| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
2
* Turku Centre for Biotechnology, University of Turku and Åbo Akademi University; and
Department of Biology, Åbo Akademi University, Turku, Finland
2Correspondence: Turku Centre for Biotechnology, Tykistokaty 6B, 20521 Turku, Finland. E-mail lea.sistonen{at}btk.utu.fi
 |
ABSTRACT |
|---|
 TOP ABSTRACT BACKGROUNDSTRUCTURAL AND FUNCTIONAL...HSFs AND THE HEAT...FUNCTIONS OF HSFs EXPAND...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
Key Words: heat shock element • hsps • HSF family • knockout and transgenic HSF models
|
BACKGROUND |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...HSFs AND THE HEAT...FUNCTIONS OF HSFs EXPAND...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
. Today it is well known that all organisms share a
common molecular stress response that includes a dramatic change in the
pattern of gene expression and the elevated synthesis of a family of
stress-induced proteins called heat shock proteins (Hsps) (for a
review, see ref 2
). Expression of Hsps usually results in
repair of damaged proteins and survival of the cell, mainly through the
chaperone function of Hsps (reviewed in ref 3
). The common
signal generated by various stress stimuli is likely to be protein
damage, but how the stress sensing mechanism in the cell exactly
operates is not known.
The heat shock response can also induce a death signal that leads to
apoptosis or rapid necrosis. The expression of small Hsps, especially
Hsp27, and the inducible Hsp70 has been shown to enhance the survival
of mammalian cells exposed to numerous types of stimuli that induce
stress and apoptosis (for reviews, see refs 4
, 5
). The
antiapoptotic Hsp27 and Hsp70 are abundantly expressed in many
malignant human tumors (reviewed in ref 5
). With regard to
the cytoprotective functions of Hsps, Hsp70 has been shown to
contribute to protection of myocardium from ischemic injuries
(6
7
8)
. The basis for the cardioprotective activity of
Hsp70 is likely to be related to its ability to prevent protein
aggregation during ischemic stress. Considering the key role of Hsps in
protection against stress-induced damage, it is of utmost importance to
elucidate the regulatory mechanisms responsible for Hsp expression.
The inducible Hsp expression is regulated by the heat shock
transcription factors (HSFs). In response to various inducers such as
elevated temperatures, oxidants, heavy metals, and bacterial and viral
infections, most HSFs acquire DNA binding activity to the heat shock
element (HSE), thereby mediating transcription of the heat shock genes,
which results in accumulation of Hsps (for reviews, see refs
9
10
11
). Since the isolation of a single HSF gene from
Saccharomyces cerevisiae (12
, 13)
and
Drosophila melanogaster (14)
, several members
of the HSF family have been found in vertebrates (HSF1–4) and plants
(Fig. 1
) (for reviews, see refs 9
, 11
, 26
, 27
). The existence of
multiple HSFs in vertebrates and plants suggests that different HSFs
mediate the responses to various forms of physiological and
environmental stimuli (for a review, see ref 28
).
|
On the basis of early studies, it has been a general assumption that
HSF1, the functional vertebrate homologue of the HSF found in yeast and
the fly, is activated by diverse forms of stress. Subsequently, HSF3, a
unique avian HSF, has been shown to function as a heat-responsive
transcription factor. In contrast to HSF1 and HSF3, HSF2 is not
activated in response to classical stress stimuli, but under
developmentally related conditions. In support of the original
findings, neither HSF2 nor any other HSF is able to functionally
substitute for HSF1 or to rescue the heat shock response in HSF1
knockout mice or in cells derived from these animals
(29
30
31)
. However, the roles of distinct HSFs have been
proposed to overlap depending on stimulatory signals. For example,
human HSF2 and HSF4, but not HSF1, are capable of complementing the
viability defect and conferring thermotolerance in S.
cerevisiae cells carrying a lethal HSF deletion (21
, 32)
. The differential activities of HSFs do not exclude the
possibility that different members of the HSF family could also
cooperate in order to regulate expression of their target genes, as has
been demonstrated for avian HSF1 and HSF3 (33)
. This
review focuses on the current knowledge on transcriptional regulation
of the heat shock response, with the emphasis on novel discoveries on
the differential functions of the HSF family members.
|
STRUCTURAL AND FUNCTIONAL FEATURES OF HSFs |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...HSFs AND THE HEAT...FUNCTIONS OF HSFs EXPAND...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
and 2
). In fact, DBD is the only domain where comparative structural data is
available, since the crystal structure of Kluyveromyces
lactis and the solution structures of K. lactis,
D. melanogaster, and tomato HSF DBDs have been determined
(36
37
38
39)
. Activation-induced trimerization of HSFs is
mediated by three arrays of hydrophobic heptad repeats (HR-A/B)
characteristic for helical coiled-coil structures, i.e., leucine
zippers (Fig. 1
and Fig. 3
) (41)
. The trimeric assembly of HSFs is unusual, as
leucine zipper proteins typically are known to associate as homo- or
heterodimers. The suppression of HSF trimerization is likely to be
mediated by another region of hydrophobic heptad repeats (HR-C)
adjacent to the carboxyl terminus of the protein (42)
. The
HR-C is well conserved among the vertebrate HSFs but poorly conserved
in plant and S. cerevisiae HSFs, which may be a reason
underlying constitutive trimerization of HSF in S.
cerevisiae (43
, 44
; reviewed in ref 9
).
|
|
Alternative splicing leads to the generation of functionally distinct
HSF isoforms, thereby providing an additional regulatory level to
control gene expression. As illustrated in Fig. 1
, two isoforms of
mammalian HSF1 and HSF2, generated by alternative splicing of exon 11
adjacent to the HR-C, have been found (18
, 19
, 45
, 46)
.
Alternative splicing appears to be a common feature of several HSFs, as
HSF4 also exists as two differentially generated isoforms
(21)
. Although the regulatory mechanisms of formation of
the isoforms are not known, tissue-specific expression has been
reported (18
, 19
, 45)
. The most recent example of
regulation via alternative splicing is zebrafish HSF1 (zHSF1), which is
expressed as two isoforms in a tissue- and temperature-dependent manner
(24)
.
|
HSFs AND THE HEAT SHOCK RESPONSE |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...HSFs AND THE HEAT...FUNCTIONS OF HSFs EXPAND...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
, 31
, 47
48
49
50)
. HSF1 is also responsible for mediating
the expression of Hsp70 in response to whole body stress in rodents,
highlighting the central role for HSF1 even during physiological stress
(reviewed in ref 51
).
Oligomeric status of HSF1
In contrast to HSF in budding yeast, D. melanogaster
HSF and HSF1 of higher eukaryotes exist as monomers in unstressed
cells. It has been of great interest to study how the monomeric status
of HSF1 is retained under nonstressful conditions. Studies showing that
HSF1 produced in bacteria and overexpressed in mammalian cells acquires
DNA binding activity spontaneously in the absence of heat shock
(15
, 48
, 52
, 53)
led to the conclusion that the intrinsic
activity of endogenous HSF1 is negatively regulated by a titratable
cellular factor in mammalian cells. However, when recombinant or
overexpressed mammalian HSF1 was expressed at lower concentrations, the
DNA binding activity did not occur spontaneously but was shown to be
regulated by heat (42
, 45
, 54
, 55)
. Therefore,
oligomerization of the factor could be repressed in the monomer,
possibly by intramolecular interactions between the HR domains (Fig. 3)
(42
, 54)
. This concept is further supported by studies
showing direct activation of purified fruit fly HSF by heat and
oxidative stress (56)
. Other inducers of the heat shock
response have no effect on activation of D. melanogaster HSF
in vitro, implying that except for heat and oxidation, heat shock
inducers exert their effects indirectly (56)
.
On the basis of the experiments on negative regulation of HSF1
activity, studies intended to identify HSF1 interacting proteins were
initiated. In fact, evidence for the regulatory role for Hsp70 in HSF1
deactivation had already been provided in earlier studies using a
variety of cell models and experimental strategies (for a review and
original references, see ref 10
). In addition to Hsp70,
other molecular chaperones have been indicated to regulate the activity
of HSF1. For example, reduced levels of Hsp90 have been demonstrated to
lead to HSF1 activation in vitro (57)
. In vivo,
geldanamycin that specifically binds and inhibits Hsp90, activates
HSF1, indicating that Hsp90-containing HSF1 complex is present in the
unstressed cells, but dissociates on stress (57)
. This
provides evidence that Hsp90 by itself or in context with
multichaperone complexes would be an important repressor of HSF1.
Similar kinds of results have been obtained in the oocytes of
Xenopus laevis, where formation of a heterocomplex between
components of the Hsp90 chaperone machinery and HSF1 has been suggested
to play a key role in modulating different steps of the HSF1
activation–deactivation pathway (58
, 59)
. Repression of
HSF1 activity by interaction with various Hsps is essential, especially
considering the need for strictly regulated chaperone levels in
maintaining the cellular homeostasis.
Besides the feedback regulation of HSF1 by Hsps, other proteins have
been found to bind HSF1. For example, in a yeast two-hybrid screen, a
novel 8.5 kDa nuclear protein termed heat shock factor binding protein
1 (HSBP1) was found to interact with the oligomerization domain of an
active HSF1, thereby negatively affecting HSF1 DNA binding activity
(60)
. During inactivation of HSF1 to an inert monomer,
HSBP1 is associated with Hsp70. In Caenorhabditis
elegans, overexpression of the homologous CeHSB-1 results in
reduced ability of the organism to cope with thermal and chemical
stresses, whereas loss of CeHSB-1 leads to similar or slightly better
survival rates than in wild-type animals (60)
. The
association of HSBP1 with HSF1 is interesting with regard to the
mechanisms allowing constitutive DNA binding activity of S.
cerevisiae HSF, since HSBP1 has been found in every organism
studied so far except in S. cerevisiae (60
;
L.-J. Tai, S. McFall, and R. I. Morimoto, personal communication).
All of the HSF1 interacting proteins reported to date act as negative
regulators of HSF1 activity, suggesting that a multicomplex might be
required to keep HSF1 in a state that can readily be activated. It is
of interest to learn whether there are any positive regulators of HSF1
or whether fast activation of HSF1 is an intrinsic property that has
been evolutionary conserved to allow a rapid autoactivation of HSF1
upon stress stimuli.
HSF1 regulation is linked to cellular signaling pathways
The strict regulation of HSF1 is further emphasized by the fact
that its DNA binding and transactivating capacities are uncoupled. This
is exemplified in S. cerevisiae, where HSF binds DNA
constitutively (12
, 43
, 61)
but fails to activate
transcription without further stimuli. Treatment of mammalian cells
with sodium salicylate and other nonsteroidal anti-inflammatory drugs
induces formation of the HSF1 DNA binding trimer without
transcriptional activity (62
, 63)
, demonstrating that in
mammalian cells the DNA binding and transcriptional activities of HSF1
are also uncoupled. In mammalian model systems, phosphorylation is an
important determinant of the transactivating potency of HSF1. The
salicylate-induced HSF1 is constitutively but not inducibly
phosphorylated on serine residues, whereas the heat-induced HSF1 is
both constitutively and inducibly phosphorylated (64)
.
Further, the drug-induced, transcriptionally inert intermediate can be
converted to a transcriptionally active state by a subsequent exposure
to heat shock, during which the only detectable change in HSF1 is its
hyperphosphorylation (64)
. These results have formed a
basis for further studies to determine the phosphorylation sites and to
understand their roles in regulation of HSF1.
As illustrated in Fig. 3
, HSF1 contains two distinct carboxyl-terminal
activation domains, AD1 and AD2, which are under the control of a
centrally located, heat-responsive regulatory domain (RD)
(65)
. Since AD1 does not appear to be heat-regulated by
itself, the regulatory domain of HSF1 has been proposed to play a key
role in sensing heat stress in humans (66)
. Constitutive
phosphorylation of two specific serine/proline motifs, S303 and S307,
is important for the function of the RD and may be critical for
negative regulation of HSF1 transcriptional activity at normal
temperatures, since substitution of serine to alanine causes
constitutive transcriptional competence (67
68
69)
.
Constitutive phosphorylation of S363 has also been found to negatively
regulate HSF1 under normal growth conditions (70)
. The
positive role of phosphorylation in regulation of HSF1 transcriptional
activity is only emerging. So far, one in vivo phosphorylated site
(S230), which has a positive effect on HSF1 transactivating capacity,
has been characterized (C. I. Holmberg, personal communication).
By phosphopeptide mapping of in vivo 32P-labeled
HSF1 and by using phosphopeptide-specific antibodies that recognize the
phosphorylated form of S230, it has been demonstrated that
phosphorylation of S230 is enhanced upon heat shock, and, consequently,
mutation of S230 to alanine correlates with decreased transcriptional
activity of the human HSF1.
To elucidate the mechanisms by which the cell senses stress, it is
important to analyze the cellular signaling pathways leading to the
stress response. On the basis of in vitro analyses and overexpression
studies, S303 and S307 have been suggested to be targeted by the
glycogen synthase kinase 3ß (GSK-3ß) and the extracellular
signal-regulated kinase, whereas S363 might be a site for
phosphorylation by the c-Jun N-terminal kinase (JNK) (67
, 68
, 70
71
72
73)
. The S230, in turn, seems to be a good substrate for
calcium/calmodulin-dependent protein kinase II (CaMK II) (C. I.
Holmberg, personal communication). The majority of the critical
phosphorylation sites and kinases/phosphatases responsible for HSF1
phosphorylation still need to be elucidated.
Stress induces formation of nuclear HSF1 granules
Much attention has recently been paid to specific structures
localized in the nucleus, collectively called nuclear bodies, which
most likely serve as organizational centers to control the activities
of various transcriptional regulators. It had already been observed by
1993 that during heat shock, HSF1 localizes into the nucleus, forming
granule-like structures in human but not in murine cells
(48)
. HSF1 granules are also formed upon other stresses,
such as proteasome inhibition, and exposure to heavy metals and the
amino acid analog azetidine, but not by treatment with the
anti-inflammatory agent sodium salicylate, which does not induce
hyperphosphorylation or transcriptional activity of HSF1
(74
75
76
77)
. This suggests that depending on the stress
stimuli, the various events associated with HSF1 activation are
differentially affected.
The formation of granules seems to be an intrinsic property of human
cells and not that of HSF1, since nuclear HSF1 granules can be detected
upon expression of exogenous mouse HSF1 in human cells
(74)
. The appearance of HSF1 granules parallels the
activation of HSF1 and the transient induction of heat shock gene
transcription (74
, 77)
. During recovery from heat shock,
HSF1 granules are no longer detected, but HSF1 rapidly relocalizes to
the same structures upon subsequent reexposure to heat
(76)
. A target for the HSF1 foci has recently been shown
to be a centromeric region on human chromosome 9 rich in repetitive
sequences (C. Jolly and R. I. Morimoto, personal communication).
Although the structure and function of granules are still unclear, as
they do not colocalize with previously described nuclear structures
(74)
, it has been proposed that granules could be sites
where activated HSF1 is stored and recycled, thereby coordinating the
regulation of heat shock gene expression. Alternatively, granules might
represent sites where HSF1 has activities distinct from transcriptional
regulation, such as a structural role in protecting hypersensitive
sites of the genome (76)
.
HSF3, an avian-specific HSF?
In avian cells, a novel heat shock factor, HSF3, has been cloned
in addition to the homologs of mammalian HSF1 and HSF2
(23)
. Like HSF1, HSF3 is a stress-responsive transcription
factor. Avian HSF2 and HSF3 are expressed at comparable levels among
various tissues, whereas the amount of HSF1 varies greatly in distinct
tissues (78
; reviewed in ref 27
). Upon
activation, a nuclear localization signal is required for
oligomerization of an inactive HSF3 dimer to a nuclear trimer that has
properties of a transactivator (79
, 80)
. The threshold
temperatures required to activate HSF1 and HSF3 are different, as only
HSF1 is activated upon mild heat shock, in contrast to coactivation of
HSF1 and HSF3 after severe heat shock (78
, 81)
. Further,
the amount of HSF3 increases after a severe heat shock, whereas HSF1
protein levels diminish (81)
, suggesting that HSF3 has a
role during severe and persistent stress in avian cells. Despite
efforts, no HSF3 homologue has been found in other than avian cells,
raising the possibility that the mechanisms used to cope with stress
might be organism specific.
A genetic approach has unraveled an interesting interdependent
relationship between chicken HSF1 and HSF3: disruption of the HSF3 gene
results in impaired heat shock response and loss of thermotolerance in
chicken B lymphocyte DT40 cells expressing normal levels of HSF1
(33)
. In the absence of HSF3, HSF1 does not trimerize upon
heat shock, even at milder temperatures, indicating that HSF3 directly
influences HSF1 activity and has a dominant role in the regulation of
heat shock response (33)
. In DT40 cells harboring a double
knockout of HSF1 and HSF3, the basal expression of hsp90
mRNA is
markedly decreased, in contrast to the unaffected basal expression
levels of other hsp mRNAs (A. Nakai, personal communication).
Therefore, the constitutive expression of Hsp90 has been proposed to
determine the capacity of stress resistance of vertebrate HSFs.
HSF4: a novel family member to be characterized
Expression of the most recently discovered mammalian HSF, HSF4, is
restricted to certain tissues, as shown by RT-PCR analysis (20
, 21)
. Similar to HSF1 and HSF2, HSF4 has two isoforms: HSF4a and
HSF4b (Fig. 1)
. The splicing of HSF4 mRNA is complex, as two
alternative 5' splice sites have been found in exons 8 and 9 with a
frameshift in the transcript (21)
. The original cDNA clone
encodes HSF4a, which has been reported to be transcriptionally inactive
and to function as a repressor of heat shock gene expression
(20)
. In contrast, the more recently reported HSF4b
isoform has the potential for transactivating heat shock genes and
substituting for the yeast HSF (21)
. HSF4 has one unique
structural feature: both isoforms lack the HR-C necessary for
suppression of HSF trimer formation (Figs. 1
and 2)
, leading to
constitutive DNA binding activity of HSF4 in vitro (21)
.
However, no constitutive binding of HSF4 to the HSEs in the human hsp70
promoter has been detected by in vivo genomic footprinting of some
human cell lines (82
, 83)
. To establish whether HSF4 is a
stress-responsive factor will require a more elaborate analysis of HSF4
functions in various cell types.
|
FUNCTIONS OF HSFs EXPAND BEYOND THE HEAT SHOCK RESPONSE |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...HSFs AND THE HEAT...FUNCTIONS OF HSFs EXPAND...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
HSF2, a nonstress-responsive member of the HSF family
The expression of Hsp70, induced during hemin-mediated erythroid
maturation of the human K562 erythroleukemia cells (84)
,
was shown in 1989 to be mediated through the HSE, suggesting that
HSF-regulated transcription may also play a role during nonstress
conditions (85)
. In 1991, the cDNAs for human and murine
HSF2 were isolated simultaneously with the HSF1 counterparts
(15
16
17)
. Subsequently, it has been well established that
the hsp70 transcription induced by hemin in K562 cells is mediated
predominantly by HSF2 (reviewed in refs 10
, 28
).
The incapability of HSF2 to respond to classical stress stimuli and the
importance of HSF2 in the hemin-mediated erythroid differentiation
pathway of K562 cells suggest a role for HSF2 in controlling
development and differentiation-specific gene expression. In addition
to the up-regulation of HSF2 expression upon activation (46
, 86)
, its diminished expression appears to be an important
control mechanism during cellular differentiation. This has been
exemplified by the rapid down-regulation of HSF2 during TPA-mediated
megakaryocytic differentiation (87)
. The opposite
HSF2 expression patterns in K562 cells differentiating along either the
erythroid or megakaryocytic lineages are intriguing, since constitutive
HSF2 expression has been observed in various cell types and tissues
(16
, 18)
. However, the possible role of HSF2 in the
distinct differentiation pathways of progenitor cells remains remote
until alternative model systems to K562 cells become available.
HSF2 has also been proposed to be responsible for the specific Hsp
expression observed during developmental processes, and the
constitutive HSF2 HSE binding activity during early embryogenesis as
well as in mouse ES and carcinoma cells has been studied extensively
(88
89
90
91)
. HSF2 binds to DNA and is abundantly expressed in
the developing mouse heart at E11.5–12.5 (92)
. In the
central nervous system, HSF2 HSE binding activity has been shown to
remain high until E15.5 (93)
. In addition, high levels of
HSF2 have been detected in mouse testis, and the expression of HSF2
mRNA has been shown to be regulated with respect to the developmental
stages of mouse and rat spermatogenesis (94
, 95)
. During
mouse embryogenesis and rat spermatogenesis, the pattern of Hsp
expression does not correlate with HSF2 DNA binding activity (92
, 93
, 95)
, whereas opposite results have been obtained during
mouse spermatogenesis (94)
. Some of the discrepancies
could be explained by the finding that the ratio of the alternatively
spliced HSF2 isoforms (Fig. 1)
varies significantly between different
tissues and cell types (18
, 19)
. In adult mouse testis,
the transcriptionally more active HSF2- isoform is predominantly
expressed, whereas the HSF2-ß isoform is more abundant during mouse
embryonic development and rat spermatogenesis (93
94
95)
. It
is plausible that HSF2-ß may not be transcriptionally active enough
to regulate heat shock gene expression. This isoform might also
activate transcription of genes different from heat shock genes or
interact with other transcription factors to either antagonize or
stimulate their transcriptional properties (93)
. The
relevance of HSF2 in development and spermatogenesis is likely to be
elucidated when HSF2 null mice will be available.
Regulation of HSF2 expression is complex and involves several
regulatory levels
HSF2 displays only a 35% overall identity to HSF1, but the DNA
binding and oligomerization domains are highly conserved (Fig. 2)
.
Cloning of the genomic structure of mouse HSF2 has revealed that the
gene is located on chromosome 10 and is composed of 13 exons spanning
at least 43 kbp in the genome (96)
. The functional domains
of mouse HSF2 appear to be related to defined groups of exons: the DNA
binding domain is comprised of exons 1, 2, and 3, the oligomerization
domain (HR-A/B) of exons 4, 5, and 6, and the HR-C of exons 10 and 11
(96)
. The human HSF2 gene shows similar overall structure
and is located on chromosome 6 (P. Nykänen et al., unpublished
results). The cytoplasmic localization of inactive HSF2 has been
implicated to require interaction between the HR-A/B and HR-C domains
(97)
. This interaction could mask the nuclear localization
signals present on both sides of the HR-A/B domain, which upon
induction would be exposed by disruption of the intramolecular
interaction (97)
.
Regulation of HSF2 expression uses both transcriptional and
post-transcriptional control mechanisms, as increased HSF2
transcription accompanied by mRNA stabilization has been shown to
precede the hemin-induced HSF2 protein accumulation in K562 cells
undergoing erythroid differentiation (86
, 87)
. There is
some evidence that HSF2 does not, at least at the transcriptional
level, regulate its own expression, as no HSE has been found in the
mouse HSF2 promoter (96)
. However, this does not exclude
the existence of regulatory elements in the context of a more complete
promoter, since only 
400 bp of the promoter sequence has been
analyzed. By using the protein translation inhibitor cycloheximide, it
has been found that the stabilization of HSF2 mRNA by hemin does not
involve novel synthesis of an HSF2 mRNA binding protein (L. Pirkkala
and L. Sistonen, unpublished results). Closer characterization of the
HSF2 promoter and regulatory factors as well as identification of the
HSF2 mRNA stability determinants are obvious subjects for future
studies.
HSF2 itself is a labile protein, which is targeted for degradation by
the 26S proteasome by a covalent attachment of multiubiquitin (L.
Pirkkala et al., unpublished results). (98)
. This suggests
that the hemin-mediated increase in HSF2 protein could be due to a
stabilizing effect of hemin on some yet unidentified HSF2 interacting
protein(s). Several candidates for HSF2 interacting proteins have been
found using yeast two-hybrid screens, including nucleoporin p62
(99)
and the testis-specific protein HSF2BP
(100)
. Yoshima and co-workers (99)
have
proposed that HSF2 would compete with importin-ß for binding to
nucleoporin p62; consequently, the lack of p62 in testis would
facilitate the nuclear localization of HSF2 and transactivation of heat
shock genes observed in mouse testis (94)
. HSF2 also
interacts with the PR65 subunit of protein phosphatase 2A (PP2A)
(101)
. A region in PR65 previously shown to be involved in
interacting with the catalytic subunit of PP2A is required for binding
to HSF2 (102)
. This suggests that HSF2 would directly
compete with the catalytic subunit for binding to PR65, thereby
modulating PP2A activity. Regulation of PP2A activity by HSF2 could
thus provide a mechanism for cross talk between heat shock gene
expression and PP2A-regulated pathways in the cell (101
, 102)
, suggesting a nontranscriptional function for HSF2. The
proposed specialized functions of HSF2 might depend on HSF2 interacting
proteins, but so far no compelling functional evidence for the
biological roles of HSF2 partner proteins has been obtained.
The 18 amino acid sequence present in the HSF2- isoform, but missing
from the HSF2-ß isoform, has been suggested to confer an increase in
transcriptional activity of HSF2. Using luciferase as a reporter gene
under control of the hsp70 promoter has demonstrated that the mouse
HSF2- isoform is a more potent transcriptional activator than the
HSF2-ß (19)
. Stable overexpression of either mouse
HSF2- or HSF2-ß isoform in human K562 cells has revealed that the
two isoforms indeed have distinct functions in vivo: overexpression of
HSF2-ß inhibits both hemin-induced hsp70 and hsp90 transcription and
erythroid differentiation (46)
. Therefore, HSF2- acts
as a potential activator and HSF2-ß as a suppressor of the
hemin-induced gene expression, and the isoform ratio might be critical
for the human hematopoietic differentiation process. Moreover, HSF2-ß
negatively controls the amount of HSF2 protein accumulated in a cell
(46)
. This type of regulation could be a relevant
mechanism to attenuate the erythroid differentiation process and
induction of heat shock gene expression. However, the mechanisms behind
the differential regulation and function of the two isoforms are still
obscure.
Differential effects of HSF2 and HSF1 on hsp70 were ultimately
suggested to be due to distinct DNA binding patterns at the human hsp70
promoter in vivo on hemin and heat treatments, respectively
(83)
, and to their slightly different binding site
specificities (103
, 104)
. This has been further supported
by the observation that synergistic activation of HSF1 and HSF2 causes
an enhanced hsp70 transcription, possibly by forming heterotrimers,
which correlates with an alteration in the DNA binding pattern in vivo
(86)
. These findings also suggested that HSF1 and HSF2
might regulate the expression of distinct genes. For example, during
erythroid differentiation of K562 cells, HSF2 might be a strong
candidate, either alone or in combination with other transcription
factors, to regulate the expression of erythroid-specific genes
(87)
. One potential HSF2 target gene is thioredoxin (TRX),
the first non-heat shock gene the expression of which is regulated in
an HSF2-dependent manner (105)
. More studies are required
to determine the role of HSF2 in regulation of TRX expression.
Knockout and transgenic organisms reveal unexpected functions for
HSF1
The yeast homologue of vertebrate HSF1 is essential for cell
growth and viability (12
, 13
, 106)
, and this requirement
has been suggested to involve a role in the regulation of basal heat
shock gene expression. Similar to the budding yeast HSF, fruit fly HSF
is essential for the heat shock response in vivo but, unlike yeast HSF,
is dispensable for growth and viability under nonstressful conditions
in adult flies (107)
. The fruit fly HSF is required during
oogenesis and early larval development, as shown by analyzing
phenotypes harboring mutant alleles of the D. melanogaster
HSF gene. However, these two genetically independent functions are not
mediated through Hsp induction (107)
.
The Hsps have vital functions during embryogenesis and in protecting
embryos against the effects of various toxic agents. Spontaneous and
inducible Hsp expression has been demonstrated in mouse early embryonic
cells and embryonal carcinoma cells (108
109
110)
. HSF1 and
HSF2 are specifically expressed at different stages of embryonic
development, as HSF1 is already abundant in oocytes, whereas HSF2 is
undetectable in oocytes and its expression increases during the
preimplantation period (111)
. At morula and blastocyst
stages, the amounts of HSF1 and HSF2 become comparable.
Until the recently reported HSF1 knockout mouse (30)
, no
evidence of the in vivo functions of mammalian HSFs during development
had been provided. HSF1-deficient mice display severe defects in the
chorioallantoic placenta, resulting in increased prenatal lethality.
Still, a small number of mice lacking HSF1 survive, demonstrating that
HSF1 is not obligatory for development (30)
. After birth,
growth of the HSF1 knockout mice is retarded and the females are
sterile (30)
. The requirement of HSF1 as a maternal factor
during early cleavage of mammalian development has recently been
demonstrated elegantly in HSF1 knockout embryos (112)
. The
HSF1 null oocytes exhibit normal morphological appearance, but the
development of fertilized embryos derived from the knockout oocytes is
arrested during preimplantation before the blastocyst stage.
Fertilization of HSF1-deficient oocytes with paternal wild-type HSF1
fails to release the blockage, indicating the essential requirement for
maternal HSF1 in early postfertilization development
(112)
. However, the HSF1 knockout zygotes spontaneously
express the Hsps, showing that zygotic transcriptional activity can
begin without HSF1 (112)
.
HSF1 null mice generated using a heterozygous mother and a homozygous
HSF1-deficient father neither exhibit classical heat shock response nor
acquire thermotolerance, and the HSF1-deficient embryonic fibroblasts
exposed to heat stress die via apoptosis (29
, 30)
. In the
absence of HSF1, exaggerated tumor necrosis factor (TNF-)
production and increased mortality after endotoxin and inflammatory
challenge have been observed (30)
. In an independent
study, inhibition of TNF- transcription by febrile range
temperatures was shown to result from partial activation and binding of
HSF1 to the proximal promoter or 5'-untranslated region of TNF-
(113)
. This finding raises the interesting possibility of
HSF1 acting as a repressor of certain genes, already proposed for
prointerleukin 1ß and c-fos genes (114
, 115)
.
Generation of transgenic mice expressing constitutively active HSF1 in
the testis has revealed an unexpected phenotype: the constitutive
expression of an active form of HSF1 results in a blockage of
spermatogenesis at the pachytene stage, accompanied by an increased
number of apoptotic spermatocytes (116)
. These results are
the opposite of previous studies in mouse and fruit fly that indicate
the requirement of HSF1 for acquisition of thermotolerance by inducing
Hsps and supporting cell survival against thermal stress (29
, 107)
. The mechanism responsible for active HSF1 being sufficient
to induce apoptosis of late pachytene spermatocytes is not known, but
it has been hypothesized that HSF1 may induce or repress target genes
involved in germ cell development or apoptosis (116)
.
Hsp70–2 is normally abundantly expressed at the pachytene stage, but
in mice deficient in Hsp70–2, spermatogenesis is blocked at this stage
and the males are infertile (117)
. Therefore, it is
tempting to speculate that HSF1 might repress hsp70–2 expression in
the pachytene spermatocytes. The spectrum of HSF1 functions has thus
become more diverse, including a requirement for developmental
processes and postnatal growth in addition to regulation of the stress
response under adverse conditions.
HSF1 plays an essential role in the ubiquitin proteolytic pathway
Ubiquitin–proteasome-mediated proteolysis is an essential pathway
for protein degradation. The selective destruction of many short-lived
regulatory proteins as well as damaged proteins is mediated by the 26S
proteasome (for a review, see ref 118
). When the
ubiquitin–proteasome network is down-regulated, certain heat shock
proteins such as Hsp70, among other molecular chaperones, are induced
(49
, 50
, 98
, 119
, 120
; reviewed in ref 118
).
It is therefore of interest to understand the functions of the distinct
HSFs and their roles in regulating Hsp expression during
ubiquitin–proteasome-mediated degradation. To resolve the
contradictory results from various laboratories concerning activation
of the different members of the HSF family (HSF1–3) on proteasome
inhibition (49
, 50
, 98)
, the specific roles of HSF1 and
HSF2 in the regulation of the ubiquitin–proteasome network have
recently been investigated. By using different strategies to abrogate
HSF1 and HSF2 activities, the key regulatory role of HSF1 in the
ubiquitin proteolytic pathway was established in HSF1 null cells by the
capability of exogenous HSF1 to restore the HSF DNA binding activity
and inducible Hsp70 expression upon proteasome inhibition
(31)
. Despite a prominent increase in HSF2 protein, the
transactivating capacity of this factor could not be detected.
It has been speculated that the distinct HSFs could have cooperative
functions. The finding that other HSFs cannot functionally substitute
for HSF1 when the ubiquitin–proteasome-mediated degradation is
inhibited indicates that not all functions of the distinct HSFs are
overlapping. This observation is further supported by studies of HSF1
knockout models (29
, 30)
. The activation of HSF1 upon
disturbances in the degradation pathway leading to enhanced expression
of Hsp70 and other molecular chaperones is an interesting subject for
future studies as to the importance of proteasome-mediated degradation
of a multitude of proteins involved in progression of cell cycle and
tumorigenesis.
Stress-independent functions of HSF3 in the cell cycle
An important role for HSFs unrelated to the heat shock response
originates from studies demonstrating HSF3 activation in unstressed
proliferating cells by direct binding to c-Myb, a transcription factor
involved in hematopoiesis and cell proliferation, through their
respective DNA binding domains (121)
. Since the c-Myb
protein is highly expressed and required for the G1-to-S transition of
the cell cycle simultaneously with expression of Hsp70, it is plausible
that the c-Myb-induced activation of HSF3 may contribute to the cell
cycle-dependent expression of heat shock genes (121)
.
Recently, the c-Myb-HSF3 interaction has been found to be disrupted by
direct binding of the p53 tumor suppressor to HSF3, leading to
proteasome-dependent degradation of c-Myb and down-regulation of hsp70
and hsp90 expression (122)
. Mutated forms of p53 found
in certain tumors are not able to inhibit c-Myb-induced HSF3
activation, suggesting an HSF3-mediated interplay between the c-Myb
proto-oncogene and the p53 tumor suppressor in regulation of the cell
cycle and apoptosis (122)
.
|
CONCLUSIONS AND PERSPECTIVES FOR FUTURE RESEARCH |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...HSFs AND THE HEAT...FUNCTIONS OF HSFs EXPAND...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
. According to the model, the various exogenous and endogenous stimuli
activate distinct members of the HSF family, in most cases leading to
induction of heat shock gene expression. In contrast to the generally
held notion, HSF1 appears not only to be a stress factor, but is also
involved in certain developmental and differentiation-related
processes. The functions of HSF1 not associated with stress have only
been established after the generation of HSF1-deficient cell and animal
models. Therefore, to elucidate the specific role of HSF2, gene
disruption experiments should be most revealing. Moreover, generation
of transgenic and double knockout animals and the cell lines derived
from them should further clarify the potential overlapping functions of
HSFs. Functional cooperativity between the avian HSF1 and HSF3 has
already been described. Unraveling the mechanistic basis of this
cooperation will undoubtedly facilitate understanding of the
interdependent relationship between distinct HSFs. This is particularly
important, since the incapability of HSFs to functionally substitute
for each other might be unique for mammalian cells: in mutated yeast
cells carrying a lethal HSF deletion, both human HSF2 and HSF4, but not
HSF1, can function as a stress-responsive factor and replace the yeast
HSF, thereby rescuing the viability defect.
|
Considering the multiple HSF isoforms yielding increased complexity in their regulatory functions, characterization of the mechanisms responsible for expression and regulation of the distinct isoforms is of great importance. For example, existence of HSF1 isoforms has been reported, but whether they possess differential features, similar to HSF2 and HSF4, has remained unexplored. The avian HSF3 might be a unique member of the HSF family, because only a single mRNA instead of multiple alternatively spliced isoforms, has been found to date. Chromosomal localization of HSFs is likely to shed light on the HSF locus and the neighboring genes. By the aid of large-scale gene analysis using microarray techniques, there remains the exciting possibility of finding new, unexpected HSF target genes.
With regard to the conservation of the heat shock response in
evolution, it is of interest to note that at least two species lacking
an inducible heat shock response have been found. The absence of heat
shock response in a highly cold-adapted, stenothermal Antarctic teleost
fish (Trematomus bernacchii) and a freshwater cnidarian
(Hydra oligactis) might be due to evolutionary adaptation to
stable subzero temperatures without a need to cope with high
temperatures (123
, 124)
. The lack of inducible stress
response could be due to elimination of the heat shock genes or some
components in the gene regulatory pathway during evolution. In the case
of T. bernacchii, the mechanism remains to be elucidated,
but in H. oligactis, the hsp70 mRNA has been shown to be
highly unstable during heat shock (125)
.
From the evolutionary point of view, phylogenetic analysis has
unexpectedly revealed that the zebrafish HSF1 is more homologous to
other vertebrate HSF1s than to the blue gill sunfish HSF
(24)
. This indicates that fish may have several distinct
HSFs or that different fish species have unusually divergent HSFs. As
the multiple HSFs in higher organisms are supposed to differentially
regulate gene expression in response to distinct signals, it is
plausible that fishes have several HSFs, considering the need of many
fish species to tolerate large fluctuations of temperature and other
physicochemical parameters in their natural environment. In further
support of this hypothesis, the complexity of HSFs in plants also
appears to be higher than in other eukaryotic organisms; in the tomato,
for example, HSFs are encoded by a gene family with up to five members
(for a review, see ref 26
). The organism-specific numbers
of HSF family members most likely will soon be established by analyzing
the complete genome sequences.
The ongoing efforts to identify HSF interacting proteins will certainly continue. Although transient protein–protein interactions have traditionally been studied in vitro, it will be most challenging to demonstrate the in vivo significance of these interactions. Understanding the structural basis of phosphorylation-mediated regulation of HSF1 and other HSFs will provide insight into regulation of the stress-sensing mechanisms. Obviously, the ultimate challenge will be to elaborate our detailed knowledge of the HSFs to understand the significance of this transcription factor family in the adaptation to the diverse biological environments to which organisms are exposed.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTBACKGROUNDSTRUCTURAL AND FUNCTIONAL...HSFs AND THE HEAT...FUNCTIONS OF HSFs EXPAND...CONCLUSIONS AND PERSPECTIVES FOR...REFERENCES |
|---|
