. 14: 733–746 (1998)

the hsf+ cDNA in the NotI site of Bluescript SK+ SK+ with ura4+ in the HindIII site. Ä411-609 was

735.

(Stratagene), generating an NdeI site. Using oligoswith NdeI ends as primers, the pCGN (Tanaka andHerr, 1990) plasmid as template, and PCR wemade a fragment containing the influenza HAepitope tag and inserted it into the NdeI siteengineered into the N-terminus of hsf. Thesequence that was inserted is shown below, wherethe epitope tag amino acids are shown in boldface and the beginning of the hsf sequence isunderlined.

MSYPYDVPDYASLGGPSMHARSHMIQ

The sequence and junctions were confirmed usingSequenase (US Biochemical).

Construction of the hsf mutants All of the hsfmutants were derived from the parental plasmidwhich contained the wild-type hsf cDNA in theNotI site of pBluescript SK+ (Stratagene). MutantÄ267-410 was made by cutting with StuI and HpaIand religating. For mutant Ä476-495, site-directedmutagenesis (Kunkel, 1985) was used to substituteL497 for a serine residue, also creating an NheIsite. Cleavage with NcoI and NheI, removing theintervening sequence, followed by the ligation ofan adapter with NcoI and NheI ends, yieldedÄ476-495. For Ä411-504 site-directed mutagenesiswas used to insert a T residue in between nucle-otides 1720 and 1721, thus creating a PvuII site.Cleavage with HpaI and PvuII, removal of theintervening sequence, and religation of the vectoryielded Ä411-504. The T insertion mentionedabove shifts the reading frame such that a stopcodon is encountered two codons downstream,thus creating Ä505-609. Ä476-609 was generatedby cleaving with NcoI and inserting a linker con-taining NcoI ends and a stop codon. Ä411-609 wasmade by cutting with HpaI (in the hsf codingregion) and XbaI (in the polylinker) and inserting afragment corresponding to the 3*UTR which hadbeen generated by PCR and cleaved with NruI andXbaI to provide the appropriate ends for ligation.The point mutations L497P and L483A/L487Awere made by site-directed mutagenesis. All of themutations that were generated by site-directedmutagenesis were sequenced, as were the junctionsof those that were generated by restriction enzymecleavage and religation. All of the mutations, withthe exception of Ä411-609, were then transferredas SacII/SpeI fragments to plasmid pS2S, thevector used for integration. pS2S is pBluescript


transferred as a SacII/SpeI fragment to the pLEV5expression vector, which contains the S. pombears1 origin of replication, the S. cerevisiae LEU2gene as a selectable marker, the SV40 earlypromoter, and the S. pombe act1 transcriptionalterminator.

Construction of the plasmid for the integration ofthe epitope-tagged hsf An EcoRI/SmaI hsf frag-ment containing the epitope tag (see above) wassubcloned into plasmid pH1B (Gallo et al., 1993),which contains a genomic fragment of the hsf+

clone, including the promoter region, the 5*UTRand the N-terminal part of the coding sequence.The ura4+ gene was inserted into the HindIII siteof the polylinker.

Construction of the mutant strains

Strain WT We used the two-step gene replace-ment method (Scherer and Davis, 1979) to replacethe endogenous, chromosomal copy of hsf+ withan N-terminally influenza HA epitope-taggedversion. The plasmid described in the previousparagraph was linearized at the unique NruI site inthe hsf coding sequence and transformed by elec-troporation (Prentice, 1992) into strain HP8 (ade6-216, ura4-294, leu1-32, h+) and ura+ transformantswere selected for. This first step of the methodproduces a configuration at the hsf+ locus in whichthe tagged and untagged copies of the gene lie oneither side of the ura4+ gene. In the second stephomologous recombination between the two cop-ies results in loss of ura4+ and leaves either thetagged or untagged copies of hsf behind. Theseura" clones were selected for using the drug5FOA, and PCR analysis was used to determinewhich copy of hsf was left behind.

Construction of the integrated hsf mutant strains(Ä267-410, Ä476-495, Ä411-504, Ä505-609, Ä476-609, L497P and L483A/L487A) The mutant hsfswere subcloned into the integration vector pS2Swhich contains ura4+ as the selectable marker, andwere linearized within the hsf sequence and trans-formed into the WT strain. The two-step genereplacement method was then applied as describedabove.

Construction of strain Ä411-609 The plasmidshuffle (Boeke et al., 1987) was used to create

. 14: 733–746 (1998)

strain Ä411-609. The plasmid expressing Ä411-609(see above) was transformed by electroporation

methanol, 150 m-glycine, 20 m-Tris base on toa nitrocellulose membrane (Schleicher and Schuell)

736 . . .

into a strain deleted for the chromosomal hsf+

gene and expressing the wild-type hsf+ gene from aplasmid containing ura4+ as the selectablemarker (Gallo et al., 1993). Leu+ transformantswere selected for and these were grown in theabsence of ura selection for several generations toallow for the loss of the plasmid expressingwild-type hsf. Ura" clones were then selected forusing 5FOA, and PCR was used to confirm thepresence of the deleted hsf and the absence of thewild-type copy.

DNA binding assaysCells were grown and heat-shock treated as for

the RNA for the primer extensions (see below).The hsf-containing extracts were then prepared asdescribed previously (Gallo et al., 1991). Theprobe consisted of a double-strand synthetic oligo-nucleotide containing bases "115 to "80 of thehuman hsp70 promoter (Figure 3; Kingston et al.,1987) or bases "231 to "196 of the S. pombewis2+ promoter (Figure 6), and was end labelledusing [c-32P]ATP and T4 kinase. The gel electro-phoretic mobility shift assays were performed asdescribed previously (Gallo et al., 1991). Briefly,15 ìg of extract was combined with 2 ng of probe,4 ìg of poly(dI-dC) (Pharmacia) and MgCl2 wasadded to a concentration of 1 m. The incubationwas carried out at 30)C for 30 min. For theantibody supershifts 200 ng of 12CA5 mono-clonal antibody (Boehringer Mannheim) wasadded to the hsf-containing extract and incubatedon ice for 15 min prior to the addition of theprobe. The products were separated on a 4% nativePAGE gel using HISB (0·25 -Tris base, 2 -glycine, 10 m-EDTA) as the running buffer. Thegels were dried down, exposed to film and aPhosphorimager (Molecular Dynamics) was usedfor quantitation.

The Kd values were determined by plotting thevalue obtained for the hsf/HSE band against theprobe concentration. The apparent Kd was takento be the probe concentration at which halfmaximal binding was attained.

Western analysisThe hsf-containing extracts used for the Western

analysis were the same as those used for the DNAbinding assays above. 150 ìg of extract was run onan 8% SDS–PAGE gel and transferred in 20%


at 50 mAmps for approximately 16 h at 4)C. Allthe remaining steps were carried out at roomtemperature. The membrane was removed andblocked in PBS containing 5% milk for 2 h. Themembrane was then rinsed in PBST (PBS+0·05%Tween 20) and incubated for 2 h in PBST contain-ing the 12CA5 antibody (Boehringer Mannheim)at a concentration of 500 ng/ml. The membranewas then washed with PBST one time for 15 minand three times for 5 min, followed by the additionof the anti-mouse IgG-HRP (Amersham) at a1:2000 dilution in PBST. After a 2-h incubationthe membrane was washed as above and the signaldeveloped by ECL (Amersham).

Northern analysisThe RNA samples were the same as those used

for the primer extension analysis described below.30 ìg of each sample was separated on to a 1%agarose, 20% formaldehyde gel using 1XFGB run-ning buffer (5 m-NaOAc, 20 m-MOPS pH 7,1 m-EDTA) and then transferred by capillaryaction for approximately 16 h on to a nitrocellu-lose membrane using 20X SSC as previouslydescribed (Ausubel et al., 1989). The membranewas then dried under vacuum at 80)C for 30 min,followed by incubation for 5 h at 42)C in hybridi-zation buffer (6X SSC, 5X Denhardts, 0·5% SDS,50% formamide, 100 ìg/ml salmon sperm DNA).The probe was made by random hexamer labelling(Ausubel et al., 1989) using plasmid pS2S,which contains both hsf+ and ura4+ sequences,as template. The probe was boiled prior toaddition to the membrane and hybridizationbuffer, and the incubation was continued forapproximately 16 h. The membrane was thenremoved from the hybridization buffer and washedas follows: 5 min followed by 15 min in 2X SSC,0·1% SDS; three 1-h washes in 0·1X SSC, 0·5%SDS, the first at 37)C and the second two at 65)C.The membrane was then blotted dry and exposedto film.

Primer extension analysisCells were grown in YE (0·5% yeast extract, 2%

glucose) at 28)C to log phase and the culture wassplit into two. Half was maintained at 28)C whilethe other half was placed at 43)C for 30 min. RNAwas prepared as described previously (Galloet al., 1993). The oligonucleotide used for primer

. 14: 733–746 (1998)

extension of the ssp1+ gene was 5*-CCAGAAA+

737.

Figure 1. (A) Schematic representation of S. pombe hsf,showing the conserved regions, the location of the epitope tag,and the six deletion and two point mutations. The conserveddomains are the DNA binding domain (DBD), the trimeriza-tion domain (T) and the C-terminal hydrophobic repeat (CTR).The sequence of the CTR is shown, with the amino acids at the‘a’ and ‘d’ positions of the heptads in bold face; the amino acidschanges made for the point mutants are indicated with arrows.The tag is the 16 amino acid epitope derived from influenzaHA. The numbers refer to the amino acids. (B) Cross-speciescomparison of the CTR from human, D. melanogaster andS. pombe. The amino acids at the ‘a’ and ‘d’ positions of theheptads are shown in bold face.

epitope-tagged versions of the mutant genes. These

AGCACAATAGAAACC-3* and for the act1gene 5*-GCACATACCAGAGCCATTATCAA TAACCAACGC-3*. The 5* end of the primer waslabelled using [c-32P]ATP and T4 kinase (NewEngland Biolabs). 2 ng of labelled oligo wasmixed with 15 ìg of RNA in 80% formamide,40 m-PIPES pH 6·4, 400 m-NaCl, 1 m-EDTA, heated to 95)C for 5 min and then allowedto cool to 36)C. The 36)C incubation was contin-ued for approximately 16 h, after which an ethanolprecipitation was performed. The RNA-primerhybrid was resuspended in 50 m-Tris pH 8,5 m-MgCl2, 50 m-KCL, 5 m-DTT, 0·5 m ofeach of four dNTPs, 50 ìg/ml actinomycin D and15 units of RNAsin (Promega) and 1 unit AMVreverse transcriptase and incubated at 42)C for 1 h.The RNA was then degraded by adding RNAse A(Boehringer Mannheim) to a concentration of0·04 mg/ml, and the primer-extended DNA prod-ucts were run on a 6% sequencing gel alongsideDNA size markers. The gels were dried down andexposed to film, and a Phosphorimager was usedto quantify the amount of radioactivity. Thevalues displayed were obtained by adding thecontrol temperature (where visible) and heat-shock-induced ssp1+ values, and dividing the sumby the act1+ value.

RESULTS

Construction of the mutant strainsWe constructed six deletion mutations and two

point mutations in hsf+ (Figure 1A), all in theC-terminal two thirds of the protein. The twopoint-mutated hsfs, L497P and L483A/L487A,contain mutations of the conserved hydrophobicamino acids at the ‘a’ and ‘d’ positions of theheptad repeat within the CTR (see Figure 1B for across-species comparison of the CTR). Both ofthese mutations would be expected to severelylimit or eliminate the ability of the CTR to form acoiled-coil interaction (Rabindran et al., 1993; Zuoet al., 1994). In four of the deletion mutations,Ä476-495, Ä411-504, Ä476-609 and Ä411-609, theCTR is deleted. The two remaining mutants, Ä267-410 and Ä505-609, delete regions flanking theconserved domain. To determine the roles thatthese regions play in regulation of S. pombe hsf, weconstructed isogenic strains using the two-stepgene replacement method (Scherer and Davis,1979) to replace the genomic copy of hsf+ with


mutant alleles are under the control of the endog-enous hsf+ promoter. Successful replacement ofthe wild-type allele with Ä411-609 was notattained, suggesting that Ä411-609 cells areinviable. Hsf+ has been shown to be an essentialgene in S. pombe (Gallo et al., 1993). We were,however, able to relace the wild-type gene with theÄ411-609 mutant borne on a multi-copy plasmid,indicating that Ä411-609 is able to substitute for

. 14: 733–746 (1998)

The DNA binding activities of the mutant hsfs

738 . . .

Figure 2. Morphologies of strains expressing WT hsf orÄ267-410 using Nomarski optics. Wild-type cells are approxi-mately 7 ìm in length.

wild-type hsf when present in high copy. Thus, incontrast to the other mutants, Ä411-609 is
expressed from a multi-copy plasmid and is underthe control of the SV40 early promoter.
Aberrant morphologies have previously beenobserved for S. pombe strains expressing hsf from amulti-copy plasmid and for strains expressing D.melanogaster HSF (Gallo et al., 1993). We thuscompared the morphologies of the hsf mutantstrains to wild type and found that they werenormal with the exception of Ä267-410, wherethe cells appeared elongated (Figure 2, and seediscussion).

Meiosis has previously been proposed to involvethe stress response: in S. cerevisiae the heat shockgenes SSA3, HSP82 and HSC82 are activatedwhen cells are induced to sporulate (Werner-Washburne et al., 1989; Szent-Gyorgyi, 1995;Erkine et al., 1996), and in S. pombe, the stress-inducible uvi15+ gene is required for efficientsporulation (Lee et al., 1995). To determine if anyof the mutant strains had meiotic defects they werecrossed with a wild-type strain and tetrad analysiswas carried out. Four viable spores were recoveredconsistently for five of the mutants (Ä476-495,Ä505-609, Ä476-609, L497P and L483A/L487A)and the mutant and wild-type hsf genes were foundto segregate 2:2, but both Ä267-410 and Ä411-504had sporulation defects. For these two mutantsfewer than four spores per ascus were frequentlyobserved and these were often unable to germinatewhen placed on rich medium. Both Ä267-410 andÄ411-504 have reduced transcriptional abilities(see below) and thus may not be capable of induc-ing the appropriate genes to sufficiently high levelsfor efficient sporulation and germination. Tetradanalysis was not carried out for the Ä411-609mutant strain because loss of the hsf-containingplasmid during sporulation led to very low sporeviability.


To determine the effects of these mutations oninduction of DNA binding, we performed electro-phoretic mobility shift assays (EMSA) using theHSE as probe and cellular extracts made from cellsgrown under normal growth conditions or follow-ing heat shock (Figure 3). As expected, the endog-enous HSF binds DNA heat inducibly and thisbinding is not super-shifted by the 12CA5 anti-body (lanes 2–4, 23–25 and 30–32). The 12CA5antibody, which recognizes the HA epitope tagthat was added to the N terminus of all engineeredhsf genes used in this study, does cause a super-shift of the induced band from the control WTstrain, which contains the epitope tag (lanes 5–7and 33–35). With the exception of Ä267-410, wherethere is no detectable DNA binding (lanes 11–13),none of the integrated mutations had a significanteffect on hsf binding activity (lanes 8–10, 14–22and 36–41). The inducible DNA binding activityin each of these extracts can be shifted by the12CA5 antibody, verifying that it contains thehsf mutant. Upon supershift the induced bandsappear to lose intensity. This can partly beexplained by antibody cross-linking of numeroushsf/HSE complexes, thus forming a highermolecular weight aggregate that does not enter thegel and remains in the well. In addition, thebinding of the antibody may reduce the ability ofhsf to bind the HSE, especially given that theepitope tag is located at the N-terminus of theprotein and is hence in close proximity to the DNAbinding domain.

In contrast to the mother mutant hsfs, signifi-cant DNA binding was detected at control tem-perature for Ä411-609 (lane 26). We show byantibody supershift that the binding activity thatwe detect at control temperature contains theÄ411-609 mutant protein (lane 27), and we showbelow that this binding activity is not caused byoverexpression. Heat shock did not increase thebinding but the mobility of the complex increased(compare lanes 26 and 28). This may be due to aheat-induced modification or conformationalchange, or even to the loss of binding of anotherfactor(s) bound to the hsf/HSE complex.

From these results we can conclude that nosingle element located in the C-terminal third ofthe protein (amino acids 411–609) is essential forkeeping hsf repressed under non-shock conditionsor for induction of DNA binding. Notably, bothdeletion of and point mutation of the CTR yields

. 14: 733–746 (1998)

739.

Figure 3. The HSE binding activities of the mutant hsf proteins. Hsf-containing extracts were made from cells growncontinuously at 28)C and from cells heat shocked (HS) for 30 min at 43)C. 15 ìg of protein and 2 ng of probe were used in eachlane. The lanes marked endogenous represent the binding activities obtained from the parental strain and the lanes marked WTrepresent the binding activities obtained from the strain containing epitope-tagged, wild-type hsf. All of the mutant hsfs are epitopetagged and the heat-inducible binding activity is supershifted by the 12CA5 monoclonal antibody (Ab) to the epitope tag in allcases. The arrows indicate the heat-shock-induced and antibody-supershifted bands.

factors that are competent for heat-inducibleDNA-binding.

been observed for HSFs from other organisms,and these may represent differentially phosphor-

Expression levels of the mutant hsfsWe expected the integrated hsf mutants to be

expressed at levels comparable to wild typebecause they were homologously integrated at thehsf+ locus and were under the control of theendogenous promoter. In addition, six of the sevenintegrated mutants had DNA binding abilitiesindistinguishable from wild type. It was neverthe-less possible that the mutations affected the trans-latability of the mRNAs or that they affectedprotein stability. This was especially a concern forthe Ä267-410 mutant, where no DNA binding wasdetected, and for the Ä411-609 mutant, which waspresent in multi-copy.

Western blot analysis was used to examine theexpression level of tagged hsf in each strain (Figure4A). Lanes 1 and 2 show the background ofcross-reacting cellular proteins present in extractsfrom the parental strain containing only endogen-ous hsf. In lane 3 a prominent band correspondingto full-length hsf is visible and can be seen toundergo a mobility shift upon heat shock (lane 4).Several hsf species are present in each lane, as has


ylated forms of the protein (Larson et al., 1988;Sorger and Pelham, 1988; Sarge et al., 1993). Theheat-shock-induced mobility shift is also typical,and has been shown to be largely due to increasedphosphorylation of the factor upon heat shock(Gallo et al., 1991).

Despite exhibiting a wild-type level of heat-shock-induced DNA binding, the hsf expressionlevels were found to be reduced for Ä476-495,Ä411-504, Ä505-609 and Ä476-609 (Figure 4A,lanes 5 and 6 and 9–14). For Ä476-495 and Ä476-609 at control temperature and for Ä505-609 atboth temperatures this reduction was so dramaticthat no hsf specific bands were apparent. In con-trast, the Ä267-410 mutant was significantly in-creased in expression (lanes 7 and 8) precludingprotein level as an explanation for the undetectableDNA binding activity of this mutant (Figure 3).Also, increased protein level cannot explain theconstitutive binding seen for Ä411-609, because itslevel is similar to WT, despite it being expressedfrom a multi-copy plasmid (lanes 15–18). Thedifferent expression levels observed for the inte-grated deletion mutants cannot be explained by

. 14: 733–746 (1998)

740 . . .

Figure 4. Expression levels of the mutant hsf proteins. (A) Western blot using proteinextracts made from the deletion mutant strains grown continuously at 28)C or followinga 30-min heat shock (HS) at 43)C. The extracts were the same as those used in Figure 3,and 150 ìg was used in each lane. The blots were probed with the 12CA5 antibody to theepitope tag followed by an anti-mouse IgG/HRP conjugate, and developed by ECL.Lanes 1 and 2 represent the background bands derived from the parental straincontaining endogenous (untagged) hsf. The arrowheads indicate the tagged hsf specificbands. The band in the marker (M) lane represents antibody cross-reaction with the220 kDa marker protein, myosin. (B) Northern blot using RNA made from the deletionmutant strains grown continuously at 28)C or following a 30-min heat shock (HS) at43)C. The probe was made by random hexamer priming using a plasmid templatecontaining hsf+ and ura4+ sequences. (C) Western blot performed as described in (A)using protein extracts made from the point mutant strains. The arrows indicate thetagged hsf-specific bands. The protein extracts were the same as those used in Figure 3.

differences in transcript levels as these were foundto be similar as determined by Northern blot

In contrast to the integrated deletion mutants,

analysis (Figure 4B). The decrease in message levelthat is seen upon heat shock for both the hsf andthe control ura4 mRNAs is typical of non-heat-shock-inducible genes. The mobilities of the Ä505-609 and Ä476-609 mRNAs are the same as WTbecause these deletions were generated by insertionof a stop codon rather than by deletion of codingsequence (lanes 11–14).


the expression levels of the two point mutants werenot significantly different from WT, and they didnot change upon heat shock (Figure 4C). Thesepoint mutations and the deletion mutation Ä411-504 affect the CTR, yet in all cases inducibleDNA binding is observed (Figure 3) underconditions where similar levels of protein can bedetected at both normal and heat shock tempera-tures. We therefore conclude that mutation of the

. 14: 733–746 (1998)

741.

Figure 5. The apparent Kd of mutant Ä476-495 is the same as the WThsf. HSE binding abilities over a range of probe concentrations for WT(lanes 1–7) and Ä476-495 (lanes 8–14) are shown. The protein extractswere made from cells that had been heat shocked at 43)C for 30 min,and 5 ìg of protein extract was used in each lane. The following probeconcentrations were used: lanes 1 and 8, 0·075 n; lanes 2 and 9,0·125 n; lanes 3 and 10, 0·25 n; lanes 4 and 11, 0·5 n; lanes 5 and12, 1·25 n; lanes 6 and 13, 5 n; lanes 7 and 14, 7·5 n. The arrowindicates the hsf/HSE complex.

CTR does not lead to constitutive DNA binding type. Ä476-495 was significantly decreased inexpression compared to WT (Figure 4A), yet
by hsf.
The HSE binding affinity of the Ä476-495 mutantis the same as WT

While our studies demonstrate that the CTR isnot required for inducible DNA binding by hsf,they show that some deletion mutations in the Cterminus of the protein result in severely reducedhsf protein levels. Nevertheless, strains bearingsome of these mutations have normal levels of hsfbinding activity following heat shock. One possibleexplanation for the discrepancy between the DNAbinding abilities and the expression levels would bethat, although the mutations are not in or near theDNA binding domain, they may have caused anincreased affinity of the factor for its binding site.Alternatively, the affinities of the wild type andmutant molecules for the HSE may be identical, inwhich case the result can only be explained by ahigher proportion of the mutant hsf moleculesbeing active for DNA binding compared to wild


EMSA (Figure 3) revealed that their heat-shock-induced DNA binding abilities were the same. Tocompare the DNA binding affinity of WT andÄ476-495, we measured the apparent Kd values foreach of these proteins by determining the probeconcentration required for half maximal binding(Figure 5). The apparent Kd values for Ä476-495(0·19 n) and WT (0·17 n) were similar. Thus,the binding affinity of the mutant is essentially thesame as for WT hsf, and we conclude that a higherproportion of hsf in the mutant strain is active forDNA binding compared to WT.

It was possible that the increased fraction ofactive hsf in the Ä476-495 strain resulted from itbeing more readily activated by heat. If this weretrue, we might expect an altered temperature pro-file for induction of hsf in this strain. We measuredthe amount of hsf binding in WT and Ä476-495strains at 28)C (normal growth temperature) andfollowing 30 min of incubation at either 34)C,

. 14: 733–746 (1998)

742 . . .

Figure 6. The temperature profiles of activation of DNA binding of WTand Ä476-495. Hsf-containing extracts were made from cells grown continu-ously at 28)C and from cells heat shocked for 30 min at 34)C, 39)C and43)C. DNA binding assays were carried out as described for Figure 3. Thelevel of binding at each temperature was quantitated and is expressed as apercentage of the maximal binding. The values shown represent the mean ofthree experiments and the error bars represent the standard deviations.

39)C or 43)C. In contrast to the WT strain, maxi- detectable HSE binding activity (Ä267-410) isnonetheless viable, despite the fact that hsf+ is an
mal induction in the mutant strain occurred at
39)C (Figure 6). This lowered temperature formaximal induction might contribute to theincrease in the fraction of hsf that is active at heatshock temperatures in this strain, and mightindicate a role for the CTR in sensing temperature(see Discussion).

Transcriptional abilities of the integrated HSFmutants

The data presented above demonstrate that sev-eral strains that contain varying amounts of hsfprotein have similar levels of HSE binding activity.They also demonstrate that one strain that has no


essential gene. To determine how these bindingresults correlated with function of hsf in vivo, wemeasured the heat-induced transcriptional acti-vation of the ssp1+ and Bip+ genes. The nuclearssp1+ gene encodes a mitochondrial hsp70 (Kasaiand Isono, 1991), and the Bip+ gene encodes aheat shock protein localized to the endoplasmicreticulum (Pidoux and Armstrong, 1992). Bothpromoters were regulated in a similar fashion; weshow data for the ssp1+ promoter for the deletionand point mutants (Figure 7). The mRNA levels ofthe non-heat-shock-inducible actin gene, act1+,are depicted in the lower panel. Upon heat shock

. 14: 733–746 (1998)

743.

+

Figure 7. Primer extension analysis of ssp1+ RNA isolated from the hsf deletion and point mutant strains. The RNA wasprepared from cells grown continuously at 28)C or following a 30-min heat shock (HS) at 43)C. The upper panel shows the threeheat-induced ssp1+ primer extension products, and the lower panel the product from the act1+ message. The control temperaturessp1+ extension product is visible in some of the lanes. The numbers below the upper panel represent quantitation of the ssp1+

extension products normalized to the act1+ product as determined by the Phosphorimager. All values above 50 were significantlyabove background. The lane marked M represents the DNA molecular weight markers.

the ssp1 transcripts start site shifts approximately transcriptional activation are separable steps in hsfactivation. Both of the point mutants activated
50 bases downstream, and the gene is induced
approximately 50-fold. Three major heat-shock-induced transcription start sites are utilized, yield-ing a set of three induced bands. The low level ofthe control temperature transcript is visible insome of the lanes. The values presented for ssp1+

transcript levels were obtained by summingthe control transcript (where visible) and theheat-induced transcripts and dividing by the actinlevel. Comparison of lanes 3 and 4, which show thessp1+ transcript levels for the WT strain, withlanes 1 and 2, which represent those for theparental strain containing only endogenous hsf,shows that the epitope tag does not affect thefunction of hsf.

All six deletion mutants were able to heat induc-ibly activate transcription, although the inducedssp1+ level was reduced compared to WT in allcases, with the exception of the smallest deletionmutant, Ä476-495 (lanes 3–20). The basal level ofactivity for the Ä411-609 mutant, which boundDNA constitutively, was somewhat elevated.Interestingly, it still activated transcription heatinducibly, indicating that DNA binding and


transcription heat inducibly, yet the induced tran-script levels were somewhat reduced compared towild type (lanes 21–28).

Thus, hsf is functional in all mutant strains butthe ability of the factor to activate transcription isdecreased in several of them. This decrease inactivation does not correlate with levels of bindingactivity. For example, transcriptional induction inthe Ä267-410 strain is greater than in the Ä411-504strain, while DNA binding activity as assayed byEMSA is too low to be measured in the Ä267-410strain and the Ä411-504 strain has wild-type levelsof binding activity.

DISCUSSION

One of the primary responses of most eukaryoticcells to stress is the induction of HSF DNAbinding ability. It is therefore critical to under-stand the molecular mechanisms that regulate thisinduction. The role of the CTR in this processhas previously been characterized using D.

. 14: 733–746 (1998)

melanogaster and human heat shock factors in the CTR is not required to maintain D. mela-nogaster HSF in an inactive state in normally

744 . . .

experimental systems that have, by necessity,required ectopic expression of hsf in cells that alsoexpress endogenous hsf. We demonstrate here thatthe CTR is not required for induction of S. pombehsf following replacement of the wild-type hsf genewith mutant versions in an otherwise isogenicbackground. Neither point mutation nor deletionof the CTR caused constitutive binding by hsf oraffected the induced level of hsf binding at 43)C(Figures 3 and 4). We have, however, found thatremoval of the C-terminal 199 amino acids ofhsf results in constitutive DNA binding. Thus,S. pombe hsf, like those of D. melanogaster andhumans, does not require any heat-inducedpositively acting modification to bind DNA;rather, the ability to bind DNA is intrinsic to theprotein and heat shock releases repressive inter-actions. These repressive interactions map to the Cterminus of the protein.

The C termini of D. melanogaster HSF andhuman HSF1 have also been implicated in theregulation of DNA-binding; however in theseinstances most of the data favour a primary rolefor the CTR in this regulation. The lack of a strictrequirement for the CTR observed in our studymay be due to differences in the mechanism ofregulation between the metazoan and S. pombeHSFs, or to differences in HSF expression levelsbetween the studies that have been conducted. Theoverall structure of HSF is conserved betweennumerous organisms; however, it is possible thatthe role of the CTR in these organisms hasdiverged. Human HSF2 contains the CTR, yet it isnot heat inducible for DNA binding. In addition,the yeast K. lactis HSF contains the CTR, yet itbinds DNA constitutively. This suggests that theCTR may play other roles that do not involveheat-induced DNA binding. Alternatively, it ispossible that previous studies reporting a require-ment for the CTR in maintaining HSF in aninactive form at control temperature (and thusmaintaining the ability of HSF to be heat-induced)were affected by over-expression of the mutantproteins. Experiments on D. melanogaster HSFand human HSF1 DNA binding induction wereperformed by expressing these factors followingtransient introduction into cells, which results inhigh-level expression. This might have titrated outa factor(s) that normally maintains HSF in aninactive state, and therefore uncovered a role forthe CTR in maintaining an inactive state that isnot normally essential. The recent observation that


growing cells when expression is from an inte-grated promoter is consistent with this latterhypothesis (Zandi et al., 1997).

Although our results show that deletion of theCTR does not derepress S. pombe hsf under nor-mal growth conditions, we have found that it doesalter its temperature profile for heat induction(Figure 6). This might indicate that the CTR playsa direct role as part of the ‘molecular thermometer’that determines the temperature at which hsfbecomes activated. Thus, the CTR might play arole in repressing DNA binding at intermediatelevels of heat shock, while other interactions mightbe sufficient to repress hsf binding at normalgrowth temperatures. Alternatively, the CTR dele-tion might alter hsf’s structural properties in such away that it indirectly alters the mechanism(s) thatsenses temperature.

Analysis of the effects of the deletion mutantsreveals mechanisms of regulation of S. pombe hsfapart from the regulation of its DNA bindingability. A large deletion of the C-terminus, extend-ing up to amino acid 410, is deregulated for DNAbinding, yet despite this constitutive DNA binding,Ä411-609 was able to heat inducibly activate tran-scription. This indicates that in S. pombe, as in thehigher eukaryotes, induction of hsf occurs by atwo-step process, induction of DNA binding fol-lowed by induction of transcriptional activation.We were unable to detect any DNA binding abilityfor Ä267-410 by the electromobility shift assay. Webelieve that despite our inability to detect bindingin vitro, it is in fact binding in vivo because it isclearly able to activate heat shock gene transcrip-tion as measured by primer extension of the ssp1+

gene (Figure 7) and of the Bip+ gene (data notshown). It is possible that the affinity of theÄ267-410 mutant for its binding site is reduced to apoint below the detection limit of the electromo-bility shift assay, but that this lower level ofbinding is sufficient to activate transcription.Alternatively, other factors that are present in vivomight increase binding.

An interesting feature of the Ä267-410 mutant isthat it has an elongated morphology compared towild-type cells. This suggests that it has caused adefect in cell cycle progression consistent with anextension of the G2 phase. Findings linking thestress response and cell cycle progression havepreviously been made. In human cells hsp70mRNA levels increase at the G1/S boundary and

. 14: 733–746 (1998)

then decline to undetectable levels by the end of S Boeke, J. D., Trueheart, G., Natsoulis, G. and Fink,G. R. (1987). 5-Fluoro-orotic acid as a selective agent

745.

phase (Milarski and Morimoto, 1986) and inS. cerevisiae a mutation in HSF was shown tocause a retardation of the cell cycle in G2 (Smithand Yaffe, 1991). These data suggest that induc-tion of the heat shock response may be requiredfor progression through various points in the cellcycle. This would also explain the requirement forhsf activity under non-stressful conditions, as evi-denced by the lethal phenotype of hsf null strains(Gallo et al., 1993).

Several of the mutants were expressed at lowlevel at normal growth temperatures and at signifi-cantly higher levels following heat shock (e.g.mutants Ä476-495 and Ä476-609, Figure 4A).Because the hsf message levels are not heat induc-ible (Figure 4B) this effect can only be explained bya post-transcriptional mechanism. Perhaps themutants are less vulnerable to degradation in theDNA binding form. Of the six deletion mutantsthat we analysed, four had significantly reducedexpression levels, while their heat-shock-inducedDNA binding abilities remained the same as forWT. For Ä476-495 this was shown not to be due toan increased affinity of the mutant hsf for the HSE,so must result from an increase in the percentageof molecules active for DNA binding. Thissuggests that a feedback mechanism exists thatmodulates the amount of hsf in the cell that isactive for DNA binding. Evidence for a feedbackmechanism involving hsp70 exists in S. cerevisiaeand D. melanogaster (Boorstein and Craig, 1990;Solomon et al., 1991). Thus, our results argueagainst a mode of regulation where the trimeriz-ation domain/CTR interaction is primary.Rather, we favour a model involving multiplecontacts in the C terminus of hsf and involvingother proteins in the regulation of S. pombe hsfDNA binding.

ACKNOWLEDGEMENTS

We wish to thank Tamar Enoch for assistance inperforming the tetrad dissection and analysis. Wealso thank Jon Lorsch, Gavin Schnitzler, DavidGross and an anonymous reviewer for helpfulcomments on the manuscript.

REFERENCES

Ausubel, F. M., Brent, R., Kingston, R. E., et al. (Eds)(1989). Current Protocols in Molecular Biology. JohnWiley and Sons.


in yeast molecular genetics. Methods Enzymol. 154,164–175.

Boorstein, W. R. and Craig, E. A. (1990). Transcrip-tional regulation of SSA3, an HSP70 gene fromSaccharomyces cerevisiae. Mol. Cell. Biol. 10, 3262–3267.

Clos, J., Westwood, J. T., Becker, P. B., Wilson, S.,Lambert, K. and Wu, C. (1990). Molecular cloningand expression of a hexameric Drosophila heat shockfactor subject to negative regulation. Cell 63, 1085–1097.

Erkine, A. M., Adams, C. C., Diken, T. and Gross, D. S.(1996). HSF gains access to the yeast HSC82promoter independently of binding sites for otherregulatory factors and antagonizes nucleosomalrepression of basal and induced transcription. Mol.Cell. Biol..

Gallo, G. J., Prentice, H. and Kingston, R. E. (1993).Heat shock factor is required for growth at normaltemperatures in the fission yeast Schizosaccharomycespombe. Mol. Cell. Biol. 13, 749–761.

Gallo, G. J., Schuetz, T. J. and Kingston, R. E. (1991).Regulation of heat shock factor in Schizosaccharomy-ces pombe more closely resembles regulation in mam-mals than in Saccharomyces cerevisiae. Mol. Cell.Biol. 11, 281–288.

Giardina, C. and Lis, J. T. (1995). Dynamic protein-DNA architecture of a yeast heat shock promoter.Mol. Cell. Biol. 15, 2737–2744.

Green, M., Schuetz, T. J., Sullivan, E. K. and Kingston,R. E. (1995). A heat shock-responsive domain ofhuman HSF1 that regulates transcription activationdomain function. Mol. Cell. Biol. 15, 3354–3362.

Harrison, C. J., Bohm, A. A. and Nelson, H. C. M.(1994). Crystal structure of the DNA binding domainof the heat shock transcription factor. Science 263,224–227.

Hightower, L. E. (1991). Heat shock, stress proteins,chaperones, and proteotoxicity. Cell 66, 191–197.

Jakobsen, B. K. and Pelham, H. R. B. (1991). Aconserved heptapeptide restrains the activity of theyeast heat shock transcription factor. EMBO J. 10,369–375.

Kasai, H. and Isono, K. (1991). Dual modes of tran-scriptional and translational initiation of SSP1, thegene for a mitochondrial HSP70, responding to heat-shock in Schizosaccharomyces pombe. Nucl. Acids Res.19, 5331–5337.

Kingston, R. E., Schuetz, T. J. and Larin, Z. (1987).Heat-inducible human factor that binds to a humanhsp70 promoter. Mol. Cell. Biol. 7, 1530–1534.

Kunkel, T. A. (1985). Rapid and efficient site-specificmutagenesis without phenotypic selection. Proc. Natl.Acad. Sci. USA 82, 488–492.

Larson, J. S., Schuetz, T. J. and Kingston, R. E. (1988).Activation in vitro of a sequence-specific DNA

. 14: 733–746 (1998)

binding by a human regulatory factor. Nature 335,372–375.

Scherer, S. and Davis, R. W. (1979). Replacement ofchromosome segments with altered DNA sequences

746 . . .

Lech, K., Anderson, K. and Brent, R. (1988). DNA-bound Fos proteins activate transcription in yeast.Cell 52, 179–184.

Lee, J. K., Kim, M., Choe, J., Seong, R. H., Hong, S. H.and Park, S. D. (1995). Characterization of uvi15+, astress-inducible gene from Schizosaccharomycespombe. Mol. Gen. Genet. 246, 663–670.

Lindquist, S. and Craig, E. A. (1988). The heat shockproteins. Annu. Rev. Genet. 22, 631–677.

Lis, J. and Wu, C. (1993). Protein traffic on the heatshock promoter: parking, stalling, and trucking along.Cell 74, 1–4.

Milarski, K. and Morimoto, R. I. (1986). Expression ofhuman HSP70 during the synthetic phase of the cellcycle. Proc. Natl. Acad. Sci. USA 83, 9517–9521.

Peteranderl, R. and Nelson, H. C. M. (1992). Trimeri-zation of the heat shock transcription factor by atriple-stranded alpha-helical coiled coil. Biochemistry31, 12,272–12,276.

Pidoux, A. L. and Armstrong, J. (1992). Analysis of theBip gene and identification of an ER retention signalin Schizosaccharomyces pombe. EMBO J. 11, 1583–1591.

Prentice, H. L. (1992). High efficiency transformation ofSchizosaccharomyces pombe by electroporation. Nucl.Acids Res. 20, 621.

Rabindran, S. K., Giorgi, G., Clos, J. and Wu, C.(1991). Molecular cloning and expression of a humanheat shock factor, HSF1. Proc. Natl. Acad. Sci. USA88, 6906–6910.

Rabindran, S. K., Haroun, R. I., Clos, J., Wisiewski, J.and Wu, C. (1993). Regulation of heat shock factortrimer formation: role of a conserved leucine zipper.Science 259, 230–234.

Ruden, D. M., Ma, J., Li, Y., Wood, K. and Ptashne,M. (1991). Generating yeast transcriptional activatorscontaining no yeast protein sequences. Nature 350,250–252.

Sarge, K. D., Murphy, S. P. and Morimoto, R. I. (1993).Activation of heat shock gene transcription by heatshock factor 1 involves oligomerization, acquisition ofDNA-binding activity, and nuclear localization andcan occur in the absence of stress. Mol. Cell. Biol. 13,1392–1407.

Sarge, K. D., Zimarino, V., Holm, K., Wu, C. andMorimoto, R. I. (1991). Cloning and characterizationof two mouse heat shock factors with distinct induc-ible and constitutive DNA-binding ability. Genes &Dev. 5, 1902–1911.

Scharf, K.-D., Rose, S., Zott, W., Schoff, F. and Nover,L. (1990). Three tomato genes code for heat stresstranscription factors with a region of remarkablehomology to the DNA-binding domain of the yeastHSF. EMBO J. 9, 4495–4501.


constructed in vitro. Proc. Natl. Acad. Sci. USA 76,4951–4955.

Schuetz, T. J., Gallo, G. J., Sheldon, L., Tempst, P. andKingston, R. E. (1991). Isolation of a cDNA forHSF2: evidence for two heat shock factor genes inhumans. Proc. Natl. Acad. Sci. USA 88, 6911–6915.

Sheldon, L. A. and Kingston, R. E. (1993). Hydro-phobic coiled-coil domains regulate the subcellularlocalization of human heat shock factor 2. Genes &Dev. 7, 1549–1558. Correction, Genes & Dev. 8, 386.

Smith, B. J. and Yaffe, M. P. (1991). A mutation in theyeast heat-shock factor gene causes temperature-sensitive defects in both mitochondrial protein importand the cell cycle. Mol. Cell. Biol. 11, 2647–2655.

Solomon, J. M., Rossi, J. M., Golic, K., McGarry, T.and Lindquist, S. (1991). Changes in hsp70 alterthermotolerance and heat-shock regulation inDrosophila. New Biol. 3, 1106–1120.

Sorger, P. K. and Nelson, H. C. M. (1989). Trimeriza-tion of a yeast transcriptional activator via a coiled-coil motif. Cell 59, 807–813.

Sorger, P. K. and Pelham, H. R. B. (1988). Yeast heatshock factor is an essential DNA-binding protein thatexhibits temperature-dependent phosphorylation. Cell54, 855–864.

Szent-Gyorgyi, C. (1995). A bipartite operator interactswith a heat shock element to mediate early meioticinduction of Saccharomyces cerevisiae HSP82. Mol.Cell. Biol. 15, 6754–6769.

Tanaka, M. and Herr, W. (1990). Differential transcrip-tional activation by Oct-1 and Oct-2: interdependentactivation domains induce Oct-2 phosphorylation.Cell 60, 375–386.

Werner-Washburne, M., Becker, J., Kosic-Smithers, J.and Craig, E. A. (1989). Yeast Hsp70 RNA levels varyin response to the physiological status of the cell. J.Bacteriol. 171, 2680–2688.

West, R. W. J., Yocum, R. R. and Ptashne, M. (1984).Saccharomyces cerevisiae GAL1-GAL10 divergentpromoter region: location and function of the up-stream activator sequence UASG. Mol. Cell. Biol. 4,2476–2478.

Wiederrecht, G., Seto, D. and Parker, C. S. (1988).Isolation of the gene encoding the S. cerevisiae heatshock transcription factor. Cell 54, 841–853.

Zandi, E., Tran, T.-N. T., Chamberlain, W. and Parker,C. S. (1997). Nuclear entry, oligomerization, andDNA binding of the Drosophila heat shock transcrip-tion factor are regulated by a unique nuclear localiza-tion sequence. Genes & Dev. 11, 1299–1314.

Zuo, J., Baler, R., Dahl, G. and Voellmy, R. (1994).Activation of the DNA-binding ability of human heatshock factor 1 may involve the transition from anintramolecular to an intermolecular triple-strandedcoiled structure. Mol. Cell. Biol. 14, 7557–7568.

. 14: 733–746 (1998)

the c-terminal hydrophobic repeat ofschizosaccharomyces pombe heat shock factor is not required for...

Documents