Search International and National Patent Collections
Some content of this application is unavailable at the moment.
If this situation persists, please contact us atFeedback&Contact
1. (WO1992009617) CELL STRESS TRANSCRIPTIONAL FACTORS
Note: Text based on automatic Optical Character Recognition processes. Please use the PDF version for legal matters
I
CELL STRESS TRANSC IPTIONAL FACTORS
Background of the Invention
Field of the Invention
The present invention relates to novel heat shock transcriptional activators or heat shock factors (HSF) and to polynucleotides encoding such factors.
The present invention pertains in particular to
Drosophila and human heat shock factors,
polynucleotides encoding the same, and antibodies made to natural and recombinant HSFs.
Background Information
All organisms respond to elevated
environmental temperatures and a variety of
environmental stresses by rapidly activating the expression of a group of proteins referred to as heat shock or stress proteins. Although the functions of heat shock proteins have remained obscure for many years since the discovery of the phenomenon by Ritossa (1962), Experientia, 18, 571-573, recent studies suggest a central role for heat shock-induced proteins and their constitutive counterparts in mediating protein-protein interactions, protein folding and the transport of proteins across membranes (reviewed by Mori oto et al, 1990, Stress proteins in biology and medicine. Cold Spring Harbor Laboratory Press. 1-36) . The synthesis of heat shock proteins is subject to both transcription and post-transcriptional control in eukaryotic cells (reviewed by Craig, 1985, Crit. Rev. Biochem. , 18, 239-280; Lindquist, 1986, Aim. Rev. Biochem. , 55, 1151-1191). Heat shock-inducible transcription is mediated by a positive control element, the heat shock element (HSE) , defined as three repeats of a 5-nucleotide [_GAA_] module, arranged in alternating orientation (Pelham, 1982, Cell, 30, 517-528; Amin et al, 1988, Hoi. Cell. Biol. 8, 3761-3769; Xiao and Lis, 1988, Science, 239, 1139-1142) . Multiple copies of the HSE are found upstream of all eukaryotes heat shock genes.
A heat shock activator protein or heat shock transcription factor now generally termed heat shock factor (HSF) , binds to HSEs and activates transcription of heat shock genes. (Wu, 1984a, Nature, 309, 229-234, 1984b; Parker and Topol, 1984, Cell, 36, 357-369; Topol et al, 1985, Cell, 42, 527-537). Although the sequence of the HSE has been highly conserved in evolution, HSF purified from yeast, Drosophila, and human cells differ in molecular size (150 kD, 110 kD and 83 kD,
respectively; Sorger and Pelham, 1987, EMBO J, 6, 3035-3041; Wu et al, 1987, Science, 238, 1247-1253;
Goldenberg et al, 1988 J. Biol . Chem. 263., 19734-19739). Yeast and higher eukaryotes also differ in the
regulation of HSF activity. In yeast, HSF bound constitutively to the HSE apparently stimulates
transcription when phosphorylated under heat shock conditions. In Drosophila and vertebrate cells, HSF is unable to bind to the HSE unless the cells are heat shocked (for a review, see Wu et al, 1990, In: Stress proteins in biology and medicine. Cold Spring Harbor

Laboratory Press. 429-442) . The heat-inducible binding of HSF appears to be a major regulatory step in the pathway to heat shock gene activation in higher
eukaryotes.
The induction and reversal of HSF binding activity in vivo does not require new protein synthesis (Zimarino and Wu, 1987, Nature, 327, 727-730 ; Kingston et al, 1987, hoi . Cell . Biol . , 7, 1530-1534; Zamarino et al, 1990a). In addition, HSF extracted from nonshocked cell cytosol can be activated in vitro by heat (Larson et al, 1988, Nature, 335, 372-375), low pH (Mosser et al, 1990, Proc. Natl . Acad. Sci USA, 87, 3748-3752), and by interaction with antibodies raised to the active form HSF (Zimarino et al, 1990b, Science, 249, 546-549).
These results suggest that the pre-existent, inactive form of HSF present in the cell cytosol can assume the active conformation without an enzymatic modification of protein structure, translocate to cell nucleus and activate heat shock protein.
The present invention relates to the structure and function of novel human and Drosophila heat shock factor (HSF) proteins and the DNA sequences that encode these proteins. In particular, the present invention relates to Drosophila and human HSF proteins produced in E. coli and Xenopus oocytes and their specific DNA binding activities with and without heat shock
induction.
Brief Description of the Drawinσs
Figure 1 shows the purification of Drosophila HSF and microsequencing of HSF peptides. (A) SDS gel electrophoresis and silver staining of a HSF
preparation from Drosophila Schneider Line 2 cells. 5% of the purified HSF fraction was electrophoresed on a 10% polyacrylamide gel. (B) A ino acid sequences of six peptides obtained by tryptic digestion of purified HSF. Oligo 27 and oligo 29 are two degenerate
sequences (boxed) , deduced from the amino acids under which they are aligned. The sequences represent the coding strand. Valine, leucine, and isoleucine codons of oligo 27 were chosen in accordance with the codon bias of Drosophila. All other codons are fully degenerate.
Figure 2 represents the cloning and sequence analysis of Drosophila HSF. (A) Schematic
representation of seven HSF cDNA clones aligned with reference to the reconstructed full-length HSF cDNA clone. HSF 302, 307, and 312 were isolated from an oligo dT-primed cDNA library, and HSF 407, 409, 410, and 412 were isolated from a random-primed cDNA
library. The open bar represents the 2073 nt HSF open reading frame. (B) Nucleotide sequence of the HSF cDNA and predicted amino acid sequence. The entire DNA sequence presented has been sequenced at least twice, from overlapping cDNA clones. Start and stop codons, and a polyadenylation signal are highlighted by reverse print. Two single restriction sites (StuI, and Apal) that were used for generation of 3 ' deletion mutants are noted. Sequences in the open reading frame that match the sequences of the six HSF tryptic peptides listed in Figure IB are boxed. (C) In situ
hybridization of digoxigenin-substituted HSF DNA
(coding sequences) to Drosophila salivary gland
polytene chromosomes. The cytologicalocus of
hybridization (55A) , is indicated by the arrow.
Figure 3 shows the DNA-binding activity of
reco binant Drosophila HSF. (A) Gel mobility shift analysis of natural and recombinant HSF. Cytoplasmic extracts from unshocked SL-2 cells (lanes 1-3) ,and HSF translated in vitro at 25βC or 30βC (lanes 4-9) were subjected to in vitro heat shock for 10 min at 34 °C or kept at 0βC. Prior to gel shift analysis, samples in lanes 3, 6, and 9 were incubated at room temperature with a 1:60 dilution of polyclonal serum raised against the national HSF protein purified from Drosophila cells (Zimarinoet al., 1990 Science 249, 546-549). Identical translations of anti-sense HSF RNA showed no DNA binding activity. (B) Gel mobility shift assay of HSF translated in vitro at 30βC, in the absence of
competitor DNA (lane 1) , with a 40-fold excess of unlabeled HSE (lane 2) or a similar excess of synthetic DNA from the hsp70 gene,positions +40 to +80 (lane 3) . (C) DNase I protection analysis. Recombinant HSF extracted from E. coli was incubated with 5" 32P-labeled hsp70 promoter DNA, digested with DNase I, and analyzed by electrophoresis on a 8% sequencing gel (left panel; non-coding strand) or 6% (right panel; coding strand) sequencing gel. Amounts of HSF used for each reaction are indicated. The total protein concentration in all samples was normalized by the addition of extracts of bacteria transformed with the expression vector alone. The lanes marked A,C,G,T are dideoxy-sequencing reactions. (D) Nucleotide sequence of the hsp70 promoter from position -185 to +10. The sequences in lower case are from the plasmid vector. Three upstream HSEs and the TATA sequence are boxes. The start site and direction of transcription are indicated. Brackets indicate sequences protected by the recombinant HSF. There is a clear DNase I footprint on both strands over the two proximal HSEs, and some protection on the coding strand also occurs on the third HSE (position -174 to -186) , the non-coding strand of which was not analyzed.

Figure 4 shows the transcriptional stimulation by recombinant Drosophila HSF in vitro. Primer extension analysis of RNA synthesized by nonshocked Drosophila embryo transcription extracts supplemented with 0.2 μl of E. coli extract from HSF expressing cells (+) , or with extract from cells transformed with the expression vector only (-) . As an internal control for
transcription from the template carrying two HSEs, the same template deleted of the HSEs (as well as a 30 bp downstream region) was mixed in the reaction. RNA originating from the template lacking HSEs is thus distinguished by a 30 nt decrease in size. As a further control for RNA recovery, a defined amount of RNA synthesized from a T7 promoter upstream of the hsp70 sequences inserted into pBluescript was
introduced into each transcription reaction along with the stop solution. Schematic drawings of the two templates are aligned with the primer extension
products of the respective transcripts.
Figure 5 demonstrates the DNA-binding activity of HSF expressed in Xenopus oocytes. Gel mobility shift assay of HSF extracted from individual oocytes.
Extracts of each of five nonshocked (18βC) oocytes
(lanes 1-5) , and five heat shocked (36βC, 10 min) oocytes (lanes 6-10) were individually analyzed. The positions of the HSF-HSE complex and free HSE are indicated.
Figure 6 shows the estimation of the native size of cloned Drosophila HSF. (A) Pore exclusion limit
analysis of HSF. Purified, cloned HSF (5μg/12 μl sample volume) was electrophoresed on a nondenaturing 4-20% polyacrylamide gel until the limit of migration was reached. The gel was stained with Coomassie Blue. The marker lane shows molecular weight markers:
thyroglobulin tetramer (1338 kD) , thyroglobulin dimer (669 kD) , apoferritin (440 kD) , catalase (232 kD) , lactate dehydrogenase (140 kD) and bovine serum albumin (67 kD) . (B) Pore exclusion limit analysis of the HSF:32P-HSE complex. 3 μl of Drosophila SL-2 cell cytosol (lanes 1,2) and 0.5 μl of an extract E. coli expressing HSF (lanes 3,4) were heat shocked (+) in vitro at 34 βC or incubated at O'C (-) for 10 min. The samples were incubated for 10 min with 32 P-labeled HSE under standard gel shift conditions, and
electrophoresed on a nondenaturing, 3-12%
polyacrylamide gradient gel until the limit of
migration. The gel was stained with Coomassie Blue, dried and subjected to autoradiography. The positions of marker proteins are indicated. (C) Glutaraldehyde cross-linking of cloned HSF. Purified HSF (2 μg/10 μl) was treated for 5 min at room temperature with
glutaraldehyde as indicated. After quenching, about 1 μg of cross-linked HSF was separated on a 4-6% SDS polyacrylamide gel, and silver stained. The minor polypeptides below the 105 kD HSF protein probably represent degradation products. The marker lane contain cross-linked phosphorylated b (Sigma) ; cross-linked thyroglobulin was also used as a marker. EGS cross-linking of cloned HSF. Lanes 1-5: purified HSF (2 μg/10 μl) was treated for 10 min at room temperature with EGS as indicated. Lanes 6,7: similar EGS
treatment of HSF diluted to 2 μg/ml. The cross-linked products were precipitated with 15% TCA, washed twice with ice-cold acetone, and dissolved in Laemmli sample buffer. Cross-linked products were analyzed by SDS gel electrophoresis as above. Introduction of ovalbumin into the cross-linking reaction revealed no interaction between HSF and the monomeric ovalbumin protein.
Figure 7 represents the deletion analysis of HSF. (A) Schematic drawing of full-length HSF open reading frame (pHSF WT) and three deletion mutants pHSF 1-367, 1-241, and 1-163. All numbers refer to amino acid positions. The open bars represent HSF coding
sequences; the shaded regions A-D represent sequences conserved between Drosophila and yeast HSF. The solid boxes indicate the φlO promoter (T7 ) , the Shine- Delgarno (SD) sequence, and the transcription
terminator (T) of the expression vector. (B) SDS polyacrylεuaide gel analysis of wild-type and mutant HSF polypeptides. Mutants were expressed in BL21(DE3) in the presence of 35 S-methionine (20 μCi/ml) . 0.1 ml of culture was precipitated and the pellet was denatured at 100°C in 10 μl of Laemmli sample buffer. Samples were electrophoresed on a 15% polyacrylamide gel and visualized by fluorography. The arrows pointing left indicate HSF polypeptides. A 26 kD protein (arrowhead) was also labeled in all samples, including E. coli transformed with the expression vector alone. (C) DNase I protection analysis of HSF mutants. A labeled fragment from the hsp70 promoter was incubated with the indicated amounts of wild-type or mutant HSF proteins. Footprinting reactions were performed essentially as described in Figure 3C.
Figure 8 shows the sequence comparison of
Drosophila and yeast HSF. (A) Dot matrix plot of conserved amino acids between Drosophila HSF (horizontal) and yeast HSF (vertical) , using the UWGCG sequence analysis programs Compare (window/stringency 30/17) and Dotplot. (B) Amino acid alignment of conserved regions A-D, using the UWGCG sequence
analysis program BestFit, with default parameters.
Vertical lines indicate amino acid identities.
(:) indicates similar amino acids, according to Dayhoff as normalized by Gribskov and Burgess (1986) .
Conserved regions A-D are shaded. There are sequence similarities that extend beyond the somewhat arbitrary boundaries imposed on each conserved region.
Figure 9 shows the comparison of the DNA binding domains of Drosophila HSF, yeast HSF, σ32 and σ70 , and the comparison of the hydrophobic amino acid heptad repeats in Drosophila HSF and yeast HSF. (A) Alignment of protein sequences conserved between Drosophila HSF, yeast HSF, σ32 , and σ70. Similar residues are boxed. The first helix of the putative helix-turn-helix motif of σ32 starts at L-253, the turn at G-261, and the second (recognition) helix at A-264, the three residues comprising the turn are boxed. The Drosophila HSF sequence shows 27% identity/ 46% similarity to the σ32 sequence in the block of 26 amino acids. (B)
Comparison of the heptad repeats of hydrophobic amino acids found in Drosophila and yeast HSF sequences. The two sequences are aligned without gaps using conserved region B as defined by the Bestfit sequence analysis program as the starting frame of alignment. The repeats are made up of hydrophobic residues at
positions a (open diamonds) and d (filled diamonds) , in the nomenclature for coiled-coils (a b c d e f g)„ . The small diamonds represent a third array of hydrophobic residues positioned out of register by one residue from the second array. Heptad repeats of the yeast HSF sequence are taken from Sorger and Nelson (1989) . Backbone illustration of hypothetical packing of α-helices are shown with the positions of
hydrophobic residues stippled.
Figure 10 shows a Western blot analysis of the molecular size of natural HSF present in cytoplasmic extracts of unshocked Drosophila cells. (A) .
Nondenaturing gel electrophoresis of cytoplasmic extracts prepared from Drosophila Schneider line-2 (SL- 2) cells according to the method of Dignam (A.D. Dignam et al.. Methods Enzymol. 1983 101, 582). 1.5 μl of nonshocked (0°C, [-]) or in vitro heat shocked (34 "C for 10 min. [+] cytoplasmic extract was diluted to 5 μl and subjected to non-denaturing gel electyophoresis followed by Western blotting. The position of high molecular weight protein standards (Pharmacia) run on the same gel are indicated. The membrane was processed for immunostaining with 1:1000 dilution of rabbit anti-Drosophila HSF polyclonal antibodies and 1:40,000 dilution of goat anti-rabbit antibody conjugated with alkaline phosphate, according to manufacturer's
instructions (Tropix) . A chemiluminescent substrate (Tropix) was employed to visualize the presence of secondary antibody. The membrane was wrapped in saran wrap and exposed to X-ray film. If the primary
antiserum was omitted or pre-incubated with 1 μg/ml recombinant Drosophila HSF protein, the specific reaction with HSF was not observed. (B) . SDS-PAGE of SL-2 cell cytoplasmic extracts, followed by Western blotting and immunostaining with anti-Drosophila HSF antibodies, as described above.
Figure 11 represents a model for HSF regulation. Heat shock or stress conditions destabilize the
inactive form of HSF symbolized by (A) a homodiner of HSF (oval) or (B) a heterodimer composed of HSF and an inhibiting protein (square) , leading to the assembly of HSF hexamers, which binds to HSEs with high affinity and activates transcription of heatshock genes.
Figure 12 shows the polymerase chain reaction using cDNA prepared from HeLa (lanes 2,3) and Drosophila (lanes 4,5) poly A+ RNA. Reaction products were analyzed on a 1% agarose gel and visualized by ethidium bromide staining. Lane 1 contains a control reaction using a Drosophila HSF control (Clos et al, 1990) . The PCR reaction was carried out according to the
manufacturer's instructions (Perkin Elmer Cetus) . 2μl (lanes 2,4) or 9μl (lanes 3,5) of the cDNA reaction was used for PCR amplification in a final volume of 50 μl, with 0.5 μl (0.7μg/μl) each of primer I: 5'
GCCGGC[N]TT[C/T]CTGGCCAA[A/G]CT[N]TGG and primer ii: 5' CTGGAGCCA[N]AC[C/T]TC[A/G]TT[C/T]TC. The reaction was programmed for 1.5 minutes at 94°C, 2 minutes at 60°C, 3 minutes at 72 βC repeated 27 times with a change of the melting step to 1 minute at 94 βC for cycles 2 to 28 and the last extension step was for 6 minutes at 72 "C. 20 μl of the reaction was applied to each gel lanes. The reverse transcription reaction contained in 50 μl: 5μl 10X PCR reaction buffer, 20μl 10 mM dNTP (each 2.5mM), 2.5μl of 0.2μg/μl pdN6 , 1 μl (20 units)
placental ribonuclease inhibitor, 1.25 μl MgCl2 , 2.5 μl Murine Leukemia Virus reverse transcriptase, and 5μgns system. Peptides were eluted with a gradient of 0% to 50% acetonitrile in 0.1% TFA and individual peaks were collected onto glass fiber filters. The filters were dried in vacuo and subjected to amino acid sequence analysis on an Applied Biosystems 477A Protein
Sequencer coupled to a 120A analyzer.
E. coli Strains and Plasmids used for Recombinant Drosophila expression
For routine cloning and plasmid amplification the strains Xl-1 Blue (STRATAGENE) or DH-5α (BRL) were used. Lambda gtll, EMBL 3 phage, and their derivatives were propagated in strains Y1090 or LE392,
respectively. The strain BL21(DE3) (Studier and
Moffatt, 1986,) served as host for bacterial expression of HSF. Subcloning of genomic DNA and cDNA inserts, and reconstruction of the full-length HSF cDNA were performed with pBluescript II KS(+) (Stratgene) . pHSF poly A contains HSF cDNA (positions -15 to +2540, combined from pHSF407 and pHSF312, see Figure 2A) inserted in the EcoRU site of pJCl. pJCl was
constructed by fusing a (dA)100 sequence derived from the plasmid pSP65AT (Baum et al, 1988, Dev. Biol. , 126, 141-149) between the Smal and BamHI sites of
pBluescript II KS (+) . This plasmid allows the
transcription of HSF RNA containing a poly A tail, under the control of the T3 RNA polymerase promoter for in vitro translation and icroinjection studies. The bacterial expression vector p ClO was constructed by ligation of the Scal/Bglll (blunted) fragment from pET 3C (Rosenberg et al, 1987, Gene, 56, 125-135) which contains the T7φl0 promoter, translation signals and transcription terminator, plus the 5' half of the ampR region, with the Scal/PvuII fragment from pBluescript II KS (+) , containing the 3 • half of the ampR region and the col El origin of replication. pJCIO is smaller than pET3 and is a high copy-number plasmid allowing high yields in analytical plasmid preparations. pHSFWT was constructed by creation of a Ndel site at the start codon of the HSF cDNA, and ligation of a Ndel-BamHI HSF fragment to pJCIO (linearized with Ndel and BamHI) .
The Ndel-BamHI fragment contains 2532 nt of HSF
sequences from the initiating AUG codon, plus 16 nt at the 3 ' end from pBluescript II KS (+) . Nested deletion mutants were generated by ExoIII/Sl digestion of pHSFWT cleaved at the StuI and Apalsites (see Figure 2C) following the manufacturer's protocol (Pharmacia).
Screening of cDNA Libraries for Drosophila HSF
The Drosophila genomic library in EMBL 3 and the oligo dT-primed cDNA library were gifts from John Tamkun and Jim Kennison. The random-primed cDNA library was a gift of Bernd Hovemann. The genomic library was screened by hybridization with two
oligonucleotides, oligo 27 and 29, at 37βC in 6 x SSC. The final wash was done at 48 °C in 3.2 M
tetramethylammonium chloride (TMA-C1) (Wood et al, 1985, Proc . Natl . Acad. Sci . USA, 82, 1585-1588; Devlin et al, 1988, DNA, 7, 499-507). Plaque hybridization of the cDNA libraries in lambda gtll was carried out as follows: hybridization and washed at 65"C in 6xSSC and 0.5xSSC, respectively, using an 1800 bp Sall-EcoRI fragment from genomic clone EMBL 3-104. Twelve cDNA clones were isolated,seven of which were sequenced after subcloning into pBluescript II KS(+).

Screening of cDNA Libraries for human HSF
Approx. 106 plaques of a human B cell lymphoma cDNA library (gift of L. Staudt, NIH) and a human activated B cell cDNA library (gift of J. Kehrl and A. Fauci, NIH; obtained through L. STaudt, NIH) in the lambda gtll vector were screened. Three nitrocellulose filter replicas were prepared from each plate
containing approx. 50,000 plagues. The replicate filters were screened with the human HSF PCR fragment, labeled with 32P-dCTP by the random prime procedure, and with two oligonucleotides derived from the sequence of the human HSF PCR fragment, labeled with 32 P-gamma-ATP by the kinase reaction. The sequences of the two oligonucleotides are:
5'GATGTTCTCAAGGAGCTGCTCCTGGCCACGCAGGAAGCATGGTGCTGGAACTC C
and
5 'AAGCACAACAACATGGCCAFC/TTTCA
The coordinates of the human HSF PCR fragment sequence are +45 to +513 on the sequence shown in Figure 12.
Filters were prehybridized with 6X SSC, 5X Denhardt's solution, 0.1% SDS for 1 hr at 65"C, and hybridized with labeled DNA under the same conditions for 12-16 hr. Filters were then rinsed trice with IX DSD, at 65*C for 15 min, rinsed briefly in IX SSC, blotted dry and exposed to X-ray film for approx. 16 hr. Only plaques which gave reaction with all three probes were considered positive. After three rounds of plaque purification, the cDNA inserts were subcloned into the vector pBluescript SK- for sequence
determination by the dideoxynucleotide technique.

Preparation of Drosophila HSF RNA and Translation in vitro
20 μg/ml pHSFpolyA was cleaved with Xbal and incubated for 60 min at 37βC in a 50 μl volume
containing 40 mM Tris-HCl, pH 8.0, 8 mM MgCl2 , 5 mM

DTT, 4 mM spermicide, 400 mM each of ATP, CTP, UPT, and m7G(5')ppp(5')Gm, 40 mM GTP, 50 μg/ml BSA, 1000
units/ml of RNase Inhibitor (Boehringer Mannheim) and 40 units/ml of T3 RNA polymerase (Boehringer Mannheim) . RNA was extracted with phenol-chloroform, precipitated with ethanol and redissolved in HPLC grade water
(Fisher Scientific) .
Rabbit reticulocyte lysate (Promega) was treated with Staphylococcus aureus nuclease (Boehringer Mannheim) as directed in Sanbrook et al. Molecular Cloning: A laboratory Manual 1983. 1 μg of in vitro-transcribed HSF RNA was translated for 2 hrs at either 25βC or 30βC, in a 25 μl volume containing 50%
translation lysate, 20 μM of each amino acid, 1000 units/ml RNase Inhibitor, and 0.2 mCi/ml 35 S-methionine (1000 Ci/mmole, DuPont-NEN) . Small aliquots of the reaction were subjected to SDS gel electrophoresis and fluorography to verify the translational efficiency and accuracy, the remainder was frozen in liquid nitrogen and stored at -80"C.
Expression and Purification of Reenmb-inant Drosophila HSF in E. coli
BL21(DE3) cells transformed with pHSFWT orits derivatives were grown at 37 βC to OD600=0.6 in M9TB/amp medium (10 g Bacto-Tryptone (Difco) , 5 g NaCl, 1 g NHAC1, 3 g KH2P04 , 6 g Na2HP04 , 4 g glucose, 1 mM
MgS0 , and 50 mg ampicillin/liter) . IPTG was added to 0.4 mM, and the cultures were transferred to 18βC.
After 40 to 60 min incubation, 40 mg of rifampicin was added to suppress transcription by bacterial RNA polymerase, and incubation was continued at 18βC overnight, with shaking. Bacteria were pelleted by centrifugation (6000 x g, 10 min, at room temperature) , and resuspended in 1/100 volume of buffer CB+400 mM KCl (buffer CB: 20 mM HEPES pH 7.6, 1.5 mM MgCl2 , 0.1 mM DTT, 2 mM leupeptin, 10% (v/v) glycerol) . After disruption by sonication at 100 mW for 2 min (B.
Braun) , the lysate was incubated for 30 min on ice.
The bacterial debris was removed by centrifugation (6000 x g, 10 min 4'C) and the supernatant was diluted 2-fold with buffer CB and centrifuged at 100,000xg at 4βC for 1 hr. The supernatant containing crude
recombinant HSF was frozen in liquid nitrogen and stored at -80βC.
In order to purify recombinant HSF, 40 ml of the crude supernatant was diluted with buffer CB to a KCl concentration of 100 mM and chromatographed on a 20 ml Heparin-Sepharose CL-6B column. HSF was diluted with a linear KCl gradient (100-500 mM) in buffer CB. HSF activity was monitored by gel mobility shift assays and active fractions were diluted to 100 mM KCl with buffer CB. HSF was further chromatographed on a 1 ml Mono Q column (Pharmacia) , and eluted with a linear KCl gradient (100-500 mM) in buffer CB. Active fractions contained the 105 kD HSF protein purified to 90% homogeneity, as determine by SDS gel electrophoresis and silver staining. The total protein concentration was 3.5 mg/ml, as determined by a dye-binding assay (Biorad) .

Gel Mobility Shift Assay
DNA-binding was monitored by the gel mobility shift assay as described previously (Zimarino and Wu, 1987, Nature, 327, 727-730), using a double-stranded, synthetic HSE carrying three [_GAA_] repeats in alternating orientation (Zimarino et al, 1990, Hoi . Cell . Biol . , 10, 752-759). The DNA wa labeled with 32 P by primer extension as described previously (Wu et al, 1987, Sciencβββ, 1247-1253). For the experiments shown in Figure 3A, 2 μl samples of protein were mixed with

10 fmole of 32P-labeled HSE, 2.5 μg of poly (dl-dC) .poly (dl-dC) , 5 μg yeast tRNA, 0.5 μg of sonicated E. coli DNA and 0.5 μg of poly (dN)5 in 10 μl of 10 mM HEPES pH 7.9, 1.5 mM MgCl2 , 0.05 mM EDTA, 120 mM NaCl, and 6% glycerol. Samples were incubated on ice for 10 min and electrophoresed on a 1.2% agarose, 0.5X TBE gel. The gel was blotted and dried onto DE 51 paper and
autoradiographed.
DNase I Footprinting
DNA fragments labeled with 32P at one 5' end were synthesized by the polymerase chain reaction (PCR) using a combination of one 5' labeled oligonucleotide primer and one unlabeled primer. An Xhol-Acc I
fragment (positions -185 to +295) from the hsp70 gene promoter (locus 87A) cloned into pBluescript I SK(+) served as template for the PCR. The oligonucleotide primers used were: hsp70 lower stand positions +149 to +177, T7 sequencing primer (Stratagene) , hsp70 upper strand positions -140 to -120, and hsp70 lower strand positions +10 to +29. 50 fmoles of the labeled DNA fragment was incubated at room temperature with
recombinant HSF extracted from E. coli under the same conditions as described for the gel mobility shift assays. After 10 min, DNase I (Pharmacia) was added (300 u/ml) and the incubation was continued for another 2 min. The reaction was stopped by the addition of EDTA and SDS to 10 mM and 1%, respectively, and the DNA was extracted with phenol-chloroform and precipitated with ethanol. Primers that were 5' end-labeled for the polymerase chain reaction were also used for dideoxy sequencing reactions as a reference.
In vitro transcription
Two supercoiled plasmid templates were used for in vitro transcription. phsp70(-50)HSE2 carries a modified hsp70 promoter in a pBluescript vector
(Stratagen) . The modified hsp70 promoter consists of hsp70 (locus 87A) sequences from -90 to +296, in which two upstream HSEs were remodeled according to Xiao and Lis 1988, Science, 239, 1139-1142, keeping the natural spacing between the HSEs and the hsp70 TATA box.
phsp70(-50) minigene is similar to phs70(-50)HSE2 , except for a deletion of a 30 bp, Alul fragment between +41 and +71, and substitution of sequences from -50 to -90 (containing the HSEs) with a synthetic
polylinker. Transcription extracts were prepared from 0-12 hr D. melanogaster (Oregon R, P2) embryos (Soeller et al, 1988, Genes Dev. , 2, 68-81; Biggin and Tjan, 1988, Cell, 53, 699-711). Care was taken not to inadvertently heatshock the embryos. Protein from the ammonium sulfate precipitation step was dialyzed to a
conductivity equivalent to HEMG:100mM KCl and stored in aliquots at -80°C (HEMG; Soeller et al, 1988).
Transcription with crude embryo extracts was performed according to Heberlein et al (1985), Cell , 41, 965-977, modified as follows for RNA recovery: after addition of 100 μl stop mix (minus SDS) and 100 μl phenol to the transcription reactions, the samples were mixed in an Eppendorf shaker for 2 min. 100 μl of
chloroform: isoamyl alcohol 241 was added and the mixing was repeated. The aqueous phase was transferred to a fresh tube, re-extracted with organic solvent, and nucleic acids were precipitated with ammonium acetate. After thorough washing with 80% ethanol, the pellet was dried in vacuo and dissolved in 9 μl of 250 mM KCl, 2 mM Tris-HCl, pH 7.9, 0.2 mM EDTA. 1 μl of 32 P-labeled primer (hsp70 positions +149 to +177) was added, and the primer was annealed by incubation at 75°C for 5 min, and at 42 βC for 20 min. After addition of 25 μl of 50 mM Tris-HCl, pH 8.3, 10 mM MgCl2 , 5 mM DTT, 1 mM EDTA, 1 mM each dNTP, the primer was extended with 7 units AMV reverse transcriptase (Promega) at 42 °C for 45 min.
Translation of HSF RNA by microinjection in Xenopus oocytes
Xenopus laevis females were obtained from Nasco or Xenopus 1. Pieces of ovary were surgically removed and the connective tissue digested with 0.2% collagenase (Sigma type II) in OR-2 medium (Wallace et al, 1973, J. Exp. Zool . ,184, 321-334). Stage VI oocytes were incubated for about 12 hours in OR-2 with ImM oxaloacetate as exogenous energy source (Eppig and Steck an, 1976, In Vitro, 12, 173-179) before
microinjection. All procedures were performed at 16-18 °C, except where indicated.
HSF RNA was adjusted to a concentration of approximately 0.4 ng/nl in injection buffer (90 mM KCl, 15mM Hepes, pH 7.5). Approximately 25 nl (10 ng) of RNA was injected into each oocyte using a micropipet attached to an adjustable 10 μl Drummond pipettor as described (Westwood, 1988, Abnormal proteins and the induction of heat-shock gene expression. Ph.D. thesis, University of California, Berkeley) . After 10 hours, groups of injected oocytes were transferred to 1.5 ml microfuge tubes containing approximately 50 μl of OR-2 medium, and heat-shocked at 36 βC for 10 min. Non- shocked oocytes were left at 18βC. The medium was removed and the oocytes rinsed quickly with 100 μl 0βC homogenization buffer (50 mM KCl, 10 mM Hepes, pH 7.9, 0.5 mM PMSE, 0.5 mM DTT) . Individual oocytes were transferred to fresh tubes, and homogenized by repeated pipetting with a micropipettor (10 μl buffer per oocyte) . The lysate was centrifuged for 5 min at
12,000 xg at 4"C, and the supernatent transferred to a fresh tube, avoiding the top lipid layer. Extracts were either frozen in liquid nitrogen or assayed immediately by the gel mobility shift technique (5 μl extract in a 10 μl final volume) .
Pore exclusion limit electrophoresis
0.5 μl (2.5 μg) of recombinant HSF purified to the Mono Q step, and high molecular weight marker proteins (Pharmacia #17-0445-01) , were electrophoresed on a 4-20% polyacrylamide gradient gel in 0.5X TBE buffer. Electrophoresis was continued for 24 hours at duration of electrophoresis was necessary for proteins to have migrated to their exclusion limit (Andersson et al, 1972, FEBS Letters, 20, 199-201).
Size estimation of the HSF-HSE complex was performed by electrophoresis of a mixture of HSF and 32P-labeled HSE (under standard gel-shift assay
conditions) on a 3-12% polyacrylamide gradient gel in 0.5X TBE buffer, as above. The gel was stained with Coomassie Blue, destained, equilibrated in water, dried,and autoradiographed.
Chemical cross-linking
2 μg of cloned Drosophila HSF (Mono Q
fraction) was incubated with glutaraldehyde or EGS (Pierce) at room temperature for 10 min in 10 μl of 175 mM NaCl, 15 mM Tris-HCl, pH 8.0, 0.1 mM EDTA, and 1.5 mM MgCl2. Reactions were quenched with 30 mM lysine and 1 volume of 2X Laemmli sample buffer. Samples were heated to 95"C for 5 min; aliquots were separated on a SDS 4-6% polyacrylamide gel without a stacking gel, and si1ver-stained.
In situ hybridization
Preparation of chromosomal squashes for in situ hybridization followed standard procedures
(Ashburner, 1989, Drosophila. A Laboratory Manual. Cold Spring Harbor Laboratory Press) . The DNA probe was substituted with digoxigenin-dUTP by a random priming reaction, and hybrids were detected according to instructions supplied with the Genius kit (Boehringer Mannheim) .
Preparation of rabbit polyclonal antibodies to cloned Drosophila HSF protein
Cloned Drosophila HSF purified as described above was used an antigen for immunization. A does of 500 μg protein per rabbit, mixed with Freund's adjuvant was injected intradermally, followed two booster injections of 250 μg protein each at 3 week-intervals. Serum was collected and stored at -70°C.

Western blotting of nondenaturing polyacrylamide gradient gels
After electrophoresis, the gel was incubated in transfer buffer (48 mM Tris, 39 mM glycine)
containing 0.25% SDS for 10 min at 75°C and allowed to cool to room temperature before electroblotting for 4 hr at 100 mA onto an Immobilon P membrane (Millipore) in transfer buffer containing 0.05% SDS, using a Novex semi-dry blotting apparatus. After blotting, the membrane was stained with 0.2% ponceau S (in 3% trichloroacetic acid, 3% sulfosalycic acid) , and destained briefly to visualize the molecular weight markers. The membrane was then processed for
immunostaining with 1:1000 dilution of rabbit anti- Drosophila HSF polyclonal antibodies and 1:40,000 dilution fo goat anti-rabbit antibody conjugated with alkaline phosphatase, according to manuf cturer's instructions (Tropix) . A chemiluminescent substrate (Tropix) was employed to visualize the presence of secondary antibody. The membrane was wrapped in Saran wrap and exposed to X-ray film.
Example 1. Purification and microseguencing of
Drosophila HSF
Drosophila HSF was purified to about 95% homogeneity by a modification of the procedure
described previously (Wu et al, 1987) (Figure 1A) . Two independently-purified, 4 μg preparations of the 110 kD polypeptide were digested with trypsin, and the
resulting peptides were subjected to reverse phase liquid chromatography (RPLC) . Essentially identical elution profiles were observed for both peptide
preparations. Individual HSF peptides were subjected to microsequence analysis, and the amino acid sequence of six peptides that yielded identical sequences in duplicate are shown in Figure IB.
Example 2. Isolation of cDNA clones for Drosophila HSF
Two 20-mer oligonucleotides with 32-fold degeneracy, based on the predicted nucleotide sequences of HSF peptide 27 and peptide 29 (Figure IB) , were used to probe a Drosophila genomic library. Initially two genomic DNA clones were identified which contained a common, 1800 nt Sall-EcoRI fragment. This Sall-EcoRI fragment, which hybridized with both oligonucleotide probes, was then used to isolate cDNA clones from a random-primed and an oligo dT-primed cDNA library. The 2.8 kb of HSF cDNA sequence reconstructed from six overlapping cDNA clones reveals a single open reading frame of 691 amino acids (2073 nt) (Figure 2A) . The sequences of all six HSF tryptic peptides within the 691-amino acid open reading frame were located, and thus concluded that this reading frame encodes
Drosophila HSF (Figure 2B) . The molecular mass of

Drosophila HSF, calculated from the deduced amino acid sequence is 77,300 daltons, significantly lower than the apparent mass of 110,000 daltons measured by SDS gel electrophoresis (Wu et al, 1987) . Evidently,
Drosophila HSF has an anomalous mobility on SDS gels; a similar anomaly was observed with yeast HSF (Sorger and Pelham, 1988, Cell, 54, 855-864; Wiederrecht et al, 1988, Cell , 54, 841-853). For purposes herein, the molecular size of HSF protein as measured by SDS gel electrophoresis will be used throughout the examples. The Drosophila HSF protein sequence predicts an acidic protein (pi = 4.7). The overall distribution of charged residues along the length of the protein sequence is nonuniform: the N-terminal one-third of HSF (amino acids 1-240) is relatively basic (predicted pi = 10.25), while the C-terminal two-thirds (amino acids 240-691) is reactively acidic (predicted pi = 4.1). In addition, there is an unusual N-terminal cluster of 9 acidic residues in a row (amino acids 18 to 26) .
DNA gel blot analysis under standard
stringency conditions shows that the Drosophila HSF gene is single-copy. The possible presence of
homologous genes that have partial sequence similarity to HSF has not yet been addressed. The Drosophila HSF gene was localized by in situ hybridization to
cytological position 55A on the Drosophila polytene chromosome (Figure 2C) .
Example 3. DNA binding activity of Recci"*""*™*-Drosophila HSF in the absence of heat shock
Naturally occurring HSF extracted from the cytosol of nonshocked Drosophila cells shows a basal affinity for DNA, which can be significantly increased by a direct heat treatment in vitro, or by reaction with polyclonal antibodies raised to the in vivo-aσtivated form of HSF (Zimarino et al. 1990, Science, 249, 546-549; Figure 3A, lanes 1-3). The slower mobility of the HSF:HSE complex upon anti-HSF treatment is due to the additional binding of antibody. When recombinant HSF was synthesized by m vitro translation in a rabbit reticulocyte lysate at 25βC, or at 30βC, neither heat treatment (34 βC) nor reaction with anti-HSF serum increased HSF affinity for DNA (Figure
3A, lanes 4-9). The low activity of HSF translated at 25βC is due to reduced translational efficiency at this temperature. The specific binding of HSF translated in vitro was demonstrated by a DNA competition experiment (Figure 3B) . The constitutive DNA binding activity of HSF synthesized in vitro could be due to an activating substance in the reticulocyte lysate. However, it was found that reticulocyte lysates do not activate HSF when incubated with cytosol from unshocked Drosophila cells.
HSF was over-expressed in E. coli at 18 βC using the T7 RNA polymerase-dependent expression system
(Studier and Moffatt, 1986, J. Hoi . Biol . , 189, 113-130). Recombinant HSF isolated from E. coli showed maximal DNA binding affinity without heat or anti-HSF treatment; see also Figure 6C, lanes 3,4). HSF expressed at low levels in bacteria also shown maximal affinity without heat or anti-HSF serum treatment; hence, over-expression per se does not lead to activation.
Specific binding of HSF produced in E. coli was
confirmed in vitro by a DNase I protection assay, which shows binding to the HSEs upstream of the hsp70 gene (Figure 3C, 3D) . The DNase I protection pattern is identical to the pattern obtained with natural HSF purified from heat shock Drosophila cells (Wu et al, 1987) . The data suggest that recombinant HSF protein synthesized outside the environment of a higher
eukaryotic cell has an intrinsic affinity for DNA.
The ability of HSF produced in E. coli to function as a transcription factor in an in vitro
transcription system derived from Drosophila embryos (Soeller er al, 1988, Genes Dev. , 2, 68-81); Biggin and Tjian, 1988, Cell , 53, 699-711) was examined next.
Addition of the recombinant protein to the transcription extract resulted in a 7-fold increase of transcription from a promoter carrying two HSEs, relative to the transcription from the same promoter lacking HSEs (Figure 4) . Hence, recombinant HSF protein is capable of functioning as a transcription factor in a binding site-dependent manner, apparently without further modification by a heat shock-induced enzymatic activity.
Example 4. Heat shock-inducible DNA binding activity of recombinant Drosophila HSF expressed in Xenopus oocytes
Naturally occurring HSF in crude extracts of unshocked Drosophila, Xenopus, and vertebrate cells shows a basal affinity for DNA by in vitro assays, which is increased about 10-fold when cells are induced by heat shock (Zimarino et al, 1990, Hoi. Cell. Biol. , 10, 752-759) . In this example, the activity of recombinant HSF synthesized after microinjection of Xenopus oocytes with HSF RNA transcribed in vitro was tested. The endogenous Xenopus HSF is undetectable in these
experiments. Although there is some fluctuation in the basal DNA binding activity is of the recombinant protein in crude extracts in individual unshocked oocytes (Figure 5, lanes 1-5) , DNA binding activity is insignificantly induced (5-fold, on average) after heatshock for 10 min (Figure 5,lanes 6-10). The amount of Drosophila HSF protein synthesized in oocytes subjected to heatshock was equivalent to the synthesis in control oocytes, as determined by 35 S-methionine incorporation and SDS gels electrophoresis. Thus, in contrast to the full DNA binding capacity of HSF synthesized in E. coli or in a reticulocyte lysate, the intrinsic affinity of HSF for DNA is suppressed in nonshocked Xenopus oocytes. The results suggest that the naturally occurring form of HSF unshocked
Drosophila cells is under negative control, which is relieved upon heat shock.
Example 5. Recombinant Drosophila HSF expressed in E. coli. associates as a τ.ay»ιm->r in solution
The apparent molecular mass of recombinant HSF, purified from E. coli extracts, was determined to be about 105 KD by SDS polyacrylamide gel
electrophoresis (see Figure 6) . This size is in agreement with the apparent mass (110 kD) of the natural protein purified from Drosophila cells (Wu et al, 1987, Science, 238, 1247-1253); the 5 kD difference could be due to gel mobility fluctuations or to post-translational modification of the natural protein. The native size of recombinant HSF was estimated by pore exclusion limit analysis (Anderson et al, 1972, FEBS Letters, 20, 199-201). In this procedure, proteins are electrophoresed for extended periods (about 24 hr) on nondenaturing polyacrylamide gradient gels; each protein migrates until it reaches the pore exclusion limit, which is dependent, to a first approximation, on the size of the protein. A native HSF molecule that migrates with an estimated size of 690 kD is observed (Figure 6A) . Minor aggregates migrate above and below the 690 kD species and very large aggregates are also visible near the origin of electrophoresis.
The native size of cloned Drosophila HSF bound to the HSE was estimated by pore exclusion limit analysis of the protein-DNA complex (Huet and Sentenac, 1987, Proc. Nat. Acad. Sci . USA, 84, 3648-3652; Hooft et al, 1987, Nucleic Acids Res . , 15, 7265-7282). Using recombinant HSF expressed in E. coli, a HSF:32 P-labeled HSF complex which migrates with a size of 690 kD was observed, similar to the HSF hexamer free in solution (Figure 6B, lanes 3,4). Since the HSE contribution to the overall protein-DNA complex is negligible (assuming one native HSF molecule binds to one or several HSEs) , this result suggests that the hexamer is the active, DNA binding form of HSF. HSF-HSE complexes are not detected in the vicinity of the origin of
electrophoresis, suggesting that the very large HSF complexes observed in Figure 6A are aggregates which lack biological activity. The complex of HSE bound to naturally occurring HSF in crude Drosophila cell cytosol after in vitro heat activation was also sized. The mobility of the natural Drosophila HSF:HSE complex was found to be similar to the mobility of the
recombinant HSF:HSE complex (Figure 6B,lanes 1,2).
Together, these results suggest that the active form of natural Drosophila HSF free in solution and when bound to DNA may be a hexamer of the 110 kD subunit.
The multimeric state of cloned Drosophila HSF was confirmed by chemical cross-linking. Cloned
Drosophila HSF protein cross-linked with limiting amounts of glutaraldehyde (Landschulz et al. , 1989, Science, 243, 1681-1688) and analyzed on an SDS gel displayed a ladder of cross-linked products whose apparent sizes are approximate multiples (up to six) of the 105 kD HSF monomer (Figure 6C, lane 2). HSF oligomers were sized relative to cross-linked
phosphorylase b markers (97 kD monomer) . Increasing the glutaraldehyde concentration enhanced the abundance of HSF trimer and hexamer, in addition to larger species at the limiting mobility of the gel. Similar results were obtained with the bifunctional reagent BGS (Abdella et al, 1979, Biochem. Biophys. Res . Com. 87, 734-742) (Figure 6D, lanes 1-5) . More importantly, a 100-fold dilution of cloned HSF protein (to 2 μg/ml) gave essentially the same abundance of HSF oligomers (Figure 6D, lanes 6 , 7), suggesting that the multimerization of HSF is not due to an artifically high concentration of the cloned protein.
Example 6. Drosophila HSF Regions important for specific and high affinity binding to DNA
As a first step towards a molecular dissection of the structure of HSF protein, progressive 3'terminal portions of the HSF coding sequence were deleted
(Figure 7A) , and the mutant genes were expressed in E. coli (Figure B) . C-terminal truncations of HSF protein, up to residue 163 (HSF 1-163) , are still capable of binding to DNA (Figure 7B) . However, HSF 1-163 shows a distinctly lower affinity for the hsp70 promoter compared to the affinity of full-length HSF. From the HSF protein concentrations required to achieve roughly 50% binding to DNA, it is estimated that HSF 1-163 binds with about 50-fold lower affinity relative to the binding of full-length HSF. The binding of HSF 1-241 and HSF 1-367 differ from full-length HSF by no more than 2-fold. These results show that HSF 1-163 is sufficient for binding specifically to HSEs, while an adjacent region, from residues 164 to 241, is important for high-affinity binding.

Example 7. Conserved sequences between Drosophila and yeast HSF
A comparison was made between the primary amino acid sequence of Drosophila HSF with the
published sequence of yeast HSF (Wiederrecht et al, 1988, Cell, 54, 841-853; Sorger and Pelham, 1988). It is striking that despite the high degree of homology among heat shock proteins between species as diverse as E. coli and Drosophila (about 50% identity, for hsp70; Bardwell and Craig, 1984, Proc. Natl . Acad. Sci. USA, 81, 848-852) , the sequences of Drosophila and yeast HSF have diverged over a large portion of the proteins. A dot matrix plot of sequence similarities revealed two major and two minor regions of local conservation
(Figure 8A) . Among the four regions, region A is most conserved between Drosophila and yeast HSF. Out of 66 amino acids, 33 are identical (50% identity; 73% similarity, allowing for conserved substitutions)
(Figure 8B) .
Conserved region B shows 44% identity and 67% similarity in 33 amino acids. Region B is contained within a larger region of yeast HSF that is required for trimerization of the yeast factor (Sorger and
Nelson, 1989, Cell, 59, 807-813). Regions C and D show 27% identity, 41% similarity, and 28% identity, 51% similarity, respectively. These regions are not involved with DNA recognition, since they can be deleted without affecting the DNA-binding function.
Regions C and D are notably represented by polar amino acids, and among the 23 identical residues combined for both regions, 10 are serines or threonines, potential candidates for phosphorylation. Four of the identical residues are acidic.
Among the four regions conserved between
Drosophila and yeast HSF, the 66 amino acid region A is most conserved (50% identity) . This region is included within the DNA binding domains of both Drosophila and yeast HSF (this example and Wiederrecht et al, 1988, Cell , 54, 841-853), and may therefore organize a
structural domain for specific DNA recognition. In E. coli , heat shock genes are positively regulated by a special sigma subunit of RNA polymerase, σ32 (Gross et al, 1990, The function and regulation of heat shock proteins in Escherichia coli. In: Stress proteins in biology and medicine. Cold Spring Harbor Laboratory Press, 167-190.) The DNA binding domains of Drosophila HSF and yeast HSF with the σ32 protein sequence were compared and a short conserved region was found which is also represented in the major E. coli sigma subunit, σ70 (Figure 9A) . Intriguingly, many of the conserved amino acids are located in the putative DNA recognition helix of the sigma factors (Gribskov and Burgess, 1986, Nucl . Acids Res. , 14, 6745-6763; Helmann and Chamber1in, 1988, Ann. Rev. Biochem. , 57, 839-872). These results suggest that the homology to the putative recognition helix of sigma factors may define an element of the HSF DNA binding domain that is important for DNA binding.
Example 8. Heptad repeats of hydrophobic >mfiτ-ιr> acids
Two lines of evidence implicate sequences within and surrounding conserved region B in the self-association of Drosophila HSF. First, C-terminal
deletions that remove 78 residues between amino acids 163 and 241 reduce the affinity for DNA, but not the specificity, by as much as 50-fold. Second, region B of yeast HSF has been shown directly to mediate trimerization of a truncated yeast HSF protein (Sorger and Nelson, 1989, Cell, 59, 807-813). These workers first noted an array of heptad repeats of hydrophobic residues in the yeast HSF oligomerization domain, and proposed a triple-stranded coiled-coil model for the yeast HSF trimer. A second, heptad array of
hydrophobic residues located 18 amino acids C-terminal to the first array was suggested to contribute to the stability of the trimeric interface.
In this example, the first and second array of hydrophobic amino acid repeats in Drosophila HSF
(Figure 9B, large diamonds) were found. In addition, a third array of hydrophobic residues, positioned one residue out of register with the second array (Figure 9B, small diamonds) was discovered. When the second and third array of heptad are viewed in a backbone model of an α-helix, it becomes evident that the helix has hydrophobic residues juxtaposed at four positions on one helical face. Such a helix would have the potential to associate with two neighboring helices of the same type by hydrophobic interactions
characteristic of leucine zipper coiled-coils
(Landschulz et al, 1988, Science, 240, 1759-1764; O'Shea et al, 1989, Science, 245, 646-648). It is likely that these three assays of hybridization repeats direct assembly of the HSF hexamer.
The conserved amino acids in the
oligomerization domain are not limited to hydrophobic residues. Identical residues include polar amino acids (three glutamines in a row [QQQ]), hydrophobic [W,F,I,L], basic [R,K] and acidic [E] amino acids.
Although hydrophobic interactions are the major
stabilizing force between coiled-coils, additional specificity may be conferred by charged or polar interactions, mediated by residues outside the heptad repeat (Cohen and Parry, 1990, Proteins, 7 , 1-15). The conserved residues may also be involved with
interactions of the HSF subunit with other proteins (see example 9) .
Example 9. The native Heat Shock Factor from
nonshocked Drosophila cell cytosol is a homo- or hetero-dimer
Using Western blots stained with Drosophila HSF-specific antibodies, the size of the inactive HSF present in nonshocked Drosophila cell cytosol was measured by pore exclusion limit electrophoresis on a nondenaturing polyacrylamide gradient gel. As shown in (Figure 10A) , the inactive form of Drosophila HSF migrates with a native size of approximately 220 kDa. In vitro activation of this HSF causes some of the 220 kDa species to be converted to 690 kDa, the native size of the active recombinant HSF protein from Drosophila. The specificity of the anti-HSF serum shown by the staining of the 110 kDa HSF subunit from crude
Drosophila cell cytosol after SDS-PAGE and Western blot analysis (Figure 10(B)). These results suggest that the native HSF protein increases from a dimeric state (2x110 kDa) to a hexameric state (6x110 kDa) upon heat stress activation. Alternatively, the inactive state of the HSF could be composed of a HSF monomer complexed to an inhibitor protein of similar size, and it is this HSF inhibitor complex which is disrupted upon heat stress, leading to the assembly of the active HSF hexamer (Figure 11) .
Example 10. A model for heat shock regulation in higher eukaryotes
The naturally occurring form of HSF in
Drosophila cells binds to DNA with high affinity only under stress conditions. Recombinant HSF synthesized in E. coli or in a rabbit reticulocyte lysate shows maximal affinity for DNA without a heat shock; this affinity is suppressed when HSF is synthesized in
Xenopus oocytes. The results herein suggest that HSF protein has an intrinsic tendency to fold to the active conformation, which is suppressed in higher eukaryotic cells. Since the inactive HSF molecule appears to be dimeric, suppression of the intrinsic HSF activity therefore occurs by a block in the assembly of the HSF hexamer. This block in assembly could be due to the preferred association under normal conditions of HSF as a homodimer or a heterodimer composed of one subunit of HSF and an inhibiting molecule. (Figure 11) . The inactive HSF dimer is thus the target for the
multiplicity of stress inducers, besides heat, which include drugs affecting energy metabolism, oxidizing agents, sulfhydryl reagents, chelating agents, heavy metals, ionophores, amino acid analogues, etc.
(Ashburner and Bonner, 1979, Cell 17, 241-254; Nover, 1984, Biol. Zentr, 103, 357-435). Applicants and others cited herein have shown that the inactive state of HSF is easily altered in vitro by physical and chemical changes in environment. If the inactive HSF dimer is maintained in a metastable state by a diverse
combination of molecular forces, for example, by hydrophobic, charged, and polar interactions, then the disruption of a subset of these forces by any one inducer of the stress response could be sufficient to trigger a change of state, and lead to the formation of HSF hexamer.
Example 11. Isolation of cDNA clones for human HSF (HuHSF)
Heat shock transcriptional activation, heat shock factor has been cloned from human. The cloning of human heat shock factor (HuHSF) was achieved by using short stretches of homologous sequences between Drosophila and yeast heat shock factors as primers in the polymerase chain reaction (PCR) . (Figure 12) . The human HSF length clone was obtained by screening human cDNA libraries with the amplified sequence. The HuHSF cDNA clone includes an open reading frame of 529 amino acids with a calculated molecular weight of 58,000. (Figure 13) . The size of HuHSF as measured by SDS-polyacrylamide gel electrophoresis is 60,000 which is in close agreement with the calculated size. (Figure 14).
Example 12. Expression of yfttf-ft- - n ion mm*τι HSF in E.coli
The open reading frame of 529 amino acids was inserted into the expression vector pJC20 by
introducing a site for the restriction endonuclease, Ndel, by site-directed mutagenesis at the initiating AUG codon. The plasmid was then restricted with Ndel and ECoRI and the fragment corresponding to the entire open reading frame was isolated from the gel and ligated into pJC20 previously restricted with the two enzymes (Figure 15) . No extra amino acids are added to the expressed protein using this system. BL21(DE3) cells carrying the T7 polymerase gene under the control of a lac uv5 promoter were transformed with the
plasmid. As a control, cells were transformed with the vector pJC20 alone. A single colony was picked from the plate and cells were grown in LB broth containing 0.4% glucose and 20 μg/ml ampicillin to an OD600 of 0.5. Isopropyl-B-D thiogalactoside (IPTG) was added to a concentration of 0.4 mM and incubation continued at 37°C for 3 hours. Cells were harvested by
centrifugation and resuspended in HEMGN (25 mM HEPES, pH 7.9, O.lmM EDTA, 12.5 mM MgCl2 , 10 % glycerol, 0.1% NP-40, 1 mM DTT) containing 300 mM KCl. The cells were disrupted by sonication using 6 pulses of 20 seconds each at 25 to 30 W power. Cells were placed in ice-water for 30 seconds between pulses. Extracts were clarified by centrifugation at 10,000 g for 10 minutes and flash-frozen in liquid nitrogen. Extract proteins were analyzed by SDS-PAGE and stained with Coomassie Blue.
Example 13. DNA binding transcription factor activity by recombinant human HSF in the absence of heat shock
HuHSF expressed in E.colinder non-shock conditions was shown to be capable of binding
specifically to the heat shock regulatory elements in vitro as determined by the gel mobility shift assay (Figure 16) and by nuclease protection experiments (Figure 17) essentially as described for the Drosophila HSF protein in Example 3.

Example 14. Transcriptional activity of τ*-«r__**mιfr_-i nunt HUHSF
The ability of the cloned human HSF protein to function as a transcription factor in vitro was
demonstrated using a heat shock plasmid template and a cell-free transcription system derived from Drosophila embryos, essentially as described in Example 4 for the recombinant Drosophila HSF protein. In this example, addition of extracts from E. coli expressing cloned human HSF to the assay caused a --6-fold stimulation of transcription in vitro (Figure 18) . This increase, similar to that observed with cloned Drosophila HSF protein, is dependent on protein binding to HSEs, since no stimulation was observed on a template with the HSEs deleted. The ability of the 529 amino acid ORF
encoding human HSF to function as a HSE-dependent transcription factor indicates that this ORF encodes most or all of a human HISF protein.
For purposes of completing this disclosure, all documents cited herein are hereby incorporated by reference.
While the foregoing invention has been
described in some detail for purposes of clarity and understanding, it will be appreciated by one skilled in the art from a reading of this disclosure that various changes in form and detail can be made without
departing from the true scope of the invention.