Mol Gen Genet (1997) 255:322±331 Ó Springer-Verlag 1997

ORIGINAL PAPER

O. Boscheinen á R. Lyck á C. Queitsch á E. Treuter V. Zimarino á K.-D. Scharf Heat stress transcription factors from tomato can functionally replace HSF1 in the yeast Saccharomyces cerevisiaeÃ

Received: 1 October 1996 / Accepted: 6 December 1996

Abstract The fact that yeast HSF1 is essential for sur- nomeric Hsf has a markedly reduced anity for DNA, as vival under nonstress conditions can be used to test shown by lacZ reporter and band-shift assays. heterologous Hsfs for the ability to substitute for the endogenous protein. Our results demonstrate that like Key words Tomato á Yeast á Heat stress á Transcription Hsf of Drosophila, tomato Hsfs A1 and A2 can func- factor á Thermotolerance tionally replace the corresponding yeast protein, but Hsf B1 cannot. In addition to survival at 28° C, we checked the transformed yeast strains for temperature sensitivity Introduction of growth, induced thermotolerance and activator func- tion using two di€erent lacZ reporter constructs. Tests The striking conservation of essential elements of the with full-length Hsfs were supplemented by assays using heat stress response is well documented by the 11 heat mutant Hsfs lacking parts of their C-terminal activator stress protein (Hsp) families, with representatives found region or oligomerization domain, or containing amino in all organisms investigated so far. As molecular acid substitutions in the DNA-binding domain. Re- chaperones, they form part of a homeostatic network markably, results with the yeast system are basically responsible for the proper folding, assembly, intracellu- similar to those obtained by the analysis of the same Hsfs lar distribution and degradation of proteins (see sum- as transcriptional activators in a tobacco protoplast as- maries by Nover 1991; Buchner 1996; Hartl 1996; say. Most surprising is the failure of HsfB1 to substitute Waters et al. 1996; Nover and Scharf 1997). Following for the yeast Hsf. The defect can be overcome by addition the initial identi®cation of the recognition element in the to HsfB1 of a short C-terminal peptide motif from HsfA2 promoters of Drosophila heat stress genes (Pelham and (34 amino acid residues), which represents a type of Bienz 1982), it soon became apparent that the conser- minimal activator necessary for interaction with the yeast vation also extends to heat stress elements (HSEs) in transcription apparatus. Deletion of the oligomerization other organisms, which show the basic structure of two domain (HR-A/B) does not interfere with Hsf function palindromic 5 bp modules in repetitive arrangements for survival or growth at higher temperatures. But mo- (-AGAAnnTTCT-). They are found upstream of the TATA box in all eukaryotic heat stress-inducible genes (Nover 1987, 1991). The long search for the putative regulatory proteins (Hsfs) binding to the HSE led to the Communicated by R. Herrmann isolation of the yeast HSF gene (Sorger and Pelham O. Boscheinen á R. Lyck á K.-D. Scharf (&) 1988; Wiederrecht et al. 1988) and subsequently to the Molecular Cell Biology, Biocenter of the J.W. Goethe University, corresponding cDNA clones from Drosophila (Clos et al. Marie-Curie-Str. 9, D-60439 Frankfurt, Germany Tel.: +49-69-798-29285; Fax: +49-69-798-29286 1990) and tomato (Lycopersicon peruvianum) (Scharf e-mail: [email protected] et al. 1990, 1993). A surprising peculiarity of the plant system was the identi®cation of three Hsf clones. Two O. Boscheinen á C. Queitsch á E. Treuter á K.-D. Scharf Institute of Plant Biochemistry, Weinberg 3, represented heat stress-inducible genes themselves, and D-06120 Halle, Germany this unique trait was also found in other plants (HuÈ bel V. Zimarino and SchoÈ ‚ 1994; Czarnecka-Verner et al. 1995; Gag- Biol. Technol. Res. Dept., San Ra€aele Sci. Institute, liardi et al. 1995). Via Olgettina 58, I-20132 Milan, Italy More than 25 Hsfs have been cloned and sequenced * Dedicated to Prof. B. Parthier (Halle) on the occasion of his 65th from di€erent plants, vertebrates, Xenopus, and yeasts birthday (Nover et al. 1996). All proteins of this Hsf family have a 323 number of common features, which are summarized in motif from the C-terminus of HsfA2. The collection of Fig. 1 and Table 2. An N-terminal DNA-binding do- yeast strains described in this paper is a valuable tool for main contains a central helix-turn-helix motif involved analysis of Hsf function and for screening of interacting in DNA recognition (Harrison et al. 1994; Vuister et al. proteins. 1994; Schultheiss et al. 1996). The C-terminal region includes an oligomerization domain with a characteristic heptad repeat pattern of large hydrophobic amino acid Materials and methods residues (HR-A/B region), a nuclear localization signal (NLS, Lyck et al. 1997) and the C-terminal activator Yeast strains and culture media region (Treuter et al. 1993). Haploid strain RSY4 of Saccharomyces cerevisiae used for func- Considering the conservation of essential functional tional Hsf substitutions was derived from strain RSY10 (MATa/ parts of eukaryotic Hsfs (Scharf et al. 1994; Wu 1995; MATa, ade2/ade2, ade6, can1/can1, his3,11,15/his3,11,15, leu2-3, Nover et al. 1996) and the reports that the yeast HSF 112/leu2-3,112, trp1-1/trp1-1, ura3-1/ura3-1), a kind gift of Mark Vidal (Northwestern University, Evanston, Io.). Growth and gene is indispensable for survival and growth even at transformation of yeast strains and selection on 5-¯uoro-orotic 28° C (Sorger and Pelham 1988; Wiederrecht et al. 1988), acid (FOA) followed standard protocols (Ausubel et al. 1993; Rose we initiated investigations of the function of heterolo- et al. 1990). For chromosomal disruption of the HSF1 gene (Wi- gous Hsfs in yeast. In this paper, we demonstrate (i) that ederrecht et al. 1988) the parental strain was transformed with a derivative of pRS303(HIS3) containing Sc-HSF1 3¢ (EcoRV-XhoI, the Drosophila Hsf and the tomato Hsfs A1 and A2 are 2772-3676) and 5¢ (EcoRI-BamHI, 1-991) fragments, inserted in able to replace the yeast Hsf; (ii) that, at least for the tandem but reverse order in the KS polylinker. Prior to yeast tomato Hsfs A1 and A2, the structural requirements for transformation, plasmid DNA was linearized with EcoRI to re- activity in yeast are very similar to those found in to- move a short polylinker segment separating the two inserts. Inte- bacco protoplasts (Treuter et al. 1993); and (iii) that the gration at the HSF locus results in excision of 71.4% of the coding region. The remaining 5¢ sequences up to position 991, encoding the lack of function of a third tomato Hsf, B1, in the yeast 66 N-terminal amino acids of Sc-Hsf1, are insucient for yeast system can be overcome by adding a short activator viability (Sorger 1990).

Fig. 1 Basic structure of Hsfs. Structure of the endogenous yeast Hsf residues. The C-terminal part contains the nuclear localization signal (Sc-Hsf1) is compared with those of Drosophila (Dm-Hsf) and of the (NLS), an additional heptad hydrophobic repeat region (HR-C)and three tomato Hsfs (Lp-HsfA1, A2 and B1). The N-terminal DNA- activator motifs (AHA1, AHA2) close to the HR-C region. The boxed binding domain (DBD) is separated from the C-terminal activator numbers below the block diagrams of the tomato Hsfs indicate domain by an oligomerization domain (HR-A/B). Note that in Lp- positions of restriction sites introduced for generation of the deletion Hsfs A1 and A2 the latter is extended by an insert of 21 amino acid and/or fusion constructs (see Table 1 and Materials and methods) 324

Plasmid constructions were raised in rabbits against recombinant (His)6-tagged Lp-Hsfs as antigens (Scharf et al. unpublished). Strain DH5a of Escherichia coli (F) /80d, lacZDM15 endA1 recA1 ) + ) hsdR17 (rK mK ) supE44 thi-1d gyrA96 D(lacZYA-argE)U169) was used for cloning and propagation of all shuttle vectors. Stan- b-Galactosidase assay dard procedures were used for cloning (Sambrook et al. 1989) and sequencing (Sanger et al. 1977). To assay the transcriptional activity of heterologous Hsf in Sc-Hsf The Dm-HSF cDNA, a 2.5 kb EcoRI fragment containing the substitution strains after FOA treatment, cells were transformed coding region and 5¢ (21 bp) and 3¢ (473 bp) untranslated se- with reporter plasmids pZJHSE2-137 (Slater and Craig 1987), quences inserted in pBluescript KS+ (Clos et al. 1990) was excised kindly provided by E. Craig, or pHSE2-BG (Sorger and Pelham with HindIII and NotI and directionally inserted in the yeast ex- 1987), kindly provided by P. Sorger. Both plasmids contain HSEs pression vector pADNS (2l, LEU2 based) (Colicelli et al. 1989) upstream of a truncated CYC1 promoter fused to the lacZ reporter between the ADH1 gene promoter and terminator. The related gene (see Fig. 6). The b-galactosidase assay was carried out as vector pAD4D (Ballester et al. 1989) was used for subcloning a described by Ellwood and Craig (1984) and Bonner et al. (1992). PvuII-EcoRI (EcoRI ®lled-in) fragment containing the coding re- gion of Sc-HSF1 (Wiederrecht et al. 1988) into the SmaI site of the vector. This vector was also used for expression of the wild-type Cross-linking forms and all C-terminal deletions of the tomato Hsfs. The corre- sponding DNA fragments were obtained as XhoI-XbaI(XbaI ®lled- Native protein extracts were prepared as described above, diluted in) fragments from the plant expression vectors described by in bu€er H to approximately 0.5 mg/ml total protein and adjusted Treuter et al. (1993), and subcloned into pAD4D digested with SalI to 150 mM NaCl. Aliquots of 100 ll were incubated for 20 min at and SmaI. room temperature before adding the cross-linking reagent ethyl- Site-directed mutagenesis (Kunkel et al. 1985) was used to eneglycol-bis (sulfosuccinimidylsuccinate) (EGS) at the indicated create unique SalI sites at di€erent positions in the three tomato concentrations. After a 30 min incubation the reaction was stopped Hsfs (Table 1) as well as mutants of the DNA-binding domain of by addition of lysine to 30 mM. Proteins were precipitated with Lp-HsfB1 (see Fig. 3). The SalI mutants were used to generate trichloroacetic acid, washed with absolute ethanol and dissolved in internal deletions (Table 2, groups II and III) or chimeric Hsfs SDS sample bu€er (50 mM TRIS-HCl, pH 6.8, 2% SDS, 10% (Table 2, groups IV and V). All mutants were initially constructed glycerol, 2.5% b-mercaptoethanol). The cross-linked products were in plant expression vectors and tested for their transcriptional ac- separated on 6% SDS-polyacrylamide gels and detected by im- tivity in the transient expression assay using tobacco protoplasts munoblotting as described above. (see also Table 2). The SalI mutations are silent with respect to the activator function of the Hsfs. They were subcloned into the yeast expression vector as described above. Instead of pAD4D a modi®ed Band-shift assays form, pAD5D, was used in which the multicloning site between the SalI and SacI recognition sequences was replaced by a linker (5¢- To assay for DNA-binding activity, native protein extracts con- taining about 30 lg total protein were adjusted to 150 mM NaCl tcgaccatggtacctagggcccgagct-3¢) to introduce a unique AvrII re- 32 striction site. This simpli®ed the directional subcloning of XhoI- and incubated with 33 fmol of P-labeled double-stranded HSE XbaI fragments. oligonucleotides in the presence of 4 lg/ll BSA, 200 ng/ll poly(dI- dC) and 5% Ficoll. After 20 min incubation at room temperature the samples were loaded on a 5% native polyacrylamide gel in 0.5 ´ TBE (Sambrook et al. 1989) and separated for 2.5 h at 160 V. Immunoblot analysis Two types of HSE oligonucleotides were used in this study with Hsf-binding motifs similar to those in the lacZ reporter constructs For protein extraction, cells from a 50 ml liquid culture grown to described above (Fig. 6): HSE3 (TCGAGCCAGAAGCTTCTA- an OD at 600 nm of 1.0 were resuspended in bu€er H (25 mM GAAAGC) with three binding motifs and HSE6 (TCGAG- HEPES, pH 7.5, 10% glycerol, 1 mM EDTA, 50 mM NaCl, 5 mM GATCCTAGAAGCTTCCAGAAGCTTCTAGAAGCAGATC) MgCl2, 0.5% b-mercaptoethanol, 10 lg/ml Pefabloc, 1 lg/ml providing one partial and ®ve perfect binding motifs. pepstatin A, 1 lg/ml leupeptin, 2 lg/ml aprotinin) and lysed by sonication. After addition of NaCl to 400 mM and Nonidet P40 to 0.2% the homogenate was kept on ice for 20 min, centrifuged for 15 min at 23 000 g and the supernatant (native protein extract) was Results stored at )70° C. Protein concentrations were determined by the Bradford method (Bradford reagent from BioRad). Survival of yeast cells with heterologous Hsfs About 20 lg protein was loaded for separation on 12% SDS- polyacrylamide gels, transferred to PVDF membrane (DuPont- NEN) and processed for Hsf detection using the ECL detection kit For construction of yeast strains expressing heterolo- (Amersham) as indicated by the manufacturer. Polyclonal antisera gous Hsfs, we used the standard technology, starting

Table 1 Lp-Hsf mutants with unique restriction sites and ex- Hsf mutant Restriction site, generated Amino acid residues exchanged changed amino acid residues. Mutants were used to generate HsfA1.6 Sal I Deletion of 322±370 ® VD the deletion and fusion con- HsfA1.2 Sal I G447A ® VD structs compiled in Table 2 (for HsfA2.7 Sal I I137E ® VD further explanation see text and HsfA2.8 Sal I L211D ® VD Fig. 1) HsfA2.9 Sal I S240D ® VD HsfA2.10 Sal I I260E ® VD HsfA2.2DC323 Sal I M289E ® VD HsfB1.25 Sal I V143D ® VD HsfB1.26 Sal I L212E ® VD HsfB1.23 Sal I G271G ® VD Table 2 Survey, growth and expression of the lacZ reporter construct in yeast strains with homologous and di€erent heterologous Hsf expression cassettes on 2l plasmids. The Hsfs expressed in the indicated yeast strains are given with their short name (Nover et al. 1996) and a block diagram indicating the size (number of codons) as well as the presence and position of functional parts (see Fig. 1) Sc Saccharomyces cerevisiae, Dm Drosophila melanogaster, Lp Lycopersicon peruvianum, DBD DNA-binding domain, n.d. not determined

Group/Hsf Growth b-Galactosidasec No. of On Temperature Activity Inducibility Tobacco codons FOAa limitb (° C) protoplastsd

I1Sc-Hsf1 833 + 37 222 3.6 ) 2 Dm-Hsf 691 + 37 594 6.8 + 3 Lp-HsfA1 505 + 33 318 3.0 + 4 Lp-HsfA2 351 + 35 346 2.2 + 5 Lp-HsfB1 301 )) ) ) +

II 1 Lp-HsfA1(WT) 505 + 33 318 3.0 + 2A1DC468 449 + 33 173 7.5 + 3A1DC451 429 + 33 128 3.0 + 4A1DC396 376 )) ) ) ) 5A1D6/2 380 + 33 51 12.0 + 6A1D6/2,C491 344 + 33 166 8.0 +

III 1 Lp-HsfA2(WT) 351 + 35 346 2.3 + 2A2DC323 328 + 35 225 1.5 + 3A2DC300 304 + 35 113 2.5 + 4A2DC285 285 )) ) ) ) 5A2D7/8,C323 254 + £ 35 )) +

IV Lp-Hsf fusion proteins 1 A2.7 ´ B1.25 295 )) ) ) n.d. 2 A2.8 ´ B1.26 300 )) ) ) n.d. 3 B1.25 ´ A2.7 357 + 35 323 1.8 + 4 B1.26 ´ A2.8 352 + 35 124 1.6 + 6 B1.23 ´ A2.10 362 + 35 77 4.0 + 7 B1.23 ´ A2.2DC323 310 + 35 118 8.0 + V DBD mutants of Lp-HsfB1 1 B1.25M2 ´ A2.7 357 )) ) ) ) 2 B1.25M4 ´ A2.7 357 )) ) ) ) 3 B1.25M7 ´ A2.7 357 )) ) ) ) 4 B1.25M11 ´ A2.7 357 + 35 335 2.0 + a Survival (+) or nonsurvival ()) on 5-¯uoro orotic acid plates at 28° C b Temperature limit of growth was determined by growing cells on YPD plates at di€erent temperatures (see Fig. 4) 325 c b-Galactosidase activity was determined in heat-stressed liquid cultures using strains transformed with the pHSE2-137 reporter plasmid. Inducibility indicates the fold increase of activity after heat stress (see Materials and methods) d Transactivator potential measured in a transient expression assay in tobacco protoplasts (data from Treuter et al. (1993) and unpublished) 326 with a strain disrupted in the chromosomal HSF gene by generation of all other strains (no. I.1 in Table 2). All insertion of an HIS3 gene (Wiederrecht et al. 1988). The strains were able to grow in YPD medium at 28° Cata viability of this strain is maintained by a URA3-coding normal rate, i.e., with a doubling time of 2±2.5 h. plasmid harboring the yeast HSF1 gene. For replace- Group I of Table 2 represents the results for strains ment of the latter, the heterologous HSF expression expressing the yeast, Drosophila and three tomato Hsfs. cassette is introduced on a LEU2-coding plasmid before With the exception of the tomato HsfB1, the other three eliminating the URA3 plasmid by selection on FOA- heterologous Hsfs can substitute for the yeast Hsf. The containing medium. The Hsf types and their derivatives lack of function of HsfB1 is clearly due to peculiarities coded by the di€erent LEU2 plasmids used in this paper of the activator domain (see below). The band-shift as- as well as the results of FOA selection are summarized in say (Fig. 2B) demonstrates that all Hsfs assigned to Table 2. group I give well-de®ned bands with the two oligonu- Strains in Table 2 are organized into ®ve groups (I± cleotides tested. The only exception is the yeast Hsf, V) based on the Hsf type. Except for the yeast Hsf which has a very low binding anity for the HSE3 (no. 1.1) and the Drosophila Hsf (no. 1.2), expression of oligonucleotide, which contains only three copies of the the corresponding Hsf was veri®ed by Western blotting (AGAAn) consensus motif. before and after FOA treatment. Results are presented Groups II and III summarize data with C-terminal in Fig. 2 for those strains surviving FOA selection. The and internal deletions of Lp-Hsfs A1 and A2, respec- proper size of the protein detected by the antisera can be tively. Only two Hsfs are defective (nos. II.4 and III.4). correlated with the size of the Hsf indicated by the block They represent truncated proteins with C-terminal de- diagrams in Table 2. The Western assay before FOA letions of all parts essential for the activator functions as treatment veri®es that nonsurvival is not a result of identi®ed by Treuter et al. (1993) using tests in tobacco nonexpression of the Hsf but of its lack of function. protoplasts. An interesting minimal version of a func- Examples are given for constructs II.4, III.1 and III.4 tional Hsf in yeast is HsfA2D7/8, C323 (no. III.5), (Fig. 2). As a control, we always included a strain with lacking the oligomerization domain (HR-A/B) and part the yeast HSF1 gene on the same 2l plasmid used for of the C-terminal activator region (HR-C). Its special properties will be discussed below. To investigate the reasons for the lack of function of the tomato HsfB1 (no. I.5), we created constructs coding for chimeric proteins with di€erent combinations of the functional domains of HsfB1 and HsfA2. The results are compiled in group IV of Table 2. Like HsfB1 itself, two fusion proteins containing the DNA-binding domain of HsfA2 and the C-terminus of B1 cannot substitute for the yeast Hsf (nos. IV.1 and 2). In contrast, all fusion proteins with C-terminal parts of HsfA2 are active. The extreme form of such a functional fusion Hsf (no. IV.6) contains most of HsfB1 plus a 34 amino acid residue motif from the C-terminus of HsfA2 with the AHA1 element (see Discussion). Interestingly, this short peptide motif from the activator region of HsfA2 can also be replaced by a similar motif from the core of the yeast Gal4 activator (data not shown).

A functional DNA-binding domain is required for survival of yeast cells

To extend the analysis of the functional requirements of heterologous Hsfs, we introduced mutations in the DNA-binding domain. Since the mutants were created in an HsfB1 background, we used a fusion protein of HsfB1 with the C-terminal activator of HsfA2 for the Fig. 2A, B Expression of heterologous Hsfs in yeast. Native protein extracts of the indicated recombinant yeast strains were processed for tests in yeast. Three of the mutants selected (M2, M4, Western blotting (A) and for band-shift assays (B) Hsfs were detected M7) contain amino acid exchanges at highly conserved with antisera a against Hsfs A1 and A2 as indicated. 32P-labeled HSE3 or invariant positions (Fig. 3A). They were found to be and HSE6 oligonucleotides were used for band-shift assays (see inactive as transcription activators in the protoplast as- Materials and methods). Extracts marked by ()) were prepared from say, and they are de®cient in DNA binding if expressed cells before FOA selection to demonstrate that the nonsurvival of strains with constructs 1.5, II.4 and III.4 is not due to nonexpression in yeast. Figure 3 presents the relevant sequence data or extreme instability of the truncated Hsfs together with the Western analyses and the band-shift 327 analysis to con®rm expression of the fusion proteins Temperature sensitivity of growth before FOA treatment. None of the three mutants was able to support growth in the absence of the yeast Hsf. To test heterologous Hsfs for their ability to support As a control we included mutant M11 with a TT ® KS growth under conditions of moderate heat stress, the exchange between b1 and b2 of the DNA-binding do- corresponding yeast strains were plated on YPD and main. It is active in DNA binding (Fig. 3C) and also incubated at the indicated temperatures. Examples for behaves like the wild-type fusion protein with respect to eight di€erent strains are shown in Fig. 4; the complete survival and growth at elevated temperatures (see information is summarized in Table 2. Clear di€erenti- Table 2, no. V.4). These results make it very likely that ation with respect to temperature sensitivity is apparent. the growth-maintaining function of Hsfs in yeast in- Strains with yeast or Drosophila Hsf grow well up to volves HSE-speci®c binding to hitherto unidenti®ed 37° C. HsfA2 and its variants, including fusion proteins genes whose expression is indispensable for growth. This with the N-terminal part of HsfB1, sustain growth up to is also supported by our observation that fusion proteins 35° C, whereas strains with tomato HsfA1 are more of the C-terminal activator region of HsfA2 with the sensitive with a limit of growth between 33° and 35° C. Gal4 DNA-binding domain are not able to replace the This peculiarity of HsfA1 is also obvious when tem- yeast Hsf for growth maintenance (data not shown). perature shift experiments with liquid cultures are done (data not shown). In contrast to other strains, there is a slowdown or cessation of growth about 4 h after shift to 35° Cor37°C, and even 3 h after shiftdown to 28° C there is no detectable cell division. Western blot analyses of HsfA1 levels showed that there is no detectable de- gradation of the protein within this time. Furthermore, the functions required for reporter gene expression (Fig. 6) and for the expression of thermotolerance are not impaired (Fig. 5), in spite of the fact that in both cases the induction temperature is 37° C.

Induced thermotolerance of yeast strains with heterologous Hsfs

An important criterion for the proper function of Hsfs is the induction of thermotolerance after conditioning by a moderate heat stress treatment (Lindquist and Kim 1996). We tested the survival of cells at 50.5° C with and without a preconditioning stress treatment for 30 min at 37° C (Fig. 5). All strains responded similarly to the pretreatment with an increase of thermotolerance by two orders of magnitude measured as percentage survival after a 20 min challenge at 50.5° C. Interestingly, the intrinsic thermoresistance of the strains is very di€erent (Fig. 5, dashed lines), depending on the Hsf present. It is relatively high for strains with either of the tomato Hsfs, intermediate with the yeast Hsf, and relatively low for the strain with the Drosophila Hsf.

Fig. 3A±C Characterization of DNA-binding domain mutants. Mutants M2, M4, M7 and M11 were created in an HsfB1 background Test for activator function with Hsf-dependent and tested in yeast as HsfB1.25 ´ A2.7 fusion proteins (see group V in Table 2) because the fusion Hsf with the wild-type DNA-binding reporter constructs domain can substitute for the yeast Hsf. The block diagram in A illustrates the Hsf DNA-binding domain with the conserved The transcription activator function of the heterologous secondary structure elements. Invariant and highly conserved amino Hsfs was tested by use of two di€erent lacZ reporter acid residues are given in bold letters. The putative DNA recognition helix is a3 of the HTH motif. The amino acid exchanges in the constructs based on an inactive CYC1 deletion promoter mutants are indicated. B presents the expression data for the wild-type in which the UAS is replaced by either of two types of and mutant proteins using Western blots of protein extracts before HSE. As indicated in Fig. 6, they contain three (reporter FOA selection (HsfA2 antiserum). C shows band-shift assays with A) and six modules (reporter B), respectively, of the these protein extracts using HSE3 (left)orHSE6(right)aslabeled oligonucleotide. In agreement with the data for FOA survival, only AGAA or TTCT consensus motifs. The data obtained the wild-type Hsf and the mutant M11 are able to bind to the HSE with reporter A are included in Table 2. In Fig. 6 we oligonucleotide present several examples of the di€erential response with 328

Fig. 4 Temperature sensitivity of growth. The indicated strains were grown in YPD medium to an OD at 600 nm of 0.2±0.3. Cells were serially di- luted (tenfold at each step), and aliquots of 4 ll spotted onto agar plates and incubated at the indicated temperatures for 48 h. The photographs show typical examples for selected strains. The corresponding data for all strains are summarized in Table 2

the two reporter constructs. Interestingly, the homolo- striking for the fusion protein HsfB1.25 ´ A2.7 (con- gous Hsf is almost four times more active with reporter struct IV.3). A than with reporter B. The opposite is true for the The reporter assay indicates yet another interesting tomato HsfA2 and the fusion proteins between Hsfs B1 point. The tomato HsfA2D7/8,C323 (no. III.5 in and A2. This preference for reporter B is particularly Table 2) is virtually inactive with reporter A but active with B. As shown in the block diagram in Fig. 1 and Table 2, the internal deletion (D7/8) removes the whole oligomerization domain (HR-A/B). Using EGS cross- linking (Fig. 7A), we demonstrated that this truncated protein is unable to form oligomers. Moreover, in a band-shift assay (Fig. 7B), it exhibits reduced binding anity for an oligonucleotide with only three HSE modules (similar to reporter A) as compared with one with six modules, corresponding to reporter B. We can conclude from this that oligomerization is not a pre- requisite for DNA binding, but it improves the binding anity for DNA, which might be important for the expression of certain heat stress-regulated genes. How- ever, we did not observe any e€ect of this deletion on the ability to support growth of this Hsf (Fig. 4), nor on the level of thermotolerance induced by a preconditioning heat stress at 37° C (Fig. 5).

Discussion

Disruptions of the single yeast HSF1 gene were origi- nally generated by Sorger and Pelham (1988) and Wi- ederrecht et al. (1988). We used a comparable strain to test the ability of tomato Hsfs and various mutant forms derived from them to replace the function of the yeast Hsf. An interesting, and for genetic screens very useful, peculiarity of yeast is the fact that the Hsf is essential for viability and growth not only under heat stress but also under control conditions. One key to understanding the dual role of Hsf may be the necessity to express chap- erone-coding genes under stress and non-stress condi- Fig. 5 Induced thermotolerance of yeast cells expressing heterologous tions. In both cases Hsf is bound to its cognate promoter Hsfs. Yeast cells with the indicated Hsfs were grown in YPD medium element (HSE), but its contacts with the basal tran- to an OD at 600 nm of 0.4. Cultures were kept at 28° C(dashed scription apparatus are modi®ed by heat stress activa- curves)orshiftedto37°Cfor30min(full curves) before incubation at 50.5° C for the indicated times. Afterwards, these cultures were tion. This model takes into account the observation that, diluted, plated on YPD agar, and colonies were counted after 48 h in contrast to other organisms, the yeast Hsf is always incubation at 28° C. Each value represents the mean of four plates bound to DNA (Sorger and Pelham 1987). Alternatively 329 or in addition, Hsf might be required for expression of Other peculiarities of the yeast system may be equally hitherto unidenti®ed housekeeping or cell cycle-speci®c helpful in elaborating details of Hsf function. The failure genes. At any rate, the data in this paper and the mu- of tomato HsfB1 to substitute for the yeast factor unless tation analysis reported by Hubl et al. (1994) demon- tagged with a short C-terminal part from the activator strate that an intact DNA-binding domain of Hsf is region of HsfA2 de®nes the minimum activator and indispensable for all functions, including survival at supports the concept of a modular architecture of these 28° C. proteins. This hybrid protein and the unmodi®ed HsfB1 However, two interesting reports are worthy of are valuable tools with which to screen for interacting mention because they modify and complement this proteins in yeast using a tomato cDNA expression seemingly simple picture. (i) Smith and Ya€e (1991) library. described a mutant Hsf of yeast with a nonsense codon The minimum activator contains a short peptide in the DNA-binding domain. Survival of the corre- motif with a central Trp residue (±DDIWEELL). It was sponding yeast strain is only possible in a PSI+ sup- found earlier (Treuter et al. 1993) to be essential for the pressor background (Lindquist and Kim 1996). This Hsf function of HsfA2 as a transcriptional activator (Trp with a missense mutation in the DNA-binding domain is element). But similar motifs with aromatic (W, F), large sucient to support growth, but it is unable to function hydrophobic (L, I, M, V) and acidic (D, E) amino acid in heat stress-induced Hsp synthesis. (ii) A yeast mutant residues have also been de®ned as functionally important with deletions in the SSA1 and SSA2 genes, coding for parts of other transcription activators, e.g., yeast Hsf, two of the four heat stress-inducible Hsp70 proteins, was Gal4, and Gcn4, or vertebrate Hsf1, p53, VP16, Fos, Jun unable to grow at 37° C. This e€ect was combined with RelA, Sp1, C/EBPa, and E2A. We have summarized an unbalanced overexpression of other Hsps (Boorstein them under the more general term AHA elements. In and Craig 1990). A search for a suppressor of this mu- several cases, they have been shown to form interaction tant phenotype led to an unexpected result. The revert- sites with components of the basal transcription complex ant had an altered HSF1 gene, coding for Hsf with an (for a summary see Nover and Scharf 1997). amino acid exchange in helix 2. The replacement of the Despite many remarkable similarities the details of invariant P residue by Q leads to a strongly reduced control of Hsf activity may vary between di€erent or- DNA-binding anity and normalization of the growth ganisms. Experiments with vertebrate and Drosophila phenotype. Evidently, an Hsf with a marked de®ciency in DNA binding as detected by band-shift assays may be sucient for growth even at higher temperatures (Halladay and Craig 1995). Taken together, our functional analyses of the Drosophila and tomato Hsfs in yeast faithfully re¯ect the data obtained with the tobacco protoplast assay (Treu- ter et al. 1993). Other examples of successful Hsf transfer into heterologous systems have been reported, e.g., ex- pression of the human and Drosophila Hsfs in tobacco protoplasts (Treuter et al. 1993), of the human Hsf1 in Drosophila cells (Clos et al. 1993) and Xenopus oocytes (Baler et al. 1993), of the Drosophila Hsf in human cells (Clos et al. 1993) and Schizosaccharomyces pombe (Gallo et al. 1993) and of the Arabidopsis HsfA1 in Drosophila and human cells (HuÈ bel et al. 1995). In many cases the regulated phenotype, i.e., heat stress-induc- ibility, is lost or markedly impaired. A fortuitous ex- ception is human Hsf1. In all three cases mentioned above, its regulation is maintained even in heterologous systems (Drosophila cells, Xenopus oocytes, tobacco protoplasts). But the threshold temperature of induction of 35°±37° C corresponds to the expression system and not to the system of origin (in mammals 42°±45° C), i.e., the Hsf itself is not the thermometer. The low or absent activity control in yeast strains expressing tomato Hsfs is particularly striking with Fig. 6 Hsf activation of the expression of two di€erent reporter HsfA2 and the constructs derived from it (Table 2, constructs. The two lacZ reporter constructs contain HSE inserts group III and Fig. 6). This system may be useful for derived from the yeast SSA1 gene (pHSE2-137), or a synthetic oligonucleotide (HSE2-BG). As indicated above (reporter A), the screening for tomato factors (coregulators) required to former contains three binding modules for Hsf, whereas reporter B restore proper regulation. To this end, we are currently harbors three additional modules. The Hsfs replacing the endogenous testing chaperones of the Hsp70 and Hsp90 families. yeast factor are indicated below the histograms 330

Fig. 7A, B Role of the HR-A/B region in oligomerization and DNA- former two Hsfs, containing the HR-A/B region. There is no binding functions of HsfA2. For cross-linking experiments (A) native detectable di€erence in the pattern of cross-linking products whether protein extracts of the indicated strains (see block diagrams in the reaction was performed in the presence (+) or absence ())ofthe Table 2, nos. III.1, 2 and 5) were treated with 750 lM ethyleneglycol- HSE6 oligonucleotide. The band-shift assay (B) demonstrates that the bis (sulfosuccinimidylsuccinate) (EGS) and after precipitation sepa- loss of the oligomerization domain in HsfA2D7/8,C323 reduces its rated by SDS-polyacrylamide gel electrophoresis (see Materials and anity for DNA-binding. No complex is detectable with HSE3, and methods). The Western blot shows the monomers with the expected only the fast-migrating complex C1 is found with the HSE6 Mr of 56 (HsfA2), 50 (HsfA2DC323) and 42 kDa (HsfA2D7/8,C323). oligonucleotide The cross-linked trimeric forms (asterisks) can be detected only for the

Hsfs indicate that heat stress-induced activation involves Acknowledgements This work was supported by grants to K.-D. a monomer to trimer transition tightly linked to nuclear Scharf from the Deutsche Forschungsgemeinschaft (Scha 577/5-1), from the Bundesministerium fuÈ r Bildung, Wissenschaft, Forschung import (Westwood et al. 1991; Baler et al. 1993; und Technologie (FKZ 0310253A) and from the Fonds der Rabindran et al. 1993; Sarge et al. 1993; Zuo et al. 1994, Chemischen Industrie. We gratefully acknowledge gifts of strains 1995). In keeping with this, DNA binding is only ob- and plasmids from M. Vidal (Northwestern University, Evanston), served after stress activation. However, in contrast to the E. Craig (University of Wisconsin, Madison). C. S. Parker (Cali- fornia Institute of Technology, Pasadena), and P. Sorger (Univer- other functional parts (DNA-binding domain, NLS, sity of California, San Francisco). The technical assistance of G. C-terminal activator), the oligomerization domain HR- Englich (Frankfurt) and H. Nixdorf and C. RuÈ lke (Halle) is greatly A/B seems to be dispensable, at least in plant cells and in appreciated. We are grateful for advice and critical discussions with yeast. The corresponding deletion form of HsfA2 L. Nover (Goethe University, Frankfurt) during the preparation of (HsfA2D7/8,C323, see no. II.5 in Table 2) is unable to the manuscript. form oligomers if EGS cross-linked (Fig. 7A), yet it can replace the yeast Hsf in all functions tested by us. The only exception is the reporter assay (Fig. 6). Transcrip- References tion is only observed with the complex reporter con- struct (B), containing six HSE modules. This Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, observation is supported by the DNA-binding assay Smith JA, Struhl K (eds) (1993) Current protocols in molecular (Fig. 7B). The results demonstrate that the multiplicity biology. J. Wiley & Sons of HSE modules usually found in many heat stress Baler R, Dahl G, Voellmy R (1993) Activation of human heat promoters might also be sucient to align monomeric shock-genes is accompanied by oligomerisation, modi®cation, and rapid translocation of heat shock transcription factor. Mol Hsf in a quasi-multimeric form, which is probably re- Cell Biol 13:2486±2496 quired for ecient transcription activation (high pro- Ballester R, Michaeli T, Ferguson K, Xu HP, McCormick F, moter occupancy). One major e€ect of oligomerization Wigler M (1989) Genetic analysis of mammalian GAP expres- is an improved binding to low-anity promoters. To sed in yeast. Cell 59:681±686 Bonner JJ, Heyward S, Fackenthal DL (1992) Temperature-de- support this idea, it will be interesting to investigate the pendent regulation of a heterologous transcriptional activation expression of the set of yeast endogenous heat stress domain fused to yeast heat shock transcription factor. Mol Cell genes in their response to the wild-type and deletion Biol 12:1021±1030 forms of HsfA2. This type of analysis may also yield Boorstein WR, Craig EA (1990) Transcriptional regulation of SSA3, an HSP70 gene from Saccharomyces cerevisiae. Mol Cell clues to the basis of the intriguing di€erences in intrinsic Biol 10:3262±3267 heat resistance (Fig. 5) and the upper temperature limit Buchner J (1996) Supervising the fold-functional principles of of growth (Fig. 4), respectively. molecular chaperones [Review]. FASEB J 10:10±19 331

Clos J, Westwood JT, Becker PB, Wilson S, Lambert U, Wu C Sanger R, Nicklen S, Coulson A (1977) DNA sequencing with (1990) Molecular cloning and expression of a hexameric Dros- chain-terminating inhibitors. Proc Natl Acad Sci USA 74:5463± ophila heat shock factor subject to negative regulation. Cell 5467 63:1085±1097 Sarge KD, Murphy SP, Morimoto RI (1993) Activation of heat Clos J, Rabindran S, Wisniewski J, Wu C (1993) Induction tem- shock gene transcription by heat shock factor 1 involves perature of human heat shock factor is reprogrammed in a oligomerization, acquisition of DNA-binding activity, and nu- Drosophila cell environment. Nature 364:252±255 clear localization and can occur in the absence of stress. Mol Colicelli J, Birchmeier C, Michaeli T, O'Neill K, Riggs M, Wigler Cell Biol 13:1392±1407 M (1989) Isolation and characterization of a mammalian gene Scharf KD, Rose S, Zott W, Schoe‚ F, Nover L (1990) Three encoding a high-anity cAMP phosphodiesterase. Proc Natl tomato genes code for heat stress transcription factors with a Acad Sci USA 86:3599±3603 region of remarkable homology to the DNA-binding domain of Czarnecka-Verner E, Yuan CX, Fox PC, Gurley WB (1995) Iso- the yeast HSF. EMBO J 9:4495±4501 lation and characterization of six heat shock transcription fac- Scharf KD, Rose S, Thierfelder J, Nover L (1993) Two cDNAs for tor cDNA clones from soybean. Plant Mol Biol 29:37±51 tomato heat stress transcription factors. Plant Physiol Ellwood MS, Craig EA (1984) Di€erential regulation of the 70K 102:1355±1356 heat shock gene and related genes in Saccharomyces cerevisiae. Scharf KD, Materna T, Treuter E, Nover L (1994) Heat stress Mol Cell Biol 4:1454±1459 promoters and transcription factors. In: Nover L (ed) Plant Gagliardi D, Breton C, Chaboud A, Vergne P, Dumas C (1995) promoters and transcription factors. Springer-Verlag Berlin, pp Expression of heat shock factor and heat shock protein 70 genes 121±158 during maize pollen development. Plant Mol Biol 29:841±856 Schultheiss J, Kunert O, Gase U, Scharf K-D, Nover L, RuÈ terjans Gallo GJ, Prentice H, Kingston RE (1993) Heat shock factor is H (1996) Solution structure of the DNA-binding domain of the required for growth at normal temperatures in the ®ssion yeast tomato heat stress transcription factor HSF24. Eur J Biochem Schizosaccharomyces pombe. Mol Cell Biol 13:749±761 236:911±921 Halladay JT, Craig EA (1995) A heat-shock transcription factor Slater MR, Craig EA (1987) Transcriptional regulation of an hsp70 with reduced activity suppresses a yeast HSP70 mutant. Mol heat shock gene in the yeast Saccharomyces cerevisiae. Mol Cell Cell Biol 15:4890±4897 Biol 7:1906±1916 Harrison CJ, Bohm AA, Nelson HCM (1994) Crystal structure of Smith BJ, Ya€e MP (1991) A mutation in the yeast heat-shock the DNA binding domain of the heat shock transcription fac- factor gene causes temperature-sensitive defects in both mito- tor. Science 263:224±227 chondrial protein import and the cell cycle. Mol Cell Biol Hartl FU (1996) Molecular chaperones in cellular protein folding. 11:2647±2655 Nature 381:571±580 Sorger PK (1990) Yeast heat shock factor contains separable Hubl ST, Owens JC, Nelson HCM (1994) Mutational analysis of transient and sustained response transcriptional activators. Cell the DNA-binding domain of yeast heat shock transcription 62:783±805 factor. Nat Struct Biol 1:615±620 Sorger PK, Pelham HRB (1987) Puri®cation and characterization HuÈ bel A, SchoÈ ‚ F (1994) Arabidopsis heat shock factor: isolation of a heat-shock element binding protein from yeast. EMBO J and characterization of the gene and the recombinant protein. 6:3035±3042 Plant Mol Biol 26:353±362 Sorger PK, Pelham HRB (1988) Yeast heat shock factor is an HuÈ bel A, Lee JH, Wu C, SchoÈ ‚ F (1995) Arabidopsis heat shock essential DNA-binding protein that exhibits temperature-de- factor is constitutively active in Drosophila and human cells. pendent phosphorylation. Cell 54:855±864 Mol Gen Genet 248:136±141 Treuter E, Nover L, Ohme K, Scharf K-D (1993) Promoter spe- Kunkel TA (1985) Rapid and ecient site-speci®c mutagenesis ci®city and deletion analysis of three heat stress transcription without phenotypic selection. Proc Natl Acad Sci USA 82:488± factors of tomato. Mol Gen Genet 240:113±125 492 Vuister GW, Kim SJ, Wu C, Bax A (1994) NMR evidence for Lindquist S, Kim G (1996) Heat-shock protein 104 expression is similarities between the DNA-binding regions of Drosophila sucient for thermotolerance in yeast. Proc Natl Acad Sci USA melanogaster heat shock factor and the helix-turn-helix and 93:5301±5306 HNF-3/forkhead families of transcription factors. Biochemistry Lyck R, Harmening U, HoÈ hfeld I, Treuter E, Scharf KD, Nover L 33:10±16 (1997) Intracellular distribution and identi®cation of the nu- Waters ER, Lee GJ, Vierling E (1996) Evolution structure and clear localization signals of two plant heat-stress transcription function of the small heat shock proteins in plants. J Exp Bot factors. Planta 202:117±125 47:325±338 Nover L (1987) Expression of heat shock genes in homologous and Westwood JT, Clos J, Wu C (1991) Stress-induced oligomerization heterologous systems. Enzyme Microb Technol 9:130±144 and chromosomal relocalization of heat shock factor. Nature Nover L (1991) Heat shock response. CRC Press, Boca Raton 353:822±827 Nover L, Scharf KD (1997) Heat stress proteins and transcription Wiederrecht G, Seto D, Parker CS (1988) Isolation of the genes factors. Cell Mol Life Sci: in press encoding the S. cerevisiae heat shock transcription factor. Cell Nover L, Scharf KD, Gagliardi D, Vergne P, Czarnecka-Verner E, 54:841±853 Gurley WB (1996) The Hsf world: classi®cation and properties Wu C (1995) Heat stress transcription factors. Annu Rev Cell Biol of plant heat stress transcription factors. Cell stress. in press 11:441±469 Pelham HRB, Bienz M (1982) A synthetic heat-shock promoter Zuo J, Baler R, Dahl G, Voellmy R (1994) Activation of the DNA- element confers heat-inducibility on the herpes simplex virus binding ability of human heat shock transcription factor 1 may thymidine kinase gene. EMBO J 1:1473±1477 involve the transition from an intramolecular to an intermo- Rabindran SK, Haroun RI, Clos J, Wisniewski J, Wu C (1993) lecular triple-stranded coiled coil structure. Mol Cell Biol Regulation of heat shock factor trimerization: role of a co- 14:7557±7568 served leucine zipper. Science 259:230±234 Zuo J, Rungger D, Voellmy R (1995) Multiple layers of regulation Rose MD, Winston F, Hieter P (1990) Methods in yeast genetics. of human heat shock transcription factor 1. Mol Cell Biol Cold Spring Harbor Laboratory. Press, Cold Spring Harbor, 15:4319±4330 NY Sambrook J, Fritsch EF, Maniatis T (1989) Molecular cloning, a laboratory manual, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY