JBC Invitrogen Ultrasensitive Cytokine Assays

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Originally published In Press as doi:10.1074/jbc.M106301200 on August 16, 2001

J. Biol. Chem., Vol. 276, Issue 43, 40254-40262, October 26, 2001
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
276/43/40254    most recent
M106301200v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Bulman, A. L.
Right arrow Articles by Nelson, H. C. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Bulman, A. L.
Right arrow Articles by Nelson, H. C. M.

The DNA-binding Domain of Yeast Heat Shock Transcription Factor Independently Regulates Both the N- and C-terminal Activation Domains*

Amanda L. BulmanDagger §, Susan T. Hubl§, and Hillary C. M. NelsonDagger ||

From the Dagger  Johnson Research Foundation and Department of Biochemistry and Biophysics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania 19104 and the  Department of Molecular and Cell Biology, University of California, Berkeley, California 94720

Received for publication, July 5, 2001, and in revised form, August 16, 2001


    ABSTRACT
TOP
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
REFERENCES

The expression of heat shock proteins in response to cellular stresses is dependent on the activity of the heat shock transcription factor (HSF). In yeast, HSF is constitutively bound to DNA; however, the mitigation of negative regulation in response to stress dramatically increases transcriptional activity. Through alanine-scanning mutagenesis of the surface residues of the DNA-binding domain, we have identified a large number of mutants with increased transcriptional activity. Six of the strongest mutations were selected for detailed study. Our studies suggest that the DNA-binding domain is involved in the negative regulation of both the N-terminal and C-terminal activation domains of HSF. These mutations do not significantly affect DNA binding. Circular dichroism analysis suggests that a subset of the mutants may have altered secondary structure, whereas a different subset has decreased thermal stability. Our findings suggest that the regulation of HSF transcriptional activity (under both constitutive and stressed conditions) may be partially dependent on the local topology of the DNA-binding domain. In addition, the DNA-binding domain may mediate key interactions with ancillary factors and/or other intramolecular regulatory regions in order to modulate the complex regulation of HSF's transcriptional activity.


    INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
REFERENCES

All cells respond to stresses, such as elevated temperatures and other changes in cellular conditions, by increasing the expression of heat shock proteins, a set of proteins that protect the cell from damage caused by stress and also aid the cell's recovery after the removal of stress. In eukaryotes, the increased expression of heat shock proteins is regulated by the heat shock transcription factor (HSF).1 HSF binds to a conserved DNA sequence called a heat shock element (HSE), which is defined as a series of alternating, inverted repeats of the conserved sequence NGAAN (1, 2). HSEs range in size from two to more than six repeats, and all heat shock promoters have at least one three-repeat HSE or two proximal two-repeat HSEs. The two-repeat HSEs can be arranged in either a head-to-head (NGAANNTTCN) or tail-to-tail (NTTCNNGAAN) orientation. The strength of HSF's activation depends primarily on the number of HSEs within the promoter and secondarily on the sequence of the non-conserved "N" residues and the orientation of the binding site.

Budding yeast, such as Saccharomyces cerevisiae, have only a single HSF, which is an essential gene that has a constitutive function (3, 4). Under all cellular conditions, yeast HSF appears to be localized in the nucleus in a DNA-binding competent form (5, 6). Key to the function of yeast HSF is the fact that it has different levels of transcriptional activity in response to different cellular conditions. Under constitutive or normal growth conditions, yeast HSF has a low level of transcriptional activity, which can vary 10-fold between growth at 15 and 30 °C (3). After heat shock or stress, yeast HSF acquires a high level of transcriptional activity, which can be up to 200-fold greater after a transient 30-min heat shock at 39 °C (3). This increased level diminishes rapidly once the transient stress is removed. If cells are shocked and maintained at the heat shock temperature for several hours, there is an early peak of activity, which then readjusts to a basal level that is significantly higher than the level prior to the increase in temperature (7). Jakobsen and colleagues (8) have shown that yeast heat shock activation responds to a change in temperature, rather than an absolute temperature.

There are at least four clearly defined domains for the S. cerevisiae gene HSF1 (Fig. 1). The central core, responsible for DNA binding and oligomerization, has been highly conserved in all HSFs. The trimerization domain consists of a three-stranded coiled-coil (9-11). The DNA-binding domain is a variation of the winged-helix-turn-helix motif (12), with an unusual alpha -helical bulge and proline-centered kink combination in its second helix (i.e. the first helix of the helix-turn-helix motif) (13, 14). There are two activation domains, located at the N and C termini of the protein. The N-terminal activation domain is found within the first 170 amino acids (7, 15), with the most potent part of the activation domain within the first 66 amino acids.2 The C-terminal activation domain is located between residues 595 and 783 (7, 15, 16). The activation domains appear to play different roles, with the N-terminal activation domain responsible for the initial transient increase in activity in response to stress and the C-terminal activation domain responsible for the more sustained stress response (7). Nevertheless, either activation domain can be deleted without affecting viability at constitutive (i.e. non-heat shock) temperatures (7, 15). Deletion of the C-terminal activation domain results in a temperature-sensitive phenotype, emphasizing its role in the ability of yeast to thrive at higher growth temperatures. In addition, the C-terminal activation domain is critical for the heat-induced or glucose starvation-induced increase of expression of the metallothionein gene CUP1, a non-canonical heat shock gene, but not for the heat-induced increase of expression of SSA1 and SSA3, two of the yeast Hsp70 homologues (17).

The regulation of yeast HSF's activity occurs at many levels, although it appears to function primarily through negative regulation of HSF's activity in the absence of stress. The level and extent of yeast HSF's phosphorylation is dependent on cellular conditions (3); therefore, it is likely that protein kinases and phosphatases are involved in regulating HSF's activity. Studies on HSF from Kluyveromyces lactis, which is closely related to S. cerevisiae, have shown that phosphorylation is likely involved in repressing HSF's activity after stress is removed (18). Several studies have shown that heat shock proteins help to repress mammalian HSF's activity after the removal of stress (19, 20), although it is not clear whether this is the case in yeast (8, 21).

At least four cis-acting regions of yeast HSF have been identified as important in the regulation of its activity. In particular, they are involved in restraint of HSF's activity in the absence of stress, as their deletion causes an increase in HSF's constitutive activity, suggesting that these regions function as negative regulators by masking the activation domains. This negative regulation could occur through intermolecular interactions with other proteins or through intramolecular interactions between these cis-acting regions and the activation domains.

The four cis-acting regions include N-terminal and C-terminal regulatory regions, as well as the DNA-binding and trimerization domains (Fig. 1). Deletion of the N-terminal regulatory region, located between residues 40 and 170, results in a total removal of the negative regulation of the C-terminal activation domain, such that HSF has the same high level of activity before and after stress (7, 22). Deletion or mutation of the CE2 regulatory region, located between residues 532 and 552 and including a short C-terminal hydrophobic heptad repeat and several critical serine residues, also results in HSF with an increased constitutive activity (18, 22, 23). Because the CE2 regulatory region is involved in the sustained, rather than the transient heat shock response, it is likely regulating the C-terminal activation domain, rather than the N-terminal activation domain (18), although no definitive studies have been reported to support this.


View larger version (5K):
[in this window]
[in a new window]
  		Fig. 1.   Schematic diagram of full-length HSF from S. cerevisiae. AD, activation domain; RR, regulatory region.

The central core, which encompasses the DNA-binding and trimerization domains, is also involved in repressing transcriptional activity (15). Studies on a series of fusions between fragments of HSF and a heterologous DNA-binding domain have shown that removal of the central core increases HSF's transcriptional activity (15). Furthermore, several studies have shown that point mutations within either the DNA-binding or trimerization domains cause an increase in constitutive and/or induced transcriptional activity (13, 14, 16, 22, 24). The DNA-binding domain mutations include M232V, identified as increasing the activity of a heterologous VP16 transcriptional activation domain, and R206S, identified as one of two mutations within HSF responsible for high basal transcription of the metallothionein gene CUP1, known to be regulated by HSF. In addition, a detailed study of the bulged-kinked second helix of the DNA-binding domain showed that it was critical for transcriptional activity; removal of the alpha -helical bulge through single amino acid deletions causes an increase in transcriptional activity that might be related to thermal destabilization of the DNA-binding domain (13). Substitution of the conserved proline, which is located at the center of the kink, by a lysine (though not alanine) causes an increase in transcriptional activity (14). None of these studies looked systematically at how these domains influenced activity, nor did they analyze whether the central core has differential regulation of the two activation domains.

In order to explore further the role of the DNA-binding domain in regulating transcriptional activity, we have studied the transcriptional activity of a series of alanine-scanning mutants that cover the surface of the DNA-binding domain (12, 25). The majority of the surface residue mutations cause a change in transcriptional activity compared with wild-type HSF. Of these, only a small number decrease transcriptional activity, whereas ~80% cause an increase in transcriptional activity. By mapping the positions of these residues onto the crystal structure of the DNA-binding domain, we can conclude that multiple surfaces of the DNA-binding domain are involved in regulation of transcriptional activity. We have focused on six of the mutants with the largest increase in transcriptional activity. We analyzed how the mutations affected the transcriptional phenotype in the absence of one or other of the activation domains. Our results show that the DNA-binding domain is involved in negative regulation of both the N- and C-terminal activation domains. Intriguingly, we also found that specific residues differentially affected the activity of the two transcriptional activation domains. The ability of the DNA-binding domain mutations to influence transcriptional activity does not appear to be related to changes in DNA-binding affinity, but may be related, in some cases, to changes in the thermal stability and/or structure of the domain. Models for how the DNA-binding domain influences negative regulation of transcriptional activity are discussed.

    EXPERIMENTAL PROCEDURES
TOP
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
REFERENCES

Plasmids and Yeast Strains-- The DNA-binding domain mutants used for yeast expression were all constructed from pHN1002, a CEN/AR4S/TRP1 yeast shuttle vector carrying a full-length copy of the S. cerevisiae HSF1 gene expressed from its own promoter (25). Construction of the full-length versions of the alanine-scanning mutants has been described (25). C-terminal truncations of specific alanine-substitutions were based on pHN2256, a version of pHN1002 that includes the first 583 residues of HSF followed by valine and asparagine as cloning artifacts and then a stop codon (gift of E. E. Powers). The DNA-binding domain mutations were transferred from the full-length background into pHN2256 using unique BamHI and StyI sites. N-terminal truncations of specific alanine substitutions used the plasmid pHN2241, a version of pHN1002 in which residues 145-833 are expressed from the natural HSF1 promoter (gift of E. E. Powers). This plasmid was digested with NcoI (which was blunted by filling-in with DNA polymerase) and StyI; this creates a backbone that includes the HSF1 promoter and the last 440 residues of HSF. Plasmids containing the full-length HSF versions of the mutants were digested with BamHI (which was blunted by filling-in with DNA polymerase) and StyI; this creates a fragment of HSF containing residues from 67 to 393. Ligation of these restriction fragments with the digested pHN2241 resulted in plasmids that used the natural HSF1 promoter to express residues 67-833 of HSF, preceded by methionine and aspartate (cloning artifact) at the N terminus.

The yeast expression plasmids were all transformed into the tester strain PS145 (gift of P. Sorger). This strain is a haploid derivative of S. cerevisiae W303 (MATa ade2-1 trp1 can1-100 leu2, 3-112 his3-11, 15 ura3), which carries the HSF1Delta 2::LEU2 chromosomal disruption and a wild-type HSF1 gene under the control of a GAL1 promoter on a derivative of the URA3-containing plasmid YCp50 (3); it can only grow when galactose is present. The transformants were tested for their ability to grown in the absence of galactose. Once it was confirmed that the particular alanine-substituted HSF was sufficient to allow viability (25), the transformant was cured of the pGAL1-HSF1 URA3-based plasmid by growth on 5-fluoroorotic acid and uracil (26). The resultant cells were transformed with the plasmid pHSE2-lacZ, a 2-µm URA3-based plasmid in which the lacZ gene is driven by a synthetic heat shock promoter (6).

beta -Galactosidase Assays-- Single colonies from a fresh plate were used to inoculate selective synthetic media. The cells were grown at 30 °C for 15-18 h until the cells were in early log phase growth (A600 ~ 0.3). To heat induce the cells, 5 ml of culture were transferred into a warm culture tube and shaken in a 42 °C water bath for 30 min, followed by a recovery growth period in a 30 °C shaking water bath for 90 min. For every strain with a mutant HSF, two to five colonies were assayed with a non-induced and induced culture for each colony.

There were some minor differences in the assays done for the original 58 full-length HSF mutants and the assays done later for the 6 HSF mutants with increased transcriptional activity. Because all of the 58 full-length HSF mutants could not be tested within a single day, care was taken to ensure reproducibility from day to day. As an additional control, at least two colonies with wild-type HSF were tested each day that the assay was performed. Each set of raw data from each mutant was compared with the entire set of data from all of the wild-type controls. In order to analyze the data, Student's t test assuming equal variance was used from the statistical package in Microsoft Excel. The significance value was set at 0.5.

For the six mutant HSFs, the assays were grouped according to the length of the HSF construct, i.e. HSF-(1-583), HSF-(66-833), or HSF-(1-833). Within a single day, all six mutants and wild-type HSF could be tested, and each set of mutants was tested over several days. Because HSF-(1-583) grows poorly, all three sets of these mutants were grown in selective synthetic media with 3% dextrose (compared with 2% dextrose found in standard recipes).

beta -Galactosidase activity was determined as described previously (6, 27). Briefly, crude yeast extract was prepared from 3-4.5 ml of cells, which were resuspended in 200 µl of Z buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 1 mM MgSO4, 50 mM beta -mercaptoethanol). 10 µl of 10% Triton X-100 and 200 µl of acid-washed glass beads (425-600 µm, Sigma) was added, followed by a 10-min incubation on ice. The tubes were then vortexed for 10 min at 4 °C, and the cell debris was separated from the crude extract by a 10-min centrifugation. Total protein concentration was estimated using a Bradford protein assay. For the assay, typically 100 µl (for the non-induced cells) or 50 µl (for the induced cells) of crude cell extract was added to Z buffer to a final volume of 500 µl. To start the assay, 100 µl of 4 mg/ml ONPG (in 0.1 M KPO4) was added and the tubes placed in a 28 °C water bath. The reaction was stopped by raising the pH with the addition 250 µl of 1 M Na2CO3; this addition was done either at the appearance of a yellow color or at a 4-h maximum time point. The amount of the cleavage produce was detected by absorbance at 420 nm. The activity was calculated as (A420)/(t·v·p), where t is time in minutes, v is volume of extract in microliters, and p is protein concentration in mg/ml.

Western Blotting-- Cells were grown to saturation or to early log phase. Typically, 1.5 ml (saturated cells) or 4.5 ml (log phase cells) were pelleted and resuspended by vortexing in 200 µl of 2× SDS-PAGE loading buffer (0.125 M Tris (pH 6.8), 4% SDS, 20% glycerol, 1% beta -mercaptoethanol). The tubes were boiled for 5 min and incubated on ice for 5 min. Following addition of 100 µl of acid-washed glass beads (425-600 µm, Sigma), the tubes were vortexed for 10 min at 4 °C and boiled for 5 min, and then the cell debris was separated from the crude extract by a 10-min centrifugation at 4 °C. The amount of crude extract was quantitated using the Pierce BCA assay. Equal amounts of protein were loaded onto an SDS-PAGE gel. The gel was electroblotted overnight onto an Immobilon filter (Millipore), which was probed with one of two different rabbit anti-HSF antibodies (gifts of P. Sorger and D. Gross). The blots were analyzed either by a colorimetric assay (using a goat anti-rabbit alkaline phosphatase conjugate) or by chemiluminescence (using a donkey anti-rabbit horseradish peroxidase conjugate).

Overexpression and Purification of HSF Fragments Containing the DNA-binding and Trimerization Domains-- The expression plasmid pHN136, previously described as Sc-DT(73L) (28), is a derivative of pET-3b (29) and contains the coding sequence for S. cerevisiae HSF amino acid residues 171-424, which includes the DNA-binding and trimerization domains. Mutations in the DNA-binding domain were introduced into pHN136 by a megaprimer polymerase chain reaction mutagenesis approach and confirmed by DNA sequencing. The plasmids are named pHN484 (K178A), pHN485 (D185A), pHN487 (R206A), pHN488 (N222A), pHN489 (H236A), and pHN490 (K237A).

The pET-3b expression vectors were transformed into the Escherichia coli strain BL21(DE3) (29) containing a derivative of the plasmid pAcYc177 (30) that overexpresses lac repressor. Cells were grown shaking at 37 °C in Terrific Broth containing 50 µg/ml each ampicillin and kanamycin until they reached an A600 of ~0.5. Cells were induced by addition of isopropyl-1-thio-beta -D-galactopyranoside to 3.75 mM, grown for an additional 3 h, harvested by centrifugation, and stored at -20 °C. The frozen pellets were thawed and resuspended in an equal volume of 50 mM Tris (pH 7.0), 600 mM NaCl, 5 mM MgCl2, and 10% glycerol. Lysozyme was added to a final concentration of 0.1 mg/ml, and the cells were incubated by shaking at 4 °C for 1 h. The cell suspension was sonicated, and the lysed cells were centrifuged at 39,000 × g for 20 min at 10 °C. The high speed supernatant was precipitated by addition of saturated ammonium sulfate to a final concentration of 18%. The mixture was centrifuged at 39,000 × g for 20 min at 10 °C. The supernatant was removed and the pellet resuspended in a minimal volume of 50 mM HEPES (pH 8.0), and 600 mM NaCl. All column chromatography was conducted on an Amersham Pharmacia Biotech AKTA Prime chromatography system at 4 °C, with absorbance and conductivity monitored simultaneously. The resuspended pellet was loaded onto a HiPrep 16/10 butyl column (Amersham Pharmacia Biotech) equilibrated in 50 mM HEPES (pH 8.0), 600 mM NaCl, 18% (NH4)2SO4. The column was washed in 50 mM HEPES (pH 8.0) 600 mM NaCl, 9% (NH4)2SO4. The column was developed with a reversed ammonium sulfate gradient; the protein started to elute at 4% (NH4)2SO4. The column fractions containing wild-type or mutant HSF fragments were pooled and concentrated to a volume of less than 20 ml with a Centriprep filtration unit (Millipore) at 10 kDa cut-off. The protein solution was desalted on HiTrap columns (Amersham Pharmacia Biotech) equilibrated in 50 mM Tris (pH 6.7). The desalted protein was loaded onto a HiTrap heparin column (Amersham Pharmacia Biotech) equilibrated in 50 mM Tris (pH 6.7). The column was developed with a NaCl gradient and the protein eluted between 650 and 850 mM NaCl. The column fractions containing wild-type or mutant HSF fragments were again pooled and concentrated to a volume of less than 5 ml with a Centriprep 10 filtration unit (Millipore) at 10 kDa cut-off. If the protein appeared greater than 95% pure by SDS-PAGE, then it was desalted on HiTrap columns (Amersham Pharmacia Biotech) equilibrated in 50 mM Tris (pH 6.7). If the protein appeared less than 95% pure by SDS-PAGE, then it was loaded onto a HiLoad 16/60 Superdex 75 column (Amersham Pharmacia Biotech) equilibrated in 50 mM Tris (pH 6.7). In either case, the purified proteins were lyophilized to dryness and stored at -20 °C. For subsequent analyses, the lyophilized protein was resuspended in the appropriate buffer and protein concentration was determined by Lowry protein assays or by absorbance at 280 nm using a protein extinction coefficient of 32,290 cm-1 M-1.

DNA-binding Assays-- Binding of wild-type and mutant HSF fragments to DNA was performed using variations of previously described protocols (13, 31). Plasmid pBL3 (31), which contains an HSE with three GAA repeats, was digested with XhoI and SpeI. The 145-base pair fragment was purified on a 4% Metaphor (BMA)-agarose gel and the ends were labeled with T4 polynucleotide kinase and [gamma -32P]ATP. Lyophilized protein was resuspended in a small volume of 10 mM Tris (pH 8.0) and 0.1 mM EDTA prior to diluting in the gel shift buffer. Binding of protein to DNA was performed in 50 mM HEPES (pH 8.0), 50 mM NaCl, 10% glycerol, and 0.5 mg/ml bovine serum albumin. gamma -32P-Labeled 145-base pair fragment (0.75 fmol) was mixed with a protein sample (1.8 fmol to 20 pmol) in a final volume of 20 µl and incubated at room temperature for 10 min. Reactions were loaded onto polyacrylamide gels and electrophoresed as described previously (31). Dried gels were used to expose Molecular Dynamics PhosphorImager plates, and data were analyzed as described previously (13).

Circular Dichroism Spectroscopy-- Lyophilized protein was resuspended in 10 mM sodium phosphate, pH 7.0. CD spectra were recorded on an Aviv CD spectrometer (model 62ADS). For scans, measurements were taken at 25 °C at 2-nm intervals with a 1-s averaging time and 1.5-nm bandwidth. For thermal melts, the wavelength was held constant at 222 nm, while the temperature was raised at 2° intervals from 4 to 94 °C with a 5-min equilibration time, 10-s averaging time, and 1.5-nm bandwidth. The program NONLIN was used to calculate the temperature at the inflection point of the unfolding curve, which was assigned as the Tm value.

    RESULTS
TOP
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
REFERENCES

Experimental Rationale-- In a previous study, we had used alanine-scanning mutagenesis, in combination with structural studies on the DNA-binding domain, to analyze systematically the surface residues of the HSF DNA-binding domain (25). The assay depended on the requirement of a functional HSF for yeast viability (3). We used the S. cerevisiae gene for the functional assay and mutated those residues predicted to be solvent accessible based on homology to the Kluyveromyces lactis structure (12). Not only are the sequences of the S. cerevisiae and K. lactis HSF DNA-binding domains highly homologous, but the proteins are functionally interchangeable (8).

Of the 64 mutants tested, we identified 6 that resulted in an inviable phenotype, suggesting that those 6 residues were essential for HSF function. One of these essential residues, Gly244, was located at the tip of the "wing" (Fig. 2); mutation of this residue to alanine caused the protein to be proteolyzed (data not shown). Another residue, Trp194, lies in a hydrophobic pocket and is likely involved in inter- or intramolecular contacts (25). Given the location of the remaining four within or located near helix 3 (Fig. 2), the DNA recognition helix, we postulated that those four residues were involved in binding to DNA. This was later confirmed from analysis of a co-crystal structure between the DNA-binding domain and a two-repeat HSE in a tail-to-tail orientation (32). Of the remaining 58 residues that were not essential for yeast viability, 7 were defined as strictly conserved and 12 were defined as highly conserved among all HSF DNA-binding domains. In this study, we focused on these mutants that were not essential for HSF function and looked specifically at their level of transcriptional activity under constitutive and heat-induced conditions.


View larger version (22K):
[in this window]
[in a new window]
  		Fig. 2.   Amino acid sequence of the DNA-binding domain of S. cerevisiae HSF. The DNA-binding domain starts at residue 170; every 10th residue is marked. All residues that were mutated in the alanine-scanning mutagenesis are in capital letters. Color coding is as follows: red, inviable; black, no change in transcriptional activity; blue, decreased constitutive activity; cyan, decreased constitutive activity and increased induced activity; purple, increased constitutive and induced activity; green, increased constitutive activity; pink, increased induced activity. Strictly conserved residues have a double underline, whereas highly conserved residues have a single underline. Secondary structure characteristics, included below the sequence, are based on the K. lactis structure (12).

Transcriptional Activity of Surface Mutants-- To analyze the transcriptional activity of the mutant HSFs, strains containing each of the remaining 58 mutant HSFs were transformed with a plasmid containing a beta -galactosidase gene whose expression depends on a synthetic HSE within a disabled CYC1 promoter. The transformed strains were assayed for constitutive activity at 30 °C (Fig. 3A), as well as heat-inducible activity after 30 min at 42 °C (Fig. 3B). Thirty-five of the mutant HSFs were determined to be statistically different than wild-type HSF at either the constitutive or induced levels (see "Experimental Procedures" for details). Only six have decreased constitutive activity, and three of those have increased induced activity. The rest of the mutants have increased either constitutive or induced levels of activity. Therefore, mutations in at least half of the solvent-accessible residues of HSF can cause an increase in its transcriptional activity. This strongly supports the idea that the DNA-binding domain is involved in negative regulation of HSF's transcriptional activity.


View larger version (40K):
[in this window]
[in a new window]
  		Fig. 3.   Transcriptional activity of alanine-scanning mutants of HSF's DNA-binding domain as measured from a synthetic HSE-dependent promoter fused to lacZ. A, constitutive activity. B, induced activity. Wild-type levels are shown at either end of the bar graph and connected by a dashed line. Error bars have been omitted for clarity. For mutants that have off-scale values, the exact value is given above the bar.

The six mutants with decreased constitutive activity include His220, a strictly conserved residue located at the turn between helices 2 and 3; Val238, a highly conserved residue located at the end of the third beta -strand; and four nonconserved residues located within the flexible wing (Fig. 2). Val238 and two of the residues in the wing also have increased induced activity.

The 29 residues with increased activity are located throughout the primary sequence of the protein. The mutants can be classified into three categories: those with increased constitutive and inducible activity (17), those with increased constitutive activity (1), and those with increased inducible activity (11). There does not appear to be a correlation with sequence conservation and activity classification, although mutations at all 7 strictly conserved surface residues and at 9 of the 12 highly conserved surface residues do cause at least some change in activity (Fig. 2).

Because of the availability of crystal structures of the K. lactis DNA-binding domain (12, 14), as well as the functional and sequence homology between the K. lactis and S. cerevisiae HSF DNA-binding domains (8, 12, 25), we can map the location of the mutants onto the structures (Fig. 4). The mutations with decreased constitutive activity appear to map at or near the wing and away from the DNA-binding surface (see cyan and blue colors in Fig. 4, B and C). This confirms a previous study that showed that deletion of the wing causes a decrease in transcriptional activity (33). On the other hand, the large subsets of mutations with either increased constitutive and heat shock activity or with only increased heat shock activity are located all over the structure (see purple and pink colors in Fig. 4, B and C). This suggests that negative regulation via the DNA-binding domain is complex and multifaceted.


View larger version (51K):
[in this window]
[in a new window]
  		Fig. 4.   Location of S. cerevisiae mutants on the structure of the K. lactis DNA-binding domain. Top, ribbon diagram, with helix 2 colored orange and helix 3 colored yellow. Middle, space-filling diagram, with side chains of residues color coded as follows: red, inviable; black, no change in transcriptional activity; blue, decreased constitutive activity; cyan, decreased constitutive activity and increased induced activity; purple, increased constitutive and induced activity; green, increased constitutive activity; pink, increased heat shock activity. Bottom, ribbon diagram, with color coding as in top panel. In each line, the structure is rotated 90° and 180° around the y axis.

Focus on Six of the Strongest Mutants-- To study further the relationship between the DNA-binding domain mutants and negative regulation of transcriptional activity, we chose to focus on six of the mutants with the largest increases in transcriptional activity. Four of the mutations are at positions that are strictly conserved in all HSFs (Lys178, Asn222, His236, and Lys237), one is at a position that is highly conserved (Asp185), and the sixth is at a position that is not conserved (Arg206). The set of residues includes positively charged, neutral, and negatively charged side chains, and they are distributed throughout the domain (Fig. 5).


View larger version (44K):
[in this window]
[in a new window]
  		Fig. 5.   Location of six specific S. cerevisiae mutants on the K. lactis co-crystal structure. In the two views, the side chains of residues are colored as follows: yellow, Lys178; pink, Asp185; cyan, Arg206; red, Asn222; orange, His236; blue, Lys237.

To confirm that the mutations did not affect stability in vivo, we used Western blot analysis to compare the levels of protein for the wild-type and mutant HSFs. We used two different polyclonal antibodies against full-length HSF (gifts of P. Sorger and D. Gross) and probed lysates from both saturated and log-phase cultures. The alanine-substitutions did not appear to have any effect on steady-state levels of HSF in vivo (data not shown).

The Mutants Differentially Affect the N- and C-terminal Activation Domains-- Although yeast HSF has two activation domains, yeast will survive with HSF lacking either the N-terminal or C-terminal activation domain, i.e. with HSF-(66-833) or HSF-(1-583), respectively. Because the domains have slightly different functions, strains containing these different versions of HSF have different constitutive and induced levels of transcriptional activity (7). In order to analyze whether the mutations within the DNA-binding domain were specific for a particular activation domain, we constructed strains that contained truncated versions of wild-type and mutant HSFs, as well as the HSE-dependent lacZ fusion. Fig. 6 shows the constitutive and induced level of transcriptional activity for the full-length and truncated HSFs. The most striking result is that some of the mutations differentially affect HSF activity dependent on the presence of a specific activation domain.


View larger version (20K):
[in this window]
[in a new window]
  		Fig. 6.   Transcriptional activity of six strong mutants in three different versions of HSF. A, HSF-(1-833); B, HSF-(66-833); C, HSF-(1-583). Note that the wild-type levels of activity differ for the different truncations, so the results can only be compared within a particular set of truncations (in other words, comparisons can only be made within each part of the figure). White, constitutive; grey, induced.

The mutants can be grouped into three classes: primary effect on the C-terminal activation domain, primary effect on the N-terminal activation domain, and mixed phenotype. For the mutation K237A, the increased activity is entirely due to the C-terminal activation domain. HSF-(1-583)K237A has similar constitutive and induced activity to HSF-(1-583) (Fig. 6C), whereas HSF-(66-833)K237A has dramatically increased constitutive and induced activity relative to HSF-(66-833) (Fig. 6B). For the mutations K178A, D185A, R206A, and H236A, the increased activity is almost entirely due to the N-terminal activation domain (Fig. 6, A and B, compare the level of activity for each mutant relative to wild-type). The mutation N222A has a mixed phenotype: HSF-(66-833)N222A retains a dramatic increase in the level of constitutive activity over wild-type HSF-(66-833) but not induced activity (Fig. 6B), whereas HSF-(1-583)N222A has the opposite (Fig. 6C).

The Mutant Phenotype Does Not Appear to Be Mediated via Changes in DNA-binding Affinity-- From the HSF-DNA co-crystal structure (32), we know that two of the residues make direct contact to the phosphate backbone (His236 and Lys237), whereas two of the residues make water-mediated contacts to the phosphate backbone (Lys178 and Asn222). Arg206 is involved in the dimer interface between two DNA-binding domains when bound to an HSE in the tail-to-tail orientation of NTTCNNGAAN, whereas Asp185 does not appear to be involved in any protein-protein or protein-DNA interfaces.

Because some of the mutations appear to be involved in DNA-binding contacts based on their location with respect to the co-crystal structure, it was important to analyze whether they affected DNA-binding affinity. Fragments of HSF containing the mutant DNA-binding domain and the wild-type trimerization domain were used in a gel-shift assay with a three-repeat HSE. We had previously shown that these fragments contained the same DNA-binding characteristics as full-length HSF (9).

None of the mutants were impaired or improved in their ability to form trimers or larger oligomeric species on DNA (data not shown). The mutants did have slightly increased apparent affinities for DNA, ranging from 1.1- to 3.8-fold over wild-type (Table I). It is possible that this increase in affinity could account for the increase in transcriptional activity seen for the mutant HSFs. However, there are several reasons to suggest that this is not the case. First, removal of an activation domain can cause some of the mutant HSF truncations to have similar levels of transcriptional activity as the same wild-type truncation (Fig. 6). Since the DNA-binding domains and DNA-binding sites were the same for all the truncations, it seems unlikely that DNA-binding affinity could account for the differences in activity for the full-length protein but not the truncations. Second, there is no correlation between the increase in DNA-binding affinity and the increase in transcriptional activity for the full-length HSFs; the mutant with the greatest increase in activity (K237A) has the lowest increase in affinity (1.1-fold over wild-type). Indeed, we had shown previously that this kind of small difference in the range of DNA-binding affinity does not correlate with differences in transcriptional activity (13, 14). Therefore, despite the contacts seen in the co-crystal structure, we conclude that the contacts to the DNA were not critical for the phenotype of these mutants.

                              
View this table:
[in this window]
[in a new window]
 
Table I
Macroscopic DNA-binding affinities (Kd values) for wild-type and mutant HSF DNA-binding and trimerization domains, calculated from gel mobility shift experimental data
Each affinity is an average of at least three experiments. WT, wild-type.

The Mutations Influence Structure and Thermal Stability-- If the mutations within the DNA-binding domain are affecting transcriptional activity by relieving negative regulation of the activation domains, it is possible that they do so by affecting the structure of the DNA-binding domain. We used circular dichroism (CD) spectroscopy to determine if the mutations influenced the secondary structure characteristics of HSF. Fig. 7 shows the CD spectra for the wild-type HSF fragment in comparison to the fragments containing each of the six strong mutations. Two of the mutants, N222A and H236A, have CD spectra that are identical to wild-type, suggesting that their secondary structures are not perturbed. The other four mutants show some differences from wild-type; K178A and K237A differ at the 208-nm minima, while D185A and R206A differ at the 222-nm minima. It is possible that the structural changes found in these mutants indicate conformational changes in the DNA-binding domain variants that increase its ability to unmask the activation domains in the absence of stress.


View larger version (28K):
[in this window]
[in a new window]
  		Fig. 7.   Circular dichroism analysis. The spectrum for each of the six mutants (black circles) is shown with respect to the wild-type spectrum (white squares).

We also used CD spectroscopy to determine whether the thermal stability of the fragments was affected (Table II). One of the mutants, K237A, was not significantly changed from wild-type, whereas the other five had decreased thermal stabilities of 3 to 11 °C relative to the wild-type HSF fragment. It is possible that the destabilization of the DNA-binding domain in some of these mutants contributes to the increase in transcriptional activity via a thermally induced transition, as has been predicted for a series of mutations within the bulged second helix of the DNA-binding domain (13).

                              
View this table:
[in this window]
[in a new window]
 
Table II
Melting temperatures of the wild-type and mutant proteins
Thermal melting was followed by circular dichroism at 222 nm. Melting temperature (Tm) is assigned to the temperature at which 50% of the protein is folded. WT, wild-type.


    DISCUSSION
TOP
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
REFERENCES

We have studied the relationship between the DNA-binding domain and HSF's transcriptional activity under constitutive and induced conditions by analyzing a set of alanine-scanning mutants of the surface residues of the DNA-binding domain. More than half of these mutants affect transcriptional activity to some extent, with the majority of the mutations causing an increase in transcriptional activity under constitutive and/or induced conditions. The mutations are located throughout the DNA-binding domain and involve residues that cover approximately half of the domain's surface. Our results extend those from previous experiments suggesting that the DNA-binding domain is involved in the negative regulation of HSF's transcriptional activity (13-15, 22, 24).

Detailed studies on six of the strongest mutants have led to the following additional conclusions. First, the DNA-binding domain regulates the activity of both the N-terminal and C-terminal activation domains, and specific residues can be specific for a particular activation domain. This can be seen most clearly by comparing the results for K178A, whose phenotype is mediated almost entirely via the N-terminal activation domain and K237A, whose phenotype is mediated almost entirely via the C-terminal activation domain (Fig. 6). Second, the mutant phenotype, i.e. increased transcriptional activity, does not appear to be mediated via an increase in DNA-binding affinity (Table I). This is supported by the fact that the DNA-binding domain mutants have different phenotypes dependent on the number and type of activation domain present. Given that all of these transcriptional assays used the same synthetic HSE-dependent promoter, it seemed unlikely that the phenotype of the DNA-binding domain mutants would be due to changes in DNA-binding affinity.

At the same time, we find that some of the mutations affect the secondary structure of a fragment containing the DNA-binding and trimerization domains, whereas other mutations affect this fragment's thermal stability. There does not seem to be an exact correlation between those mutants whose structure is affected and those whose thermal stability is affected. K178A and K237A have the most dramatic effects on secondary structure, but K178A has a melting temperature that is 7° less than wild-type HSF, whereas K237A is as stable as wild-type. D185A and R206A, with moderate structural changes, have moderate decreases in thermal stability. N222A and H236A have identical secondary structure to wild-type HSF, but the most dramatic decrease in thermal stability.

In previous studies, we had discussed two potential models for how mutations within the DNA-binding domain might affect transcriptional activity (13, 14). Removal of residues from the alpha -helical bulge in the second helix of the DNA-binding domain significantly decreases the thermal stability of the domain by at least 15 °C without appearing to affect the overall structure. These mutants had an increase in transcriptional activity of 4-6-fold. We suggested that the DNA-binding domain might be sensing heat directly, with the decrease in thermal stability causing the mutants to be structurally analogous to the heat-induced state of HSF (13). Alternatively, we found that substitution of the proline that is located at the center of the kink in the same helix with a lysine, but not an alanine or aspartate, caused a significant increase in transcriptional activity. The lysine substitution did not affect the structure of the kink, but it did alter the molecular surface of that face of the protein. We suggested that this might affect the interaction of that surface of the DNA-binding domain with either an ancillary factor or a regulatory region of HSF (14).

Our current studies support those two models. N222A and H236A behave in a manner similar to the bulge-deletion mutants. They have decreased thermal stability of 10-11 °C with no apparent affect on structure, but they have an average of 4-fold increase in activity over wild-type HSF. K237A, on the other hand, has no decrease in thermal stability but a significant change in secondary structure and transcriptional activity. It is likely that this mutant is dramatically affecting the surface structure of the DNA-binding domain, perhaps causing a structural transition that mimics the heat-induced state.

It is also interesting to look at the thermal stability of the six mutants with respect to their influence on a specific activation domain. Three of the four mutants whose increase in constitutive activity depends on the N-terminal activation domain (K178A, D185A, and H236A) have the largest decrease in thermal stability (6 to 11 °C). This implies that the activity of the N-terminal activation domain is influenced by a thermally induced transition, which makes sense as it is more involved in the transient response to heat shock.

Our studies cannot differentiate between models of direct or indirect interactions by the DNA-binding domains with the activation domains. Using cross-linking studies, Bonner and colleagues (34) have shown that the DNA-binding domain can interact directly with the N-terminal activation domain, but not the C-terminal activation domain. This would suggest that mutations such as K237A affect the ability of the DNA-binding domain to interact with another protein that might function to unmask the C-terminal activation domain.

    ACKNOWLEDGEMENTS

We are indebted to Ramona Urbauer for help with cloning and to the DeGrado laboratory for the use of their CD spectrometer. We also thank Karen Flick, Eric Powers, Jeanne Hardy, Marco Cicero, Jamie Schlessman, and Scott Ferguson.

    FOOTNOTES

* This work was supported by National Institutes of Health Grant GM44086.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

§ These authors contributed equally to this work and should be considered joint first authors.

|| To whom correspondence should be addressed: Dept. of Biochemistry and Biophysics, University of Pennsylvania School of Medicine, 813 Stellar-Chance, 422 Curie Blvd., Philadelphia, PA 19104-6089. Tel.: 215-573-7473; Fax: 215-573-2503; E-mail: hnelson@mail.med.upenn.edu.

Published, JBC Papers in Press, August 16, 2001, DOI 10.1074/jbc.M106301200

2 E. E. Powers and H. C. M. Nelson, unpublished data.

    ABBREVIATIONS

The abbreviations used are: HSF, heat shock transcription factor; HSE, heat shock element; PAGE, polyacrylamide gel electrophoresis.

    REFERENCES
TOP
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
REFERENCES

1. Amin, J., Ananthan, J., and Voellmy, R. (1988) Mol. Cell. Biol. 8, 3761-3769[Abstract/Free Full Text]
2. Xiao, H., and Lis, J. T. (1988) Science 239, 1139-1142[Abstract/Free Full Text]
3. Sorger, P. K., and Pelham, H. R. (1988) Cell 54, 855-864[CrossRef][Medline] [Order article via Infotrieve]
4. Wiederrecht, G., Seto, D., and Parker, C. S. (1988) Cell 54, 841-53[CrossRef][Medline] [Order article via Infotrieve]
5. Jakobsen, B. K., and Pelham, H. R. (1988) Mol. Cell. Biol. 8, 5040-5042[Abstract/Free Full Text]
6. Sorger, P. K., and Pelham, H. R. (1987) EMBO J. 6, 3035-3041[Medline] [Order article via Infotrieve]
7. Sorger, P. K. (1990) Cell 62, 793-805[CrossRef][Medline] [Order article via Infotrieve]
8. Hjorth-Sorensen, B., Hoffmann, E. R., Lissin, N. M., Sewell, A. K., and Jakobsen, B. K. (2001) Mol. Microbiol. 39, 914-923[CrossRef][Medline] [Order article via Infotrieve]
9. Sorger, P. K., and Nelson, H. C. M. (1989) Cell 59, 807-813[CrossRef][Medline] [Order article via Infotrieve]
10. Peteranderl, R., and Nelson, H. C. M. (1992) Biochemistry 31, 12272-12276[CrossRef][Medline] [Order article via Infotrieve]
11. Peteranderl, R., Rabenstein, M., Shin, Y. K., Liu, C. W., Wemmer, D. E., King, D. S., and Nelson, H. C. (1999) Biochemistry 38, 3559-3569[CrossRef][Medline] [Order article via Infotrieve]
12. Harrison, C. J., Bohm, A. A., and Nelson, H. C. (1994) Science 263, 224-227[Abstract/Free Full Text]
13. Hardy, J. A., Walsh, S. T., and Nelson, H. C. M. (2000) J. Mol. Biol. 295, 393-409[CrossRef][Medline] [Order article via Infotrieve]
14. Hardy, J. A., and Nelson, H. C. M. (2000) Protein Sci. 9, 2128-2141[Abstract]
15. Nieto, S. J., Wiederrecht, G., Okuda, A., and Parker, C. S. (1990) Cell 62, 807-817[CrossRef][Medline] [Order article via Infotrieve]
16. Chen, Y., Barlev, N. A., Westergaard, O., and Jakobsen, B. K. (1993) EMBO J. 12, 5007-5018[Medline] [Order article via Infotrieve]
17. Tamai, K. T., Liu, X., Silar, P., Sosinowski, T., and Thiele, D. J. (1994) Mol. Cell. Biol. 14, 8155-8165[Abstract/Free Full Text]
18. Hoj, A., and Jakobsen, B. K. (1994) EMBO J. 13, 2617-2624[Medline] [Order article via Infotrieve]
19. Shi, Y., Mosser, D. D., and Morimoto, R. I. (1998) Genes Dev. 12, 654-666[Abstract/Free Full Text]
20. Zuo, J., Guo, Y., Guettouche, T., Smith, D. F., and Voellmy, R. (1998) Cell 94, 471-480[CrossRef][Medline] [Order article via Infotrieve]
21. Bonner, J. J., Carlson, T., Fackenthal, D. L., Paddock, D., Storey, K., and Lea, K. (2000) Mol. Biol. Cell 11, 1739-1751[Abstract/Free Full Text]
22. Bonner, J. J., Heyward, S., and Fackenthal, D. L. (1992) Mol. Cell. Biol. 12, 1021-1030[Abstract/Free Full Text]
23. Jakobsen, B. K., and Pelham, H. R. (1991) EMBO J. 10, 369-375[Medline] [Order article via Infotrieve]
24. Sewell, A. K., Yokoya, F., Yu, W., Miyagawa, T., Murayama, T., and Winge, D. R. (1995) J. Biol. Chem. 270, 25079-25086[Abstract/Free Full Text]
25. Hubl, S. T., Owens, J. C., and Nelson, H. C. (1994) Nat. Struct. Biol. 1, 615-620[CrossRef][Medline] [Order article via Infotrieve]
26. Sikorski, R. S., and Boeke, J. D. (1991) Methods Enzymol. 194, 302-318[Medline] [Order article via Infotrieve]
27. Ausubel, F. H., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A., and Struhl, K. (1995) Current Protocols In Molecular Biology , 2nd Ed. , Greene Publishing Associates/Wiley-Interscience, New York
28. Flick, K. E., Gonzalez, L., Jr., Harrison, C. J., and Nelson, H. C. (1994) J. Biol. Chem. 269, 12475-12481[Abstract/Free Full Text]
29. Studier, F. W., Rosenberg, A. H., Dunn, J. J., and Dubendorff, J. W. (1989) Methods Enzymol. 185, 60-89
30. Stark, M. J. (1987) Gene (Amst.) 51, 255-267[CrossRef][Medline] [Order article via Infotrieve]
31. Drees, B. L., Grotkopp, E. K., and Nelson, H. C. (1997) J. Mol. Biol. 273, 61-74[CrossRef][Medline] [Order article via Infotrieve]
32. Littlefield, O., and Nelson, H. C. M. (1999) Nat. Struct. Biol. 6, 464-470[CrossRef][Medline] [Order article via Infotrieve]
33. Cicero, M. P., Hubl, S. T., Harrison, C. J., Littlefield, O., Hardy, J. A., and Nelson, H. C. (2001) Nucleic Acids Res. 29, 1715-1723[Abstract/Free Full Text]
34. Bonner, J. J., Chen, D., Storey, K., Tushan, M., and Lea, K. (2000) J. Mol. Biol. 302, 581-592[CrossRef][Medline] [Order article via Infotrieve]

Copyright © 2001 by The American Society for Biochemistry and Molecular Biology, Inc.


This article has been cited by other articles:


Home page
 J. Biol. Chem.Home page
D. L. Eastmond and H. C. M. Nelson
Genome-wide Analysis Reveals New Roles for the Activation Domains of the Saccharomyces cerevisiae Heat Shock Transcription Factor (Hsf1) during the Transient Heat Shock Response
J. Biol. Chem., October 27, 2006; 281(43): 32909 - 32921.
[Abstract] [Full Text] [PDF]

Home page
 GeneticsHome page
K.-W. Park, J.-S. Hahn, Q. Fan, D. J. Thiele, and L. Li
De Novo Appearance and "Strain" Formation of Yeast Prion [PSI+] Are Regulated by the Heat-Shock Transcription Factor
Genetics, May 1, 2006; 173(1): 35 - 47.
[Abstract] [Full Text] [PDF]

Home page
 GeneticsHome page
H. Singh, A. M. Erkine, S. B. Kremer, H. M. Duttweiler, D. A. Davis, J. Iqbal, R. R. Gross, and D. S. Gross
A Functional Module of Yeast Mediator That Governs the Dynamic Range of Heat-Shock Gene Expression
Genetics, April 1, 2006; 172(4): 2169 - 2184.
[Abstract] [Full Text] [PDF]

Home page
 J. Biol. Chem.Home page
N. Hashikawa, Y. Mizukami, H. Imazu, and H. Sakurai
Mutated Yeast Heat Shock Transcription Factor Activates Transcription Independently of Hyperphosphorylation
J. Biol. Chem., February 17, 2006; 281(7): 3936 - 3942.
[Abstract] [Full Text] [PDF]

Home page
 Mol. Cell. Biol.Home page
N. Hashikawa and H. Sakurai
Phosphorylation of the Yeast Heat Shock Transcription Factor Is Implicated in Gene-Specific Activation Dependent on the Architecture of the Heat Shock Element
Mol. Cell. Biol., May 1, 2004; 24(9): 3648 - 3659.
[Abstract] [Full Text] [PDF]

Home page
 Proc. Natl. Acad. Sci. USAHome page
T. Chen and C. S. Parker
Dynamic association of transcriptional activation domains and regulatory regions in Saccharomyces cerevisiae heat shock factor
PNAS, January 24, 2002; (2002) 32681299.
[Abstract] [Full Text] [PDF]

Home page
 Proc. Natl. Acad. Sci. USAHome page
T. Chen and C. S. Parker
Dynamic association of transcriptional activation domains and regulatory regions in Saccharomyces cerevisiae heat shock factor
PNAS, February 5, 2002; 99(3): 1200 - 1205.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
276/43/40254    most recent
M106301200v1
Right arrow