Basolateral Membrane-associated 27-kDa Heat Shock Protein and Microfilament Polymerization

The in vivo activity of the 27-kDa heat shock protein, a barbed-end microfilament capping protein, may be localized to the plasma membrane. To investigate this pu- tative association, bovine endothelial cells expressing the human wild type or a mutant nonphosphorylatable 27-kDa heat shock protein were subjected to subcellular fractionation and immunoblot analysis. The 25-kDa en- dogenous bovine homolog and both exogenous gene products partitioned with cytosolic or plasma mem- brane components, indicating that phosphorylation is not required for membrane association. Phorbol ester treatment resulted in phosphorylation of only mem-brane-associated 25-kDa and wild type 27-kDa heat shock protein and did not induce redistribution. In a second fractionation protocol, streptavidin-agarose pre- cipitation of extracts prepared from cells biotinylated at either the apical or basal surface localized membrane 25- and 27-kDa heat shock protein exclusively to the basolateral surface. Stimulation of transfectants ex- pressing the wild type 27-kDa heat shock protein resulted in its phosphorylation and a doubling in the amount of membrane-associated F-actin precipitated, whereas the mutant protein decreased the amount of F-actin precipitated. These data suggest that mem-brane-associated 25- and 27-kDa heat shock proteins in- hibit the generation of basolateral microfilaments and that phosphorylation releases this inhibition. Endothelial cell F-actin exists a dynamic and microfilament tightly regulated by a of spatial and temporally regulated microfilament modulating proteins (1–4). One protein, a barbed-end luminographic and Autoradiographic Signals— Densitometry of chemi- luminographs and autoradiographs of immunoblots was performed with a Stratagene Eagle Eye II computerized digital camera and Eagle Sight 2.0 software. Densitometric data are generated as integrated density units. Chemiluminographic signals for immunoreactive mate- rial generated in the experiments presented in this report were within a linear range determined by analyzing serially diluted Triton lysates of BAECs or wtHSP27-expressing BAECs (Fig. 1, panels A and B ) with the primary antibodies recognizing HSP25 (rabbit antisera raised to mu- rine HSP25, StressGen, Inc.), HSP27 (the monoclonal antibody G3.1, StressGen, Inc.), and (cid:98) -actin (clone AC-15, Sigma). Likewise, autora- diographic signal intensities were within the linear range of exposures as defined by densitometric analysis of the autoradiograph of an immu- noblot on which serial dilutions of a Triton X-100 lysate prepared from radiolabeled wtHSP27-BAECs were analyzed (Fig. 1, panels A and B ). For fractionation experiments of radiophosphorylated BAECs, the autoradiographic integrated densities (Rad) of cytosolic HSP25/27 and membrane-associated HSP25/27 were determined for quiescent and PMA-stimulated BAECs. These values were divided by the integrated densitometric values of the immunostained HSP25/27 bands on chemi-luminographs ( i.e. antigen intensity, Ag). The resultant Rad/Ag ratios were averaged and presented with the calculated standard deviations.

The in vivo activity of the 27-kDa heat shock protein, a barbed-end microfilament capping protein, may be localized to the plasma membrane. To investigate this putative association, bovine endothelial cells expressing the human wild type or a mutant nonphosphorylatable 27-kDa heat shock protein were subjected to subcellular fractionation and immunoblot analysis. The 25-kDa endogenous bovine homolog and both exogenous gene products partitioned with cytosolic or plasma membrane components, indicating that phosphorylation is not required for membrane association. Phorbol ester treatment resulted in phosphorylation of only membrane-associated 25-kDa and wild type 27-kDa heat shock protein and did not induce redistribution. In a second fractionation protocol, streptavidin-agarose precipitation of extracts prepared from cells biotinylated at either the apical or basal surface localized membrane 25-and 27-kDa heat shock protein exclusively to the basolateral surface. Stimulation of transfectants expressing the wild type 27-kDa heat shock protein resulted in its phosphorylation and a doubling in the amount of membrane-associated F-actin precipitated, whereas the mutant protein decreased the amount of F-actin precipitated. These data suggest that membrane-associated 25-and 27-kDa heat shock proteins inhibit the generation of basolateral microfilaments and that phosphorylation releases this inhibition.
Endothelial cell F-actin exists as a dynamic and responsive microfilament cytoskeleton tightly regulated by a number of spatial and temporally regulated microfilament modulating proteins (1)(2)(3)(4). One such protein, a barbed-end filament capping protein that is inhibited by its phosphorylation, is the 27-kDa heat shock protein (HSP27) 1 (5)(6)(7). The in vivo activity of HSP27 has been inferred from the effects of overexpression of the protein in cultured fibroblasts (8,9). Expression of human HSP27 in these cells stabilized cortical F-actin microfilaments that normally disaggregate in response to heat shock, acute cytochalasin D treatment, or oxidative stress (10,11). In unstressed cells, overexpression of HSP27 leads to increased pinocytotic activity and membrane ruffling (9), processes dependent on a dynamic membrane-associated microfilament cytoskeleton (12,13). Whether the in vivo changes that result from the enhanced expression of HSP27 are due to the microfilament capping activity demonstrated for HSP27 in vitro has yet to be firmly established, however.
HSP27 is phosphorylated by kinase activity induced by a variety of stress, cytokine and mitogenic stimuli (11, 14 -17). The control of HSP27 in vivo activity has been linked to its phosphorylation state, a conclusion based on results demonstrating that a non-phosphorylatable mutant (mu) HSP27, expressed in cultured cells, failed to promote the same effects promoted by the wild type (wt) HSP27 (8,9). Evidence has been presented, however, that contradicts the reliance of HSP27 activity on its phosphorylation. For example, Knauf et al. (18) have found that overexpression of muHSP27 confers the same thermo-resistant traits as the wild type protein and that the in vitro ability of muHSP27 to act as a molecular chaperone is also not affected by the lack of phosphorylation.
To investigate a putative role of HSP27 in controlling the endothelial microfilament cytoskeleton, we have generated stable transfectants of bovine arterial endothelial cells (BAECs) expressing the human HSP27 gene (wtHSP27) (19) or a mutagenized gene product (muHSP27) lacking the three known phosphorylation sites (9). The transfected BAECs constitutively express the exogenous gene products in addition to the bovine HSP27 homolog, HSP25 (20,21). Bovine HSP25 and human wtHSP27, but not muHSP27, are phosphorylated to similar levels by kinase activity induced in response to chemical (phorbol 12-myristate 13-acetate, PMA) or mechanical (laminar flow) stimuli (21). The transfected clones express 20 -40 ng of exogenous HSP27/20 g of total transfectant protein, which has no effect on the expression of the endogenous HSP25 (21). In addition, expression of either exogenous gene product does not effect the basal or stimulated level of HSP25 phosphorylation (21). Thus, at the level of expression generated in the BAEC clones, the exogenous gene products simply add to the cellular pool of HSP25. This addition can induce phenotypic alterations. For example, expression of wtHSP27 induces accelerated growth and culture senescence of the BAECs (20).
Expression of exogenous HSP27 in fibroblasts results in increased cortical actin structure and an enhancement of activities dependent on the membrane-associated microfilament cytoskeleton such as pinocytosis and membrane ruffling (9 -11). These data suggest that a component of cellular HSP27 activity is focused to the membrane. To investigate this possibility, we performed subcellular fractionation on BAECs and transfected BAECs expressing human wtHSP27 and muHSP27. In this report, we demonstrate that a significant portion of cellular HSP25/27 fractionates with plasma membrane components, that phosphorylation is not required for this localization, and that this subpopulation of HSP25/27 is a substrate for kinase activity induced by brief treatment of the cells with phorbol ester. Using a second fractionation protocol, we further demonstrate that a portion of the membrane-associated HSP25/27 localizes to the basolateral membrane where it is substrate for HSP27 kinase activity. Concurrent with phorbol ester-induced HSP25/27 phosphorylation is the generation of additional membrane-associated F-actin. Expression of muHSP27 (which cannot be phosphorylated) inhibits the generation of additional membrane-associated F-actin. These data are consistent with the demonstrated in vitro functioning of HSP27 (i.e. that nonphosphorylated HSP27 inhibits actin polymerization and the phosphorylation abrogates this activity) (7). These data suggest that membrane-associated HSP25/27 may play a role in the regulation basolateral membrane-associated microfilament dynamics.
Generation and Culture of BAEC Transfectants Expressing HSP27-All culture reagents, except where otherwise noted were obtained from BioWhittaker, Inc. Stable clonal cells lines of BAECs were prepared with modifications of what has been previously described (20). Briefly, low passage BAECs, isolated from pulmonary arteries, were transfected with a plasmid containing a genomic clone of human HSP27 (8), a plasmid containing that clone subjected to site-directed mutagenesis in which the codons for the principal sites of HSP27 phosphorylation (Ser 15 , Ser 78 , and Ser 82 ) were altered to encode glycine residues (9) or vector plasmid without insert (pBluescript, pBS) using the cationic lipid LipofectAMINE (Life Technologies, Inc.). The cells were plated at a density of 4 ϫ 10 4 /cm 2 in 96-and 6-well plates. A plasmid conferring neomycin resistance, pCDM8 neo , was co-transfected at a one-tenth molar ratio for the purpose of antibiotic selection. Transfected populations were grown for 72 h at 37°C under 5% CO 2 in Dulbecco's modified Eagle's media containing 25 mM HEPES and 4.0 g/liter glucose supplemented with 1 mM sodium pyruvate, 1 mM nonessential amino acids, 1 mM penicillin/streptomycin, and 10% fetal calf sera (Intergen Corp.). The cells grown in the 6-well plates were screened for transfection efficiency via immunoblot analysis and immunofluorescence microscopy using the anti-HSP27 monoclonal antibody G3.1 (StressGen, Inc.). Selection media, culture media containing 700 g/ml geneticin (Life Technologies, Inc.), was then added to the 96-well plates. Media were changed every 3 days as non-transfected cells died. After 2 weeks of culture in selection media, the 96-well plates were viewed via phase contrast microscopy, and the wells containing single colonies were noted. These clones were subcultured in selection media. Following this protocol, a single transfection performed in three 96-well plates generated 15-25 clones. Clones developed from several different transfections were used for this study and demonstrated by densitometric analysis of immunoblots on which recombinant human HSP27 (Stress-Gen, Inc.) was analyzed to contain approximately 20 -40 ng of exogenous HSP27/20 g of total cellular protein.
Labeling with 32 P i -Cells were cultured in complete media prepared with phosphate-free Dulbecco's modified Eagle's media (Sigma) containing 100 Ci/ml [ 32 P]orthophosphate (NEN Life Science Products; 900 Ci/mmol) for 45 min at 37°C. At this point, 100 nM phorbol 12myristate 13-acetate (PMA, Sigma) or vehicle was added directly to the cultures and incubated for an additional 10 min. The cells were then rinsed three times with Dulbecco's phosphate-buffered saline (BioWhittaker) and used for subcellular fractionation or surface biotinylation.
Cell Fractionation-Transfected BAECs were fractionated using a slightly modified protocol from that described (22). Briefly, 6 -10 ϫ 10 6 BAECs or transfected BAECs were radiophosphorylated as described, stimulated with 100 nM PMA for 10 min, and then released from the substratum by trypsinization. After quenching the trypsinization with media containing 10% fetal calf sera, the cells were centrifuged at 200 ϫ g and washed three times in Dulbecco's phosphate-buffered saline. The final cell pellet was resuspended in a cold hypotonic buffer consisting of 2 mM Tris, pH 8.0, 0.2 mM adenosine triphosphate, 2 mM CaCl 2 , 10 mM sodium fluoride, 1 mM sodium vanadate, 100 g/ml leupeptin, 1 mM benzamadine hydrochloride, and 1 mM phenylmethylsulfonyl fluoride. For efficient release and subsequent fractionation, hypotonic buffer was added at a ratio of at least 1 ml to approximately 2-3 ϫ 10 6 cells. After incubation on ice for 15 min, the resultant nucleated ghosts were pelleted by centrifugation at 500 ϫ g for 5 min at 4°C. The hypotonic lysate was collected, and a volume of 5 M NaCl was added to yield a final concentration of 145 mM. The lysate was then cleared by ultracentrifugation at 100,000 ϫ g for 1 h at 4°C. The pelleted nucleated ghosts were then resuspended in the hypotonic buffer, pelleted, and washed twice more. The final supernatant was removed, and a volume of 0.1% Nonidet P-40 (v/v) in the hypotonic buffer was added to the pelleted nucleated ghosts. The nucleated ghosts were then subjected to homogenization using a Dounce homogenizer. The resultant homogenate was cleared of nuclei by centrifugation at 500 ϫ g for 5 min and then cleared of endosomal material and mitochondria by ultracentrifugation at 100,000 ϫ g for 1 h at 4°C. A volume of 5 M NaCl was added to the resultant homogenate to yield a final concentration of 145 mM. Protein content of both the hypotonic releasate and Nonidet P-40 homogenate was determined using the bicinchoninic acid method (Pierce).
Surface Biotinylation and Streptavidin-Agarose Precipitation-Apical and basolateral biotinylation of cells was performed essentially as described (23). For apical biotinylation, transfected clones were cultured in tissue culture wells. For basal biotinylations, transfectants were cultured in 6.8-cm 2 tissue culture-treated Transwell culture inserts with 3-micron pores. Cells were cultured 1-2 weeks past the point of confluence. For each experiment, fresh succinimidyl-6-(biotinamido)hexanoate (Pierce) was dissolved in chilled 50 mM NaHCO 3 , pH 8.2, 145 mM NaCl to yield a concentration of 55 g/ml. The cells were washed 3 times in prechilled biotinylation buffer; the biotin solution was added and incubated with the cells at 4°C for 30 min. For the basal biotinylation, the biotin solution was stirred in the lower chambers of the Transwell plate with microstir bars. After biotinylation, the cell monolayers were washed 3 times with Dulbecco's phosphate-buffered saline and lysed in 0.5% Triton X-100 in 10 mM imidazole, pH 7.15, 40 mM KCl, 10 mM EGTA containing 10 mM benzamidine, 0.5 mM phenylmethylsulfonyl fluoride, 100 g/ml leupeptin, 2 mM sodium vanadate, and 0.5 mM sodium fluoride, a buffer that stabilizes and preserves F-actin (3). Cells were lysed in a volume of 200 l per Transwell insert. Lysis and all subsequent steps were performed at room temperature. The lysates were briefly centrifuged at 500 ϫ g for 2 min, and a sample was saved for protein determination and immunoblot analysis. The remaining portion of the lysates was incubated with streptavidin-conjugated agarose (SA, Pierce) equilibrated to the lysis buffer at a ratio of 50 l of prepared SA-agarose beads per 200 l of lysate. After incubating for 30 min at room temperature while gently mixing, unbound material was removed by three consecutive washes in 10 bead volumes of lysis buffer. Cellular material was eluted off the SA-agarose by boiling in Laemmli SDS-PAGE sample buffer for 3 min. Samples were reduced by the addition of 5% ␤-mercaptoethanol and then analyzed for the presence of HSP27 and actin by immunoblot analysis. In some experiments, transfected BAECs were radiophosphorylated (as described above) prior to basal biotinylation. In every experiment, serially diluted Triton X-100 lysates were also subjected to immunoblot and autoradiographic analyses for the purpose of demonstrating linearity of the immunoreactive and autoradiographic signals (see below).
Immunoblotting-Cell subfractions, lysates, or streptavidin eluates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions on 12% (w/v) polyacrylamide gels (24). Upon completion of electrophoresis, proteins were transferred to nitrocellulose membranes that were then blocked for 1 h at room temperature in 5% nonfat milk in 20 mM Tris, pH 7.4, 145 mM NaCl with 0.05% Tween 20 (TTBS). The membranes were then incubated with dilutions of rabbit antisera raised against murine HSP25 (StressGen), a murine monoclonal antibody (G3.1, StressGen) specific for human HSP27 or a monoclonal antibody specific for non-muscle ␤-actin (clone AC-15, Sigma) in TTBS containing 1.0 mg/ml bovine serum albumin. Bound antibody was detected using donkey anti-mouse or anti-rabbit IgG conjugated to horseradish peroxidase (Jackson Laboratories) and the enhanced chemiluminescence reagent (Amersham Corp.). The stained immunoblots were placed against x-ray film (BioMax, Eastman Kodak Co.) to generate chemiluminographs.
Densitometric Analyses and the Demonstration of Linearity of Chemi-luminographic and Autoradiographic Signals-Densitometry of chemiluminographs and autoradiographs of immunoblots was performed with a Stratagene Eagle Eye II computerized digital camera and Eagle Sight 2.0 software. Densitometric data are generated as integrated density units. Chemiluminographic signals for immunoreactive material generated in the experiments presented in this report were within a linear range determined by analyzing serially diluted Triton lysates of BAECs or wtHSP27-expressing BAECs (Fig. 1, panels A and B) with the primary antibodies recognizing HSP25 (rabbit antisera raised to murine HSP25, StressGen, Inc.), HSP27 (the monoclonal antibody G3.1, StressGen, Inc.), and ␤-actin (clone AC-15, Sigma). Likewise, autoradiographic signal intensities were within the linear range of exposures as defined by densitometric analysis of the autoradiograph of an immunoblot on which serial dilutions of a Triton X-100 lysate prepared from radiolabeled wtHSP27-BAECs were analyzed (Fig. 1, panels A and B).
For fractionation experiments of radiophosphorylated BAECs, the autoradiographic integrated densities (Rad) of cytosolic HSP25/27 and membrane-associated HSP25/27 were determined for quiescent and PMA-stimulated BAECs. These values were divided by the integrated densitometric values of the immunostained HSP25/27 bands on chemiluminographs (i.e. antigen intensity, Ag). The resultant Rad/Ag ratios were averaged and presented with the calculated standard deviations.

HSP25, wtHSP27, and muHSP27 Are Present in Cell Fractions Enriched for Plasma Membrane Components-Overex-
pression of HSP27 in transfected cell lines generates enhanced cortical F-actin structure and alters cellular processes dependent on a dynamic membrane-associated microfilament cytoskeleton (e.g. pinocytosis and membrane ruffling) (9 -11), suggesting a functional association with the plasma membrane. To determine if endothelial cell HSP25 localizes to this cell compartment, BAECs were fractionated by sequential hypotonic lysis, homogenization in the presence of Nonidet P-40, and differential centrifugation (22). Using this protocol, BAE cell fractions enriched for nuclear (N), plasma membrane (P), microsomal (M), and cytosolic components (C) were generated and subjected to immunoblot analysis using rabbit antisera raised against murine HSP25 (Fig. 2, panel a). HSP25 primarily fractionated with either the Nonidet P-40 homogenate containing solubilized plasma membrane components (P) or the hypotonic lysate containing cytosolic proteins (C).
The protocol employed efficiently fractionates the plasma membrane compartment from the cytosolic compartment as demonstrated by the lack of anti-glucose-6-phosphate dehydrogenase (G6PDH) immunoreactive material in the plasma membrane fraction (P) but staining of immunoreactive material in the cytosolic (C) fraction (Fig. 2, panel b). Conversely, cytosolic material was not contaminated with plasma membrane components as indicated by the lack of anti-integrin ␤ 3 immunoreactivity in the cytosolic fraction but positive staining of a band with the appropriate apparent molecular mass (100 kDa) in the plasma membrane fraction (Fig. 1, panel C, lane P).
To determine whether phosphorylation of HSP27 is necessary for the partitioning of this protein into the membrane fraction, clones of BAECs expressing HSP27 or the non-phosphorylatable muHSP27 were also fractionated by this method and analyzed with the monoclonal antibody G3.1 that is specific for the human protein. Both wtHSP27 (Fig. 2, panel d,  lanes 1-3) and muHSP27 (Fig. 2, panel d, lanes 4 and 5) partitioned with the plasma membrane fraction indicating that phosphorylation of HSP27 is not required for the association of HSP25/27 with the membrane components. In panel d, a greater percentage of the plasma membrane fractions was subjected to analysis than the hypotonic lysates containing cytosolic components. This accounts for the greater signals obtained for the HSP25/27 in the plasma membrane fractions of the individual clones.
Quantitation of the Membrane-associated HSP25/27-Cell fractionation experiments detailed above were performed, and the relative antigen level in each fraction was determined by immunoblot analysis. The antigenic signals obtained for the membrane-associated or cytosolic HSP25/27 from the fractionated transfectants fell within the linear range of immunodetection (Fig. 1), allowing the comparison of signal intensities. The relative amounts of HSP25/27 in the cytosolic fractions and the plasma membrane fractions of BAECs, wtHSP27-, or mu-HSP27-expressing BAECs were calculated from the densitometric signals obtained from the immunoblot and the percentage of each fraction analyzed. The average percentages of total cellular HSP25, wtHSP27, and muHSP27 that partitioned with the membrane fractions were determined to be 27 Ϯ 9 (n ϭ 14), 26 Ϯ 12, and 24 Ϯ 15%, respectively.
Thus a significant portion of the endogenous HSP25 and the exogenous HSP27 gene products fractionate with membrane components. These results are not due to the nonspecific trapping of large oligomeric complexes containing HSP25/27 since glucose-6-phosphate dehydrogenase, which also exists in hetero-oligomeric complexes exhibiting molecular mass of up to 250 kDa (25, 26), was not detected in the plasma membrane fractions. Membrane-associated HSP25 Is Substrate for HSP25/27 Kinase Activity-The microfilament capping activity of HSP25, which inhibits actin polymerization in vitro, is abrogated by HSP25 phosphorylation (7). To investigate whether membrane HSP25 is phosphorylated in response to kinase activity induced by phorbol esters, BAECs were radiolabeled with 32 P i , stimulated with PMA, and then fractionated by the above procedure. To evaluate the relative levels of HSP25 phosphorylation in the two cellular compartments, autoradiographic and immunoblot analyses of the cytosolic and plasma membrane fractions obtained from PMA-stimulated, 32 P i -labeled BAECs were per-formed (Fig. 3, panel A). Because two-dimensional isoelectric focusing/SDS-PAGE demonstrated that no other phosphoproteins in endothelial cell lysates co-migrate in the second dimension with phospho-HSP25 (27,28), one-dimensional SDS-PAGE was deemed adequate for this study. Immunostaining for HSP25 demonstrated equivalent loading of HSP25 in the membrane and cytosolic fractions for both stimulated (ϩ) and non-stimulated (Ϫ) cells. PMA stimulation did not result in a change in the relative distribution of cellular HSP25 between the two compartments (Fig. 3, panel A). The percentage of cellular HSP25 that partitioned with the plasma membrane fraction was determined to be 37 Ϯ 9 and 35 Ϯ 10% (n ϭ 5) for non-stimulated and stimulated BAECs, respectively. The autoradiographic intensity of the cytosolic HSP25 did not increase as a result of PMA stimulation (Fig. 3, panel A). In contrast, the autoradiographic intensity of the membrane-associated HSP25 exhibited a 2.7-fold increase upon treatment with PMA.
To compare the relative levels of PMA-induced phosphorylation of membrane-associated HSP25 to cytosolic HSP25, the autoradiographic integrated densities (Rad) of cytosolic HSP25 and membrane-associated HSP25 were determined for quiescent and PMA-stimulated BAECs and then normalized to the chemiluminograph antigenic integrated densitometric values (Ag) obtained for the HSP25 immunostained bands. The resultant Rad/Ag ratios were averaged and presented with the calculated standard deviations in Fig. 3, panel B. The basal phosphorylation level of the plasma membrane-associated HSP25 was slightly higher than that of the cytosolic HSP25. Upon PMA stimulation, the Rad/Ag ratio of the membrane HSP25 increased 3.3 Ϯ 0.7-fold (n ϭ 4). In contrast, PMA stimulation did not affect the level of cytoplasmic HSP25 phosphorylation.  4 and 5), and a vector-transfected control clone (lane 6) were fractionated and the plasma membrane components subjected to immunoblot analysis using both anti-HSP25 and anti-HSP27 antibodies. B, quantitation of membrane-associated HSP25/27. Vector, wtHSP27, and muHSP27 transfected clones were fractionated as described for Fig. 1 and subjected to quantitative immunoblot analysis. The relative percentage of HSP25/27 present in the membrane fraction was calculated from the densitometric value obtained from the immunostaining and from the percentage of each fraction loaded. The percentages of cellular HSP25/27 present in the membrane fraction obtained for four different sets of transfected clones were averaged and are presented with the calculated standard deviation.

FIG. 3. Immunoblot and autoradiographic analyses of cytosolic and plasma membrane HSP25/27. Panel A,
BAECs were radiophosphorylated, allowed to remain quiescent (Ϫ), or stimulated with PMA (ϩ), fractionated into a cytosolic (Cyt.) and membrane (Mem.) fraction, and subjected to autoradiographic (Autorad) and Immunoblot analyses. Panel B, the autoradiographic densitometric values (Rad) obtained for the cytosolic and membrane-associated HSP25 present in fractions obtained from quiescent or PMA-stimulated BAECs were divided by the densitometric values (Ag) obtained for the anti-HSP25 immunostained bands of those fractions. The Rad/Ag ratios were averaged and are presented with the calculated standard deviations.
These data indicate that it is the portion of cellular HSP25 that fractionates with the plasma membrane which is the primary substrate for PMA-induced kinase activity.
HSP25/27 Localizes to the Basolateral Membrane-To determine if membrane HSP25/27 is evenly distributed in the cell, HSP27 transfectants were biotinylated at either the apical or basal surface, lysed with Triton X-100, and subjected to streptavidin-agarose precipitation, and then the eluted material was analyzed via immunoblotting for HSP27/25. Fig. 4, panel A, demonstrates the presence of HSP27, muHSP27, and endogenous HSP25 in extracts prepared from apically biotinylated cells (lanes 1, 2, and 7, respectively). The corresponding eluates from the streptavidin-agarose columns (panel A, lanes 4, 5, and 8), however, lacked HSP25/27. Horseradish peroxidase-SA staining of the blots (Fig. 4, panel B) demonstrates successful biotinylation and subsequent adsorption of the biotin-labeled material to the streptavidin-agarose except when biotin was omitted from the culture (lanes 3 and 6).
In contrast, wtHSP27 and muHSP27 were present in the eluates from streptavidin-agarose incubated with extracts of cells biotinylated from the basal surface (Fig. 4, panel C, lanes  4 and 5, respectively). Endogenous HSP25 could also be isolated from wtHSP27, muHSP27 (data not shown), and from vector-control BAECs (Fig. 4, panel C, lane 8). Horseradish peroxidase-biotin staining of immunoblots on which the SA flow-through was analyzed demonstrated that the SA quantitatively adsorbed all of the biotinylated cellular material. WtHSP27 was not detected in the eluate (Fig. 4, panel C, lane  6) obtained from a non-biotinylated wtHSP27 extract, demonstrating that the precipitation of HSP25/27 was not due to nonspecific adsorption.
Basolateral HSP25/27 May Effect the Generation of Membrane-associated Microfilaments-Since a significant portion of cellular HSP25/27 is associated with the plasma membrane and a functional association of HSP27 with membrane-associated microfilaments has been suggested (9 -11), it is likely that the expression and phosphorylation of HSP25/27 has an effect on these microfilaments. To determine if this is the case, the blots on which the SA precipitates were analyzed were reprobed with an anti-actin antibody (Fig. 5, panel A). Actin was present in the SA eluates of the basally biotinylated wtHSP27, muHSP27, and vector control BAECs (Fig. 5, panel A, lanes  5-7, respectively) but not in the eluted material from the nonbiotinylated control (lane 8) or in the eluates of extracts prepared from apically biotinylated cells, indicating that the isolated actin does not represent trapped or nonspecifically adsorbed material.
The actin that is precipitated from extracts prepared from BAECs biotinylated on the basal surface represents F-actin. This is indicated by the co-elution of the microfilament binding proteins ␣-actinin and myosin light chain (Fig. 5, panel B). These proteins have filament bundling and contractile activities, respectively, and do not interact with G-actin (29 -31). That the SA-precipitated actin represents filamentous and not monomeric actin is also suggested by the fact that endothelial cell F-actin, including cortical structures, exists below tight junctions (1) that are impermeable to apically applied biotin. The fact that apical biotinylation failed to result in the isolation of any actin suggests that surface biotinylation and subsequent streptavidin precipitation does not precipitate monomeric actin and that the actin isolated from the basally biotinylated cells represents F-actin.
PMA stimulation did not result in a significant change in the amount of HSP25/27 precipitated from basally biotinylated cells via SA precipitation (Fig. 6, panel A). For example, densitometric analysis revealed that the ratios of HSP27, muHSP27, and HSP25 precipitated from stimulated cells over that precipitated from quiescent wtHSP27 BAECS, mu-HSP27 BAECs, and control cells were 1.1, 1.2, and 0.9, respectively.
In contrast to the precipitation of basolateral HSP25/27, PMA stimulation resulted in the isolation of additional actin from basally biotinylated HSP27 BAECs and control BAECs (Fig. 6, panel A). For the wtHSP27 BAECs, PMA stimulation resulted in an average 1.8 Ϯ 0.3-fold (n ϭ 5) increase in the amount of actin precipitated as determined by densitometric analysis of the immunoblot stained with a monoclonal antibody specific for non-muscle ␤-actin (Fig. 6, panel A). PMA stimulation of control BAECs resulted in a more modest 1.3 Ϯ 0.18-fold (n ϭ 3) increase in the amount of actin precipitated (Fig. 6,  panel A). In contrast, expression of muHSP27 at the mem- FIG. 4. Surface biotinylation and streptavidin-agarose precipitation. Transfectants were biotinylated either at the apical (panels A and B) or basal (panels C and D) surface, lysed with 0.5% Triton X-100, and then subjected to streptavidin-agarose precipitation. The starting and eluted material were analyzed via immunoblotting using the anti-HSP27 monoclonal antibody G3.1 (panels A and C) and then with horseradish peroxidase-streptavidin (panels B and D) to detect the biotinylated proteins isolated via the precipitation. Lysates prepared from biotinylated wtHSP27-BAECs, muHSP27-BAECs, and from a non-biotinylated wtHSP27-BAE clone were analyzed in lanes 1-3, respectively. The eluted material from these lysates were analyzed in lanes 4 -6, respectively. In a separate experiment, extracts of biotinylated vector-transfected BAECs (lane 7) were subjected to SA precipitation, and the eluate was analyzed in lane 8.
FIG. 5. Streptavidin-agarose precipitation of membrane-associated F-actin. Panel A, lysates prepared from apically (Apical Biot.) or basally biotinylated (Basal Biot.) wtHSP27, muHSP27, vector control, and non-biotinylated wtHSP27 BAECs (lanes 1-4, respectively) were subjected to SA precipitation and, together with the corresponding eluted material (lanes 5-8, respectively), subjected to immunoblot analysis using a monoclonal anti-␤-actin. Panel B, extracts prepared from basally biotinylated wtHSP27, muHSP27 or vector control BAECs (lanes 1-3, respectively) were subjected to SA precipitation and, together with the corresponding SA-precipitated material (lanes 4 -6), analyzed by immunoblot analysis with antibodies specific for ␣-actinin and myosin light chain (MLC). brane, which is not a substrate for kinase activity, inhibited the PMA-induced generation of membrane-associated actin. The amount of actin present in the SA precipitates obtained from PMA-stimulated muHSP27 BAECs depicted in Fig. 6, panel A, was 0.65 Ϯ 0.09 times that of actin precipitated from nonstimulated muHSP27 BAECs. Thus while the amount of HSP25/27 precipitated via SA precipitation did not change upon PMA stimulation, the amount of actin isolated did in fact change.
Concurrent with the PMA-induced change in the amount of F-actin precipitated from extracts prepared from basally biotinylated cells, PMA stimulation resulted in an increase in the phosphorylation of basolateral wtHSP27 and HSP25 of 2.4 Ϯ 0.3-(n ϭ 5) and 1.7 Ϯ 0.15-fold, respectively. In contrast, lysate HSP27 and HSP25 phosphorylation increased 1.38 Ϯ 0.33-and 1.1 Ϯ 0.1-fold, respectively (Fig. 7). As indicated above, PMA stimulation did not result in a change in the amount of HSP25/27 isolated via streptavidin precipitation of basally biotinylated cells. These results, and the fact that muHSP27 was isolated to a similar extent as wtHSP27, indicate that association of HSP25/27 at the basolateral membrane is independent of its phosphorylation state. This is in agreement with the results obtained for the entire membrane HSP25/27 pool. Likewise, basolateral HSP25/27 exhibited a greater propensity to serve as substrate for PMA-induced HSP25/27 kinase activity than the remaining HSP25/27, a finding consistent with the partitioning of phospho-HSP25 with the plasma membrane in BAECs fractionated by sequential hypotonic lysis and homogenation.

DISCUSSION
The small molecular weight heat shock protein HSP27 has been demonstrated to be an important microfilament modulating protein involved in the control of microfilament dynamics and organization that is regulated by inducible kinase activity. For example, phosphorylation of HSP25/27 retards the inhibitory filament capping activity demonstrable in vitro (5)(6)(7). In cultured cells, transfection and expression of the wild type human HSP27 stabilizes cortical filaments that normally disaggregate in response to cytokines, oxidative stress, and hyperthermia, conditions that result in the phosphorylation of HSP27 in vivo (11, 14 -17). Enhanced HSP27 expression in these cells conveyed increased thermo-tolerance, whereas expression of a non-phosphorylatable mutant HSP27 did not convey resistance and had no effect on microfilament dynamics or, in some cases, reversed the effect of the wild type protein (9). In non-stressed cultures, expression of the phosphorylatable wild type protein enhanced pinocytotic activity and membrane ruffling (9), processes dependent on a dynamic membrane-associated cytoskeleton (12,13). The reliance on the ability of HSP27 to be phosphorylated to impart biological consequences may not be universal, however. For example, Knauf et al. (18) have shown that a non-phosphorylatable mutant HSP27 conveys the same thermo-tolerance in Swiss 3T3 cells as did the wild type protein. These data indicate that HSP27 phosphorylation, and thus regulation of HSP27 activity, may be spatially and/or temporally regulated.
The stabilization of cortical actin filaments in stressful situations, enhanced membrane ruffling, and increased pinocytosis in cells expressing elevated levels of HSP27 suggests that the HSP25/27 activity is spatially regulated. Data presented in this report identified a subpopulation representing nearly 30% of cellular HSP25/27 that fractionates with plasma membrane components. muHSP27, which is not phosphorylated in response to agonists such as PMA, partitioned with this fraction to the same extent as exogenous wtHSP27 or endogenous HSP25, indicating that phosphorylation is not necessary for the association(s) responsible for its partitioning with membrane components. Membrane-associated HSP25/27 was demonstrated to be distinct from the cytoplasmic HSP25/27 in that PMA stimulation induced the phosphorylation of the former pool, but not the latter. Importantly, the HSP25/27 antigen FIG. 6. Streptavidin-agarose precipitation of radiophosphorylated HSP25/27 and membrane-associated F-actin. Panel A, wtHSP27-BAECS, muHSP27-BAECs, and vector control BAECS were radiophosphorylated, allowed to remain quiescent (lanes 1 and 3), or stimulated with PMA (lanes 2 and 4). The cells were biotinylated on the basal surface and then lysed with 0.5% Triton X-100. The lysates were then subjected to streptavidin-agarose precipitation. The lysates (lanes 1 and 2) and eluted material (lanes 3 and 4) were subjected to SDS-PAGE and transferred to nitrocellulose membranes. The membranes were subjected to immunoblot analysis using monoclonal antibodies specific for human HSP27 and non-muscle ␤-actin. Panel B, the streptavidin-agarose eluates obtained from quiescent or PMA-stimulated basally biotinylated wtHSP27 (n ϭ 4), muHSP27 (n ϭ 3), and vector control cells (n ϭ 3) were subjected to immunoblot analysis using the anti-␤-actin monoclonal antibody. The ratio of densitometric values obtained for the membrane-associated F-actin obtained for the actin bands of stimulated cells over that obtained for the nonstimulated cells was then calculated.

FIG. 7.
Autoradiographic analyses of streptavidin-agarose eluates. wtHSP27, muHSP27, and vector control cells were radiophosphorylated with 32 P i , left quiescent (lanes 1 and 3), or stimulated with PMA (lanes 2 and 4) while growing on Transwell membranes. Biotin was placed in the subluminal chamber and then Triton X-100 extracts were prepared (lanes 1 and 2). The extracts were subjected to SA precipitation and the eluates and extracts (lanes 3 and 4) were subjected to SDS-PAGE and transferred to nitrocellulose which was then exposed to film. After radiography, the blots were immunostained for HSP25 and HSP27. level in the membrane fraction did not change upon PMA stimulation, indicating that phosphorylation did not induce a change in HSP25/27 distribution. Thus, any alteration in the activity of the membrane-associated HSP25/27 caused by PMA stimulation is likely a result of its phosphorylation. Although the kinase present at the membrane responsible for HSP27 phosphorylation has yet to be defined, one kinase that has been shown to phosphorylate HSP27 is the mitogen-activated protein kinase-activated protein kinase 2 (11, 14 -17).
Although extremely useful in identifying a distinct pool of cellular HSP25/27, the hypotonic buffer used in the fractionation protocol employed does not preserve F-actin structure and therefore is of little use in addressing whether the functional consequences of HSP25/27 phosphorylation involve microfilaments. A second fractionation method in which membrane-associated F-actin was isolated and quantitated was therefore employed. BAEC transfectants were biotinylated at their basal surface, lysed in Triton X-100 in a buffer that stabilizes microfilaments (3), and then were subjected to SA precipitation. In these experiments, 1) HSP25 and HSP27 were detected exclusively in the eluates obtained from basal biotinylated cells, demonstrating an association with basolateral membrane components, 2) muHSP27 was detected in these eluates indicating that membrane association does not require HSP27 phosphorylation, and 3) PMA induced the phosphorylation of the eluted HSP25/27 to over twice that of untreated cells whereas lysate HSP25/27 was less affected. Thus, by two different procedures, it is apparent that membrane association of HSP27 is independent of its phosphorylation state and that this subpopulation of HSP25/27 is substrate for PMA-induced HSP27 kinase activity.
The exact nature of the interaction(s) of HSP25/27 at the basolateral membrane remains to be determined. It is possible that HSP27 associates with the membrane itself, with integral membrane proteins, or through a direct interaction with basolateral microfilaments. The former two possibilities are indicated by the persistence of HSP25/27 with the plasma membrane fraction in buffer conditions in which actin filaments readily depolymerize. An association of HSP25/27 with the membrane would not be novel since other actin filament modulating proteins (e.g. vinculin) also interact with the membrane (32). The possibility that HSP25/27 was fractionated with membrane components due to the interaction with the membrane-associated microfilaments is suggested by the in vitro studies demonstrating that HSP25/27 acts as a filament capping protein (5,6).
The localization of HSP27 to the basolateral membrane suggests that this subpopulation of HSP27 may have a pivotal role in regulating basolateral membrane-associated microfilaments, particularly when the membrane-associated cytoskeleton is undergoing dynamic changes. In Swiss 3T3 cells, PMA treatment has been shown to stimulate membrane ruffling, a process associated with the enhanced polymerization of cortical actin (33). Likewise, PMA stimulation of wtHSP27 BAECs generated a substantial increase in the amount of F-actin isolated via streptavidin precipitation, whereas muHSP27 expression actually decreased the amount of membrane-associated F-actin precipitated. These changes do not appear to be artifacts of the biotinylation of activated versus non-activated cells since SA precipitated equivalent amounts of biotinylated material and associated proteins. For example, the amount of wtor muHSP27 isolated from activated or quiescent cells was found to be identical. The increased membrane-associated Factin in the wtHSP27 BAECS is consistent with the observation that stress, which results in HSP27 phosphorylation, leads to an accumulation of cortical actin in fibroblasts (8, 10, 11). A reduction in the amount of membrane-associated F-actin in muHSP27 expressing cells is congruous with the dominant negative effects of muHSP27 expression on membrane ruffling and pinocytosis (9).
The mechanism by which wtHSP27 or muHSP27 expression affects PMA-induced changes in the basolateral membraneassociated microfilament cytoskeleton is suggested by the in vitro observations of HSP25/27 activity (5-7). The barbed-end capping activity of HSP25/27 inhibits actin polymerization when HSP25/27 is unphosphorylated. Since both wtHSP27 and muHSP27 localize to the membrane, PMA stimulation would result in the phosphorylation of the wild type protein but not the mutant gene product. Thus, cell activation would inhibit wtHSP27 activity, allowing the generation of F-actin. Introduction of muHSP27 would establish a population that is constitutively active and continuously inhibits F-actin generation. In this manner HSP25/27 may contribute to the regulation of membrane-associated F-actin polmerization.
A putative consequence of this regulation is the control of endothelial cell migration. It has been recently demonstrated that the expression of wtHSP27 or muHSP27 enhances or retards BAEC migration, respectively, in a wound assay. 2 Since endothelial cell migration is dependent on the polymerization of a meshwork of lamelipodial F-actin that is associated with the basolateral membrane (4), HSP25/27 may be regulating migration at the level of lamelipodial extension. This is substantiated by the fact that the F-actin demonstrated to be affected by HSP25/27 and subject to SA precipitation is part of the Triton X-100 soluble pool of cellular F-actin. This pool has been demonstrated in other cell types to be highly dynamic and includes filaments involved in lamelipodial extension and chemotaxis (2,3).