(Received for publication, December 9, 1996, and in revised form, January 6, 1997)
From the Departments of Vascular Biology and
Molecular and Experimental Medicine, The Scripps Research
Institute, La Jolla, California 92037 and the § Department
of Medicine, University of California, San Diego,
La Jolla, California 92093-0618
Basic fibroblast growth factor is an important mitogenic and angiogenic factor that stimulates endothelial cell growth and migration. This hormone is not secreted via the classical vesicular pathway, and the identification of intracellular proteins that facilitate its release remains lacking. Transfection and expression of the 27-kDa human heat shock protein in bovine arterial endothelial cells doubles the rate of estrogen-induced basic fibroblast growth factor secretion, preferentially inducing the release of high molecular weight forms of the hormone. The secreted basic fibroblast growth factor is mitogenic to breast adenocarcinoma cells cultured in the conditioned medium obtained from the transfected endothelial cells. In contrast, decreasing the level of the endogenous heat shock protein homolog with an antisense vector markedly decreases basic fibroblast growth factor release. Anti-heat shock protein or anti-basic fibroblast growth factor antibodies co-precipitate both proteins from endothelial cell extracts, demonstrating a direct association between the two proteins. This interaction is likely to be an important step in the mechanism of basic fibroblast growth factor secretion.
The growth of tumors beyond several millimeters in diameter
requires the recruitment of capillaries into the tumor and the establishment of a tumor blood supply (1). Controlling or influencing this process is the paracrine communication between tumor, stromal and
endothelial cells (2, 3). Key regulatory factors in breast tumor
development include insulin-like growth factors, basic fibroblast
growth factor (bFGF1/FGF-2), and
-estradiol, the latter potentiating the mitogenic effects of the
former peptide hormones on the tumor cells (4, 5). Breast tumor bFGF is
predominantly of paracrine origin (6, 7); thus factors facilitating the
release of bFGF from endothelial cell and stromal cells facilitate
tumor growth. In addition, since bFGF regulates endothelial cell growth
and migration, factors that induce the release of bFGF would also
facilitate tumor angiogenesis (8).
The mechanism by which bFGF is released from cells remains to be
elucidated. bFGF is actually four different translation products of a
single message species exhibiting apparent molecular masses of 18, 22, 22.5, and 24 kDa (9-11). The translation of the 18-kDa species begins
at a classical AUG start site whereas the higher molecular weight
species (HMW bFGF) are initiated at CUG codons 5 to the AUG start site
(11). Neither the 18-kDa nor the HMW bFGF species possess a signal
sequence and therefore are not secreted via the classical vesicular
pathway involving the endoplasmic reticulum (ER) and Golgi apparatus.
Indeed, agents that block trafficking through the Golgi-ER pathway do
not inhibit the release of bFGF from cells, and novel pathways for bFGF
release have been suggested (12, 13). The identification of proteins
involved in the release of bFGF remains lacking however. Data generated in this report suggest that the 27-kDa heat shock protein (HSP27) is
involved in the non-lytic release of bFGF.
Endothelial cells express a basal level of the HSP27, which is further enhanced by transcriptional up-regulation in response to estrogens (14). Higher levels of HSP27 expression have been shown to affect microfilament assembly and morphology as well as cell growth (14, 15). For example, we have demonstrated that expression of human HSP27 in bovine arterial endothelial cells (BAECs) via transfection results in a 2-3-fold enhancement in the rate of cell growth (14). These characteristics make HSP27 a likely control point in angiogenic processes where estrogens play key regulatory roles, e.g. breast tumor angiogenesis (2, 3). HSP27 has several demonstrable functions that are attributed to its ability to act as a molecular chaperon, a property it shares with the highly homologous lens crystallins (16, 17) and other HSPs (18, 19). Putative ligands for the chaperoning activity of HSP27 may include steroid-receptors (18, 20-22) and nascent or partially denatured proteins (23, 24).
To begin to investigate a role for HSP27 in breast tumor angiogenesis,
stably transfected BAECs expressing human HSP27 or control BAECs (14)
were co-cultured with the breast adenocarcinoma MCF-7 cells and the
growth of each cell type was measured. We consistently found that the
culture of the HSP27 BAECs with the tumor cells resulted in increased
tumor cell growth if -estradiol was included in the cultures. This
was true even if the MCF-7 cells exhibited decreased estrogen receptor
levels and were no longer capable of exhibiting a mitogenic response to
-estradiol alone (i.e. were estrogen unresponsive).
Conditioned media obtained from
-estradiol-treated HSP27 BAECs also
induced MCF-7 cell growth, indicating that a soluble factor was
released from the HSP27 BAECs. An anti-bFGF blocking antibody inhibited
the potentiated MCF-7 growth, and analysis of HSP27 BAEC-conditioned
media confirmed that bFGF was the paracrine factor released from the
HSP27 BAECs. Reducing the expression of the endogenous HSP27 homolog
(HSP25) by infection with HSP27 antisense adenovirus resulted in
decreased bFGF secretion. Immunoprecipitation using either anti-HSP27
or anti-bFGF antibodies co-precipitated the two proteins, suggesting that a direct interaction exists between HSP27 and bFGF. These data
suggest that HSP27 may act as a chaperon of bFGF, facilitating bFGF
release from endothelial cells.
Low passage bovine pulmonary arterial endothelial cells were a generous gift of Dr. W. Laug, Children's Hospital, Los Angeles, CA. All cell culture reagents were obtained from Bio Whittaker, Inc. except where otherwise noted. Cells were cultured under 5% CO2 in Dulbecco's modified Eagle's medium (DMEM) containing 25 mM HEPES and supplemented with 10% fetal calf serum (Intergen) and 1 mM each of sodium pyruvate, penicillin, streptomycin, and nonessential amino acids. Cells were plated at a density of 0.7-2 × 104 cells/cm2 and passaged when confluent (approximately 1 × 105/cm2).
Plasmids and Stable BAEC TransfectantsClones of stably transfected BAECs expressing human HSP27 were generated as described (14). Briefly, pHS2711, a plasmid containing a genomic clone of the human HSP27 gene (16) or the vector plasmid (Bluescript) were transfected into low passage BAECs using cationic lipid (LipofectAMINE, Life Technologies, Inc.). The neomycin resistance-carrying plasmid, pCDM8neo, was co-tranfected for the purpose of selecting transfected cells from non-transfected cells. Cells were cultured in the presence of 700 µg/ml Geneticin (G418-sulfate, Life Technologies, Inc.).
Culture of MCF-7 CellsThe breast adenocarcinoma cell line
MCF-7 was obtained from the American Type Culture Collection at passage
149 (ATCC HTB-22). This cell line is estrogen receptor positive and
exhibits a mitogenic response to -estradiol. Cells were maintained
in minimal essential medium containing 10% fetal calf serum, 1 mM sodium pyruvate, and 10 µg/ml bovine pancreas insulin.
To generate estrogen-unresponsive cells (i.e. cells that do
not exhibit a mitogenic response to
-estradiol), MCF-7 cells were
subcultured into Phenol Red-free minimal essential medium containing
sodium pyruvate, insulin, and 10% steroid-depleted fetal calf serum
prepared by dextran/charcoal stripping (25). These cultures represent
non-clonal populations of cells selected for the ability to grow in
steroid-deficient media that express reduced estrogen receptor
levels.
Triplicate wells in 12-well plates were
seeded with 2.5 × 104 HSP27 BAECs (14) or vector
control BAECs in Dulbecco's modified Eagle's medium with 10% fetal
calf serum. After 24 h in culture, the BAECs were washed three
times with Phenol Red-free and serum-free minimal essential medium
containing 0.2% lactalbumin hydrolysate (assay media). Assay media,
with or without 100 nM -estradiol was then placed onto
the BAECs or into empty wells for the generation of mock conditioned
media for the No BAEC controls. The endothelial cells were then
cultured for 18 h, after which the conditioned media were
collected. 2.5 × 104 MCF-7 cells were seeded into 12 wells and cultured for 1-2 days prior to an experiment. The MCF-7
cells were washed three times in assay media, and the BAEC-conditioned
media, to which tritiated methyl thymidine (1 µCi/ml, Amersham Corp.)
was added, was placed on the MCF-7 cells. The MCF-7 cells were cultured
for 18-22 h, and the amount of tritiated thymidine incorporated into
the MCF-7 cells was determined by trichloroacetic acid precipitation as described (14). In experiments testing the effect of antibodies on
growth, 20 µg/ml of either anti-bFGF (Sigma, clone FB-8) or an
irrelevant isotype matched monoclonal antibody was added to the
BAEC-conditioned media and incubated for 30 min prior to the addition
to the MCF-7 cells.
T-75 flasks of transfected HSP27 BAEC
or vector control BAEC, at approximately 75% confluence, were placed
in assay media (6 ml/flask) with or without 100 nM
-estradiol and cultured for 18-24 h. The media were collected and
the cells were washed three times in Dulbecco's phosphate buffered
saline and then lysed with 300-500 µl of 0.5% Triton X-100 in 10 mM imidazole, 40 mM KCl and 10 mM
EGTA (plus 10 mM benzamidine, 1 mM
phenylmethylsulfonyl fluoride, and 100 µg/ml leupeptin). The lysate
was cleared by centrifugation at 14000 × g for 5 min.
Protein content of the Triton lysates was determined by the
bicinchoninic acid assay (Pierce). 75 µl of Triton lysate was
solubilized immediately in 4 X reducing SDS-PAGE sample buffer (26).
The remaining material on the flasks (the Triton X-100-insoluble
material) was washed 3 times in lysis buffer and then solubilized in
reducing SDS-PAGE sample buffer. The conditioned media were cleared of
cellular material by centrifugation at 800 × g, and
5.5 mls transferred into 5.5 ml of ice-cold 12% trichloroacetic acid.
After an incubation on ice for 30 min, the precipitated material was
pelleted by centrifugation at 46000 × g for 1 h
at 4 °C. The pelleted material was resuspended in 0.75 ml of cold
5% TCA and transferred to microcentrifuge tubes. An additional wash of
the centrifuge tubes with 0.75 ml of 5% trichloroacetic acid was
combined with the resuspended precipitated material and centrifuged at
14000 × g for 30 min at 4 °C. The resulting pellets
were allowed to air dry and then solubilized in reducing SDS-PAGE
sample buffer. The conditioned media samples were neutralized with NaOH
and all the samples applied to a 12% w/v acrylamide gel. Equivalent
loading of each type of sample was obtained using the bicinchoninic
acid protein assay results to calculate the volume of sample to add.
Also loaded were serially diluted (2-fold) samples of rbFGF, beginning
with 15 ng of rbFGF for the purpose of generating a standard curve.
After electrophoresis, the proteins were transferred to nitrocellulose
and stained with the blocking anti-bFGF antibody using horseradish
peroxidase-conjugated donkey secondary antibody (Jackson ImmunoResearch
Laboratories, Inc.) and the enhanced chemiluminescence reagent (ECL,
Amersham Corp.). Densitometric analysis was performed using the Eagle
Eye II computerized digital camera and Eagle Sight 3.0 software
(Stratagene).
Adenovirus constructs were generated
from replication-deficient adenovirus in which the EIA and EIB regions
had been deleted. Sequence representing human HSP27 cDNA antisense
was inserted behind a cytomegalovirus promoter which drove expression.
Control adenovirus was the replication-deficient adenovirus lacking
insert. HSP27 antisense or control virus was added to confluent T-75
flasks of a vector-transfected control clone at a concentration of
108 virons/ml of Dulbecco's modified Eagle's medium
containing 2% fetal calf serum for 1 day. The media were removed, and
the cells were washed three times in assay media. Assay media
containing 100 nM -estradiol were added, and the cells
were cultured for an additional day. Samples were prepared and analyzed
for bFGF content by the immunoblot analysis described above. In
addition, the blot was reprobed with a rabbit polyclonal antibody
raised against murine HSP25 (StressGen) and demonstrated itself to
cross-react with bovine HSP25 (14).
Triton lysates of HSP27 BAECs were prepared as described above and subjected to immunoprecipitation with anti-HSP27-Sepharose (G3.1, StressGen), anti-bFGF-Sepharose (FB-8, Sigma), or non-immune mouse IgG-Sepharose. The affinity matrices were prepared per manufacturer instructions using cyanogen bromide-activated Sepharose CL-4B (Pharmacia) with coupling performed at a ratio 1.0 mg of antibody per 1.0 ml of swelled gel. The lysates were precleared by incubation with non-immune mouse IgG-Sepharose (100 µl of beads/500 µl of lysate) for 1 h at room temperature and then incubated with 100 µl of anti-HSP27-, anti-bFGF-, or non-immune mouse IgG-conjugated Sepharose CL-4B for 2 h at room temperature. The beads were washed three times with 1.5 ml of lysis buffer and then eluted with 0.1 M glycine, pH 3.0. The eluates were neutralized and subjected to SDS-PAGE and immunoblot analysis along with the starting material. The blots were first probed with the anti-bFGF antibody and then probed with the anti-HSP27 monoclonal antibody.
To demonstrate a
role for HSP27 in breast tumor angiogenesis, breast adenocarcinoma
cells (MCF-7 cells) were cultured in Transwell inserts above growing
BAECs. We consistently found that the growth of the MCF-7 cells could
be enhanced if these cells were either cultured with BAECs expressing
human HSP27 (HSP27 BAECs) or cultured in the media conditioned by
growing HSP27 BAECs, if the HSP27 BAECs were first treated with
-estradiol (Fig. 1). This mitogenic response of the
MCF-7 cells was absolutely dependent on treatment of the HSP27 BAECs
with
-estradiol. In the absence of
-estradiol, HSP27 BAEC culture
had no effect on MCF-7 growth. For example, the culture of
estrogen-unresponsive MCF-7 cells (see "Discussion") in the
conditioned media obtained from
-estradiol-treated HSP27 BAECs
generated thymidine incorporation 1.54 ± 0.2 times that of
control MCF-7 cultures (compare filled bars in Fig. 1,
mean ± S.D., n = 5). Importantly, the addition of
fresh
-estradiol to the conditioned media obtained from HSP27 BAECs
not treated with
-estradiol failed to elicit a mitogenic response.
In addition, MCF-7 culture in mock conditioned media containing
-estradiol (i.e. media cultured in the absence of cells)
also failed to enhance growth (Fig. 1). This demonstrates that the
estrogen-dependence of the enhanced MCF-7 growth is a consequence of
-estradiol affecting the HSP27 BAECs and not the MCF-7 cells. The
enhanced MCF-7 growth was absolutely dependent on HSP27 expression in
the BAECs, since co-culture with vector-transfected control cells or
culture in their conditioned media did not result in enhanced thymidine
incorporation into the MCF-7 cells. Thus, both enhanced HSP27 levels
and
-estradiol treatment are required for the stimulation of MCF-7
cells. These data suggest that
-estradiol releases a paracrine
factor from HSP27 BAECs and that HSP27 somehow facilitates this
release.
Stimulation of both tumor
growth and tumor angiogenesis is partly dependent on paracrine
communication (3) such as that suggested by the induction of MCF-7 cell
growth by -estradiol treatment of HSP27 BAECs. bFGF is a potent
mitogen of endothelial cells that synthesize and store the growth
factor (27). To determine if the paracrine factor in the HSP27
BAEC-conditioned media responsible for the potentiated MCF-7 growth was
indeed bFGF, 20 µg/ml of a function blocking anti-bFGF monoclonal
antibody was included in the HSP27 BAEC-conditioned media. As shown in
Fig. 1, this antibody completely abrogated the enhanced tritiated
thymidine incorporation, whereas an isotype-matched control antibody
did not. These data indicate that bFGF is the paracrine mitogenic factor released from the estrogen-treated HSP27 BAECs.
To determine if the enhanced MCF-7 growth is correlative with increased
bFGF secretion from the HSP27 BAECS, HSP27-expressing and control BAECs
were placed in serum-free assay media (defined in Fig. 1), with or
without -estradiol, and cultured for 18 h. The conditioned
media were cleared of cellular material and then precipitated with
trichloroacetic acid, solubilized for SDS-PAGE, and subjected to
immunoblot analysis using the monoclonal anti-bFGF antibody. Also
included on the gels were Triton X-100 lysates of the cultures, the
Triton-insoluble material (solubilized by SDS extraction), and, for the
purpose of generating a standard curve, titrating amounts of rbFGF. In
a representative experiment (Fig. 2, panel
A), HSP27 BAECs and vector control BAECs, in the absence of
-estradiol, secreted 4.8 and 5.1 ng of bFGF/mg of protein in the
Triton extract, respectively. Upon treatment of the transfectants with
100 nM
-estradiol, the amount of bFGF released by the
HSP27 BAECs increased to 10.2 ng of bFGF/mg of Triton lysate protein,
with increases in all the bFGF forms evident. The amount of bFGF in the
conditioned media of the control BAECs treated with
-estradiol in
Fig. 2 was 5.2 ng of bFGF/mg of Triton lysate protein. Estrogen
treatment of the HSP27 BAECs resulted in an average 2.4 ± 0.2-fold enhancement (n = 4) of bFGF released into the
media, increasing the media concentration from 240-320 pg/ml to
512-730 pg/ml. The higher concentration range, which supported a
mitogenic response in our culture system, is above the ED50
of bFGF for other cell types in culture (e.g. 3T3
fibroblasts (11)). In contrast, the control BAECs did not release
additional bFGF upon
-estradiol treatment (+estrogen/
estrogen = 1.1 ± 0.07, n = 4).
Although the transfectants released equivalent amounts of bFGF, HSP27
BAECs preferentially secreted HMW bFGF, with little or undetectable
amounts of 18-kDa bFGF released. As demonstrated in Fig. 2, the HSP27
BAECs secrete a disproportionate amount of HMW bFGF. In the absence of
-estradiol treatment, the average percentage of the secreted bFGF
representing HMW bFGF was 94.5 ± 6.5% and 62 ± 8% for
cultures of HSP27 BAECs and vector control BAECs, respectively
(n = 4). Upon
-estradiol treatment, the average relative amount of HMW bFGF in the HSP27 BAEC-conditioned media dropped
to 76 ± 6.8% whereas the relative amount of HMW bFGF released by
the control cells did not change significantly. The average absolute
amount (ng/mg of cellular protein) of HMW bFGF in the HSP27
BAEC-conditioned media increased 2-fold whereas 18-kDa bFGF levels
increased at least 3-fold. The greater release of the 18-kDa bFGF
accounts for the drop in the relative percentage of HMW bFGF in the
HSP27 BAEC conditioned media.
In addition to probing the immunoblots with anti-bFGF, the blots were
reprobed with antibody specific for glucose-6-phosphate dehydrogenase
as a measure of lytic and non-directed release. Importantly, estrogen
treatment of the BAECs had little effect on the percentage of cellular
glucose-6-phosphate dehydrogenase released from the cells. Densiometric
analysis determined that BAECs released an average of 3 ± 1%
(n = 3) of the total cellular glucose-6-phosphate
dehydrogenase regardless of HSP27 expression or the presence or absence
of estrogen in the culture media. In contrast, the percentage of total
culture bFGF that was secreted into the media increased from 9.1 ± 1.4% to 25 ± 1.4% upon -estradiol treatment of HSP27 BAEC
cells. Given this difference and the fact the HSP27-expressing clones
exhibited the same viability as control clones with or without estrogen
treatment (92 ± 4%, n = 6), it is apparent that
the estrogen-induced release of bFGF from HSP27 BAECs is not due to
cell lysis.
To determine if the enhanced release of bFGF is the result of a general enhancement of protein secretion, cultures of HSP27 BAECs and vector control BAECs were cultured in media containing [35S]methionine for 5 h and then placed in serum-free assay media for 18 h. The media were trichloroacetic acid precipitated, and Triton extracts were prepared from the cells. The amount of radioactivity incorporated into the precipitated proteins was determined and found to be 990 and 1040 cpm/mg of cellular protein for the HSP27 BAECs treated with estrogen and untreated, respectively. Thus, there was no general increase in the level of protein secretion as measured in this manner. Vector-transfected BAEC-secreted proteins had specific activities of 820 and 890 cpm/mg Triton lysate protein for estrogen-treated and non-treated cells, respectively.
Reduction of Endogenous HSP25 via Infection with HSP27 Antisense Adenovirus Reduces the Amount of bFGF Secreted into the MediaIncreased levels of HSP27 in endothelial cells are thus
correlative with increased estrogen-induced bFGF release. To determine if reduced levels of HSP27 would abrogate this response, BAEC cells
were infected with either non-replicating control or HSP27-antisense adenovirus vectors. The expression of the bovine HSP27 homolog (HSP25)
was dramatically decreased in the antisense-infected cells (Fig.
3). Infection of 107 virus
particles/cm2 of culture area for 2 days reduces HSP25
levels by a third to one-half as measured by densitometry of
immunoblots using a rabbit polyclonal antibody raised against murine
HSP25 (Fig. 3). Infection with control virus had no effect on HSP25
levels. The viability of all the infected BAEC cultures was 96 ± 1.3% (n = 4), and neither virus affected overall
secretion of proteins into the conditioned media. The amount of
radioactivity of [35S]methionine-labeled proteins
secreted into the media were determined to be 13,000 and 15,200 cpm/mg
of cellular protein for antisense- and control-infected BAECs,
respectively. As shown in Fig. 3, infection of the antisense adenovirus
dramatically reduced the level of bFGF in the culture media, whereas
the control virus had no effect. No quantitative differences were noted
in the amount of bFGF present in either the Triton-soluble or
Triton-insoluble cellular fractions (Fig. 3). It thus appears that, not
only does HSP27 expression correlate with bFGF release, but HSP27 may
be necessary for bFGF release.
bFGF Co-precipitates with HSP27 by Immunoaffinity Chromatography
To begin to elucidate the role of HSP27 in bFGF
release, immunoprecipitations of cell extracts using immobilized
anti-HSP27 and anti-bFGF monoclonal antibodies were performed. Eluted
material was then subjected to immunoblot analyses using the same
antibodies. As demonstrated in Fig. 4, anti-HSP27
antibody precipitated HSP27 (panel C, lane 7) and
anti-bFGF immunoreactive material (panel B, lanes
3 and 4). The latter was overwhelmingly HMW bFGF,
whereas the anti-bFGF precipitated all forms (panel B,
lane 5), with the main form representing the 18-kDa species.
Anti-bFGF chromatography precipitated HSP27 (panel C,
lane 8) along with cellular bFGF. Since the antibodies do
not recognize a common determinant, as demonstrated by the failure of
the anti-bFGF antibody to stain HSP27 in cell extracts (panel
A, lane 1) and failure of the anti-HSP27 antibody to
stain bFGF (panel A, lane 2), the
co-precipitation of both proteins is the result of an interaction
between the HSP27 and bFGF.
The small molecular weight heat shock protein (HSP27) is an
estrogen-responsive protein, which has profound effects on cell growth
(14) and microfilament cytoskeletal dynamics
(28).2 Expression of HSP27 in estrogen
responsive breast tumors (i.e. tumors requiring
-estradiol for growth and therefore susceptible to anti-estrogen
therapy) is correlative with the growth of these tumors (29, 30). The
exact mechanisms by which enhanced HSP27 levels impart a phenotype of
accelerated growth have not been firmly established.
The data presented in this report demonstrate that enhanced HSP27 expression facilitates the release of bFGF from endothelial cells (Fig. 2). This secreted bFGF, which was predominantly HMW bFGF, was mitogenic to breast tumor cells (Fig. 1). The 18-kDa and HMW bFGF have different capacities to modulate cellular function and may work through different binding mechanisms and intracellular signals (31). For example, HMW bFGF can support growth in low serum media at levels at which the 18-kDa bFGF form does not (31). This is consistent with the results of the assays detailed in this report in which MCF-7 growth is observed in the serum-free media obtained from the HSP27 BAECs (Fig. 1). The difference in cellular response to the two forms of bFGF may be due to unique signals induced from the binding of HMW bFGF to an intracellular or nuclear membrane receptor or from direct regulation of transcriptional activity in the nucleus (32, 33).
The MCF-7 cell line obtained for this study expresses an estrogen
receptor and exhibits a mitogenic response to -estradiol. Co-culture
of these cells with the HSP27 BAECs accelerated the growth of the
estrogen-responsive cells induced by
-estradiol alone (data not
shown). Subculture of the MCF-7 cells in steroid-deficient media
selected for cells that can grow in the absence of estrogens and
generated cultures that exhibited decreased expression of the estrogen
receptor. These cells no longer exhibited a mitogenic response to
-estradiol alone (Fig. 1) and thus roughly approximated estrogen-unresponsive breast tumors. Use of these cells in this study
abrogated the estrogen-induced autocrine effects of
-estradiol and
allowed the direct test of the effects of
-estradiol on the paracrine activity of the BAECs (Fig. 1).
The paracrine communication responsible for the results depicted in
Fig. 1 may have relevance to the pathogenesis of breast adenocarcinomas. Endothelial cells isolated from a wide range of tissue
and vessel types, including the BAECs employed in this study, possess
demonstrable estrogen receptors and are capable of exhibiting genomic
responses to -estradiol (34-36). For example, culture of the BAECs
employed in this study in the presence of
-estradiol for several
days induced enhanced HSP25 expression (14). Thus, the potential exists
for the capillary endothelial cells involved in the vascularization of
breast tumors to respond in a similar fashion and increase HSP27
expression, which then would facilitate bFGF release. The growth of
estrogen-unresponsive tumors, therefore, would still be affected by
-estradiol present in the circulation even though the tumors are
refractory to the steroid itself.
Secretion of bFGF has been postulated to occur via pathways that are independent of the ER-Golgi vesicular pathway (12, 13). Florkiewicz et al. (12) have demonstrated that the 18-kDa form of bFGF is translocated through the plasma membrane in an energy-dependent manner. This translocation could be arrested if the transmembrane domain of an integral membrane protein was affixed to the carboxyl terminus of the bFGF. A putative role for HSP27 in this model is suggested by the fact that a significant portion of cellular HSP27 is associated with the plasma membrane.2 The data presented in this report, however, demonstrate that HSP27 associates predominantly with HMW bFGF, whereas the model described is specific for the 18-kDa form of bFGF. Mignatti et al. (13) demonstrated that agents which stimulate exocytosis (e.g. treatment with the ionophore A23187) facilitate bFGF release and that conditions which inhibit exocytosis (e.g. culture at 18 °C or with methylamine) inhibited the release of bFGF which stimulated cellular migration in an autocrine fashion. Since inhibitors of ER-Golgi protein trafficking failed to inhibit the autocrine induction of migration, it was concluded that bFGF is released via exocytosis that is independent of the ER-Golgi pathway. A role for HSP27 in this putative mechanism of bFGF release is suggested by the fact that overexpression of HSP27 in fibroblasts has been demonstrated to increase the rate of pinocytotic activity (15). This phenomenum was attributed to the role of HSP27 as a modulator of F-actin dynamics at the plasma membrane (15). If similar mechanisms are involved in exocytosis, increased HSP27 expression may facilitate the release of bFGF via this pathway.
The HSP27-dependent release of bFGF release requires
-estradiol treatment of the HSP27 BAECs. The nature of this
estrogen-dependence is as yet unknown but is not due to greater bFGF
(Fig. 2) or HSP27 expression (data not shown). The latter is consistent
with the report demonstrating that the induction of HSP27 production by
-estradiol requires 72 h (14).
-Estradiol exerts a myriad of
receptor- and non-receptor-mediated events that result in both genomic
and non-genomic changes in the cell. For example, the BAECs employed in
this study have a demonstrable estrogen receptor, the activation of
which could lead to altered transcriptional activity and the
up-regulation of specific genes (37). These gene products may influence
HSP27-mediated bFGF release. Another of the many mechanisms by which
estrogen may induce HSP27-dependent release is the
possibility that activated estrogen receptor may directly associate
with the HSP27-bFGF complex, acting as part of the putative chaperon
machinery. Interactions of peptide hormones with the estrogen receptor
have been proposed as one explanation for the peptide hormone induction
of estrogen receptor-dependent transcription (38). Whatever
the mechanism, the association of HSP27 with bFGF is likely to be an
important step in the release of bFGF and/or the estrogen dependence of
the release.
Plasmid containing the genomic clone of human HSP27 was a gift of L. Weber, University of Nevada, Reno, Nevada.