Identification and Characterization of hic-5/ARA55 as an hsp27 Binding Protein*

hsp27 has been reported to participate in a wide vari-ety of activities, including resistance to thermal and metabolic stress, regulation of growth and differentiation, and acting as a molecular chaperone or a regulator of actin polymerization. We hypothesized that these di-verse functions are regulated in a cell- or tissue-specific manner via interaction with various binding proteins. To investigate this hypothesis, we used hsp27 as a “bait” to screen a yeast two-hybrid cDNA library from rat kidney glomeruli and identified a novel hsp27 binding protein, hic-5 (also known as ARA55), a focal adhesion protein and steroid receptor co-activator. Biochemical interaction between hsp27 and hic-5 was confirmed by co-immunoprecipitation, and critical protein (cid:1) protein interaction regions were mapped to the hic-5 LIM domains and the hsp27 C-terminal domain. Initial analysis of the functional role of hsp27 (cid:1) hic-5 interaction revealed that hic-5 significantly inhibited the protection against heat-induced cell death conferred by hsp27 overexpression in co-transfected 293T cells. In contrast, when a non-hsp27-interacting hic-5 truncation mutant (hic-5/ (cid:1) LIM4) was co-expressed

Heat shock protein 27 (hsp27) 1 is a member of the low molecular weight heat shock protein (hsp) superfamily. It is constitutively expressed in many mammalian cell types, and its expression increases in response to sublethal stresses such as heat shock, exposure to toxicants, and oxidative stress (1)(2)(3)(4). hsp27 also becomes rapidly phosphorylated in response to similar stresses, as well as to exposure to cytokines and mitogens (5). The accumulation and phosphorylation of hsp27 have been correlated with increased cell survival and recovery from stress conditions (6,7).
Two predominant hypotheses have been proposed to explain the mechanism by which hsp27 may protect cells from harmful stresses. One hypothesis is that hsp27 acts as a chaperone to facilitate folding/renaturation of proteins that have become denatured under stress conditions. In vitro, both hsp27 and the related small hsp, ␣B-crystallin, were shown to be able to rescue citrate synthase or alcohol dehydrogenase from temperature-induced aggregation (8,9). In addition, cultured cells overexpressing hsp27 were reported to exhibit accelerated recovery from heat-induced nuclear protein aggregation (10). This chaperone activity has been correlated with the presence of large oligomers of hsp27 (11,12). Upon phosphorylation, however, large hsp27 oligomers have been reported to dissociate into small oligomers or dimers (12)(13)(14). Reduction in the size of hsp27 oligomers has been correlated with elimination of in vitro chaperone activity (11,12).
The other hypothesized mechanism of protection is that hsp27 stabilizes actin filaments in an as yet undefined manner resulting in increased resistance to stresses. This function has been correlated with small hsp27 oligomers or dimers (2) and appears to be regulated by the phosphorylation of hsp27 (15,16). For example, hsp27 in its unphosphorylated form can inhibit actin polymerization in vitro whereas phosphorylated hsp27 lacks the ability to inhibit actin polymerization (16). In addition, microinjection of recombinant non-phosphorylated hsp27, but not phosphorylated hsp27, into ROC 17/2.8 cells inhibited the restoration of heat shock-disrupted actin stress fibers (17). Furthermore, overexpression of wild type hsp27, which can become phosphorylated under stress conditions, stabilized stress fibers during hyperthermia, prevented cytochalasin D-induced actin depolymerization, increased cortical polymerized actin (F-actin) content, and increased ruffling and pinocytotic activity (7,15). In contrast, overexpression of a non-phosphorylatable hsp27 mutant (S15G, S75G, and S82G) destabilized stress fibers, reduced cortical F-actin content, and decreased pinocytotic activity.
Although hsp27 may regulate the dynamics of actin polymerization, the molecular basis for the interaction between hsp27 and actin is not yet clear. hsp27 co-localizes with actin filaments in several highly differentiated cell types, including Sertoli cells (18), striated muscle cells (19,20), cultured differentiated C2C12 muscle cells, and kidney glomerular podocytes, 2 but appears to be distributed throughout the cytoplasm in most other cell types. Importantly, however, although direct interaction between hsp27 and actin has been demonstrated in vitro (16,21,22), no direct interaction between hsp27 and actin in vivo has been reported. Together, these findings led us to hypothesize that hsp27 may regulate actin via a novel mechanism involving hsp27 binding proteins (hsp27-BP) that are expressed in a cell and/or tissue-specific manner. Several hsp27-BPs have previously been identified, including mammalian transglutaminase (platelet factor XIII) (23), a Drosophila ubiquitin-conjugating enzyme (24), and a novel protein, PASS1 (protein associated with small stress proteins 1) (25).
The present study was undertaken in an attempt to identify hsp27-BP(s) that might play a role in the regulation of the actin cytoskeleton by hsp27. We present both in vitro and in vivo evidence demonstrating that hsp27 binds to a known focal adhesion/steroid receptor co-activator protein, hic-5/ARA55 (hydrogen peroxide-inducible clone-5/androgen receptor co-activator, 55 kDa), and define the C termini of both hsp27 and hic-5 that are required for their interaction. We also demonstrate that the interaction of hic-5 with hsp27 abolishes the protection against heat stress-induced cell death conferred by hsp27 overexpression.

EXPERIMENTAL PROCEDURES
Isolation of Normal Rat Kidney Glomeruli-Kidneys harvested from normal adult male Sprague-Dawley rats (175-200 g) were rinsed in cold PBS, and the renal capsules were removed. Glomeruli were isolated by mincing kidney cortical tissue with a razor blade, rinsing with PBS, pressing the tissue through a no. 140 stainless steel sieve (W. S. Tyler, Mentor, OH), and collecting the glomeruli on a no. 200 sieve, as previously described by Kreiserg et al. (26). Isolated glomeruli were assessed under the microscope, and the procedure was repeated until glomerular samples were Ͼ95% pure.
Preparation of the Normal Rat Kidney Glomerular Two-hybrid cDNA Library-Poly(A) mRNA was extracted from the isolated glomerular preparations and used to generate a HybriZap-2.1 two-hybrid cDNA library (Stratagene, La Jolla, CA). In vivo excision of the lambda vector was then performed to release the GAL4 activation domain vectors (pGAL4-AD) containing the cDNA inserts, and the cDNA library was amplified according to the manufacturer's instructions.
Two-hybrid Library Screening-The Y190 yeast host strain carries three auxotrophic markers, leucine, tryptophan, and histidine, for transformation selection, and two reporter genes, lacZ and HIS3, for the detection of in vivo protein⅐protein interactions. The expression of the reporter genes are controlled by GAL4 elements positioned upstream of each of the two reporter genes. Y190 cells that had been previously transformed with the bait plasmid pGAL4-BD-hsp27 (GAL4-DNA binding domain fused with the full-length rat hsp27 cDNA) (27) were transformed with prey plasmids containing the GAL4 activation domain fused with cDNAs from the rat glomerular cDNA library. Yeast colonies containing plasmids encoding potential hsp27-BPs were selected on tryptophan-, leucine-, and histidine-deficient plates, and colonies were further selected by the reporter gene expression assay (␤galactosidase lift assay) according to the manufacturer's instructions (Stratagene). The total number of cDNA copies that were screened was calculated by counting the colonies from an aliquot of the above-transformed yeast cells grown on leucine/tryptophan-depleted plates. Plasmids containing potential hsp27-BP cDNA were isolated from the positive yeast colonies after eliminating the pGal4-BD-hsp27 plasmid according to manufacturer's instructions (Stratagene). Plasmids were then transformed into Escherichia coli, re-isolated, and sequenced by the DNA Sequencing Core of the University of Michigan.
Preparation of cDNA Constructs-cDNA fragments were generated by polymerase chain reaction (PCR) using an Expand High Fidelity PCR kit (Roche Molecular Biochemicals, Indianapolis, IN). PCR was performed using a GeneAmp 2400 PAR system (PerkinElmer Life Sciences, Boston, MA) with 30 cycles at different temperatures for denaturing (95°C), annealing (50 -60°C), and extension (60°C) for various times (30 s to 2 min).
The mouse full-length hic-5 cDNA (GenBank accession number L22482) was used in PCR reactions as the template to prepare FLAGtagged hic-5 and its truncation mutant cDNA constructs. The oligonucleotide that encodes the FLAG epitope (N-AspTyrLysAspAspAspAs-pLys-) was fused to hic-5 and its truncation mutant sequences at the 5Ј-end of the following primer, AGGGGTACCACCATGGACTACAAAG-ACGATGACGACAAGGAATTCATGTCACGGTTAGGGGCTCCAAAA-GAGCGC (primer 1). Full-length hic-5 cDNA was produced by PCR with primer 1 and GCTCTAGATCAGCCGAAGAGCTTCAGGAAGCA (primer 2). The N-hic-5 truncation mutant cDNA was produced by PCR with primer 1 and GCTCTAGAGGCCTTTGGCCTGTGTGGGGAC, hic-5⌬LIM3/4 with primer 1 and GCTCTAGACTAGCGTGGGGCGAACA-GCTGCAG, and hic-5⌬LIM4 with primer 1 and GCTCTAGACTACAG-CGAACCACGCTGAGCATG. The C-hic-5 truncation mutant cDNA was produced by PCR with the 5Ј-end primer, AGGGGTACCACCATGGA-CTACAAAGACGATGACGACAAGTGTGGCTCCTGCAATAAACCTA-TAGCT, and primer 2. External restriction sites were introduced at the ends of the fragments for cloning purposes. The PCR reactions were digested with KpnI/XbaI for hic-5, N-hic-5, hic-5⌬LIM3/4, and hic-5⌬LIM4 and EcoRI/XbaI for C-hic-5 cDNAs (all restriction enzymes from Roche Molecular Biochemicals) and separated by electrophoresis on 1% agarose gels. PCR products were excised from the gel as determined by their calculated molecular weights and inserted into the expression vector pcDNA3.1 (Invitrogen, San Diego, CA) at the corresponding restriction sites.
The monoclonal anti-hsp27 antibody (28) used in these studies was generated against a peptide corresponding to amino acid residues 123-137 of human hsp27. This monoclonal antibody does not react with the hsp27 truncation mutant proteins we prepared for the domain mapping studies. A pcDNA3.1/Myc-His vector (Invitrogen) was used to construct a C-terminal Myc-tagged full-length rat hsp27 and its truncation mutant constructs, allowing detection of the Myc-tagged fusion proteins with a commercially available anti-Myc antibody. Rat hsp27 cDNA (GenBank accession number M86389) (18) was used as template in PCR reactions to generate a full-length hsp27 cDNA and its truncations. The full-length hsp27 cDNA was produced with a 5Ј-end primer of TTCGAATTCGCCACCATGACCGAGCGCCGCGTGCCCTTCTCGCTA (hsp primer 1) and a 3Ј-end primer of GCCGAATTCCTTGGCTCCAG-ACTGTTCCGACTC (hsp primer 2), and cloned into the vector at EcoRI sites. The N-hsp27 truncation mutant was produced using hsp primer 1 and GCTCTAGACCGAGAGATGTAGCCATG and cloned into the vector at EcoRI/XbaI sites. The C-hsp27 truncation mutant was generated with primers of CGGGGTACCACCATGCCCGCCTTCAGCCGG and hsp primer 2 and cloned into the vector at KpnI/EcoRI sites. To confirm the sequence accuracy of all constructs, DNA sequencing was performed by the DNA Sequencing Core of the University of Michigan.
Cell Culture and Transient Transfections-Kidney epithelial (293T) cells of human embryonic origin were grown in medium consisting of Dulbecco's modified Eagle's medium supplemented with non-essential amino acids, 100 units/ml penicillin, 100 g/ml streptomycin, and 10% fetal bovine serum (Life Technologies, Inc., Gaithersburg, MD) at 37°C in a 100% relative humidity and 5% CO 2 atmosphere. Transient transfection was performed using FuGENE 6 (Roche Molecular Biochemicals) according to the manufacturer's protocol. Briefly, 24 h prior to transfection, cells were dispensed into 6-well plates at 1 ϫ 10 5 cells/ well. After culture for 1 day, plasmid cDNA (1 g/well) and FuGENE 6 (3 l/well) were added to the cells. Forty-four hours after transfection, cells were either collected for immunoprecipitation studies or used in heat shock studies.
Immunoprecipitation Studies-Protein extracts from immortalized adult rat Sertoli (ASC-17D) cells were used in immunoprecipitation studies to detect an endogenous hsp27⅐hic-5 interaction. ASC-17D cells (29) were cultured in Dulbecco's modified Eagle's medium:nutrient mixture F-12 with 4% fetal bovine serum at 33°C and differentiated in the same medium at 40°C for 3 days. Cells were rinsed three times with ice-cold PBS containing 1 mM sodium orthovanadate and lysed in immune precipitation buffer (50 mM Tris-Cl, pH 7.5, 150 mM NaCl, 2 mM EGTA, 0.1% Triton X-100) containing protease inhibitors (1ϫ complete protease inhibitor mixture, Roche Molecular Biochemicals). The lysate was collected and centrifuged at 12,000 ϫ g for 10 min at 4°C, and the supernatant (800 -1000 l) was incubated with 4 l of anti-hic-5 monoclonal antibody (H-82420, Transduction Laboratories, Lexington, KY) or 2 l of anti-rat hsp27 monoclonal antibody (8A7) (18,28) at 4°C for 1 h. The resulting immune complexes were isolated from the lysates by addition of protein G-Sepharose beads (Roche Molecular Biochemicals), incubated at 4°C for 1 h on a rotating platform, followed by collection of the beads by a brief centrifugation. Bound immunocomplexes were released from the beads after three washes with lysis buffer by resuspension and a 3-min, 100°C incubation in SDS sample buffer. Controls consisted of samples with normal mouse sera substituted for anti-hic-5 or anti-hsp27 antibodies.
To define the critical interaction regions between hsp27 and hic-5, immune precipitation buffer lysates of transfected 293T cells were prepared as described above and incubated with 1 l of anti-FLAG antibody (Sigma Chemical Co., St. Louis, MO), anti-Myc antibody (Roche Molecular Biochemicals) or anti-paxillin antibody (P13520, Transduction Laboratories, Lexington, KY) at 4°C for 1 h. The resulting immune complexes were isolated and solubilized as described above. Controls consisted of samples with normal mouse sera substituted for anti-FLAG or anti-Myc antibodies.
Preparation of Rat Tissue Extracts-Tissues from various organs were collected from adult rats followed by immediate homogenization in SDS sample buffer (2% SDS, 62.5 mM Tris-Cl, pH 6.8, 10% glycerol) with protease inhibitors (Roche Molecular Biochemicals). Homogenates were boiled at 100°C for 5 min and centrifuged at 12,000 ϫ g for 10 min. The protein concentration in the resulting supernatants was determined by BCA microassay, using albumin standards also prepared in solubilization buffer. Prior to electrophoresis, samples were brought to 5% 2-mercaptoethanol and incubated at 100°C for 5 min.
SDS-PAGE and Western Blotting-For SDS-PAGE, 30 g of total protein/sample in SDS buffer was resolved on 12% polyacrylamide gels using a mini-PROTEAN II Electrophoresis Cell (Bio-Rad, Hercules, CA), and the separated proteins were transferred onto polyvinylidene difluoride membranes (Millipore, Bedford, MA) using a submarine electrophoretic transfer unit in the same apparatus.
For Western blotting, membranes were blocked for 1 h in 0.1% Tween 20 in PBS (PBS-T) containing 5% nonfat dry milk. Membranes were then incubated for 1 h at room temperature with one of the following primary antibodies diluted in blocking solution: 1:2000 mouse antihic-5 (H82420), 1:5000 mouse anti-Myc, anti-FLAG, anti-rat hsp27 (8A7), or 1:10,000 mouse anti-paxillin antibodies (P13520, Transduction Laboratories). Following incubation with the primary antibody, membranes were washed three times in PBS-T and incubated for 1 h with either goat anti-mouse IgG F(ab) or goat anti-mouse IgG Fc antibodies conjugated to horseradish peroxidase (Jackson ImmunoResearch Laboratories, West Grove, PA) at a dilution of 1:10,000 in blocking solution. The membranes were then washed three times in PSB-T and processed for detection by enhanced chemiluminescence (ECL reagent, Amersham Pharmacia Biotech, Arlington Heights, IL).
Heat Shock Treatment and Neutral Red Uptake-293T cells were detached from six-well cell culture plates with trypsin solution (Life Technologies, Inc.), and an equal number of cells (1 ϫ 10 5 cells/well) was dispensed into 24-well culture plates. Cells were cultured for 6 h to allow them to attach to the plates and then heat-shocked at 45.0, 46.0, or 47.0 Ϯ 0.1°C for 30 min in a circulating water bath. After 14 h of recovery at 37°C, neutral red uptake assays were performed as described previously (30) with some modifications. Briefly, 200 l of PBS containing 0.033% (w/v) neutral red (Sigma) was added to cell culture medium. After a 3-h incubation at 37°C, cells were collected and washed twice in PBS. The cells were then resuspended and lysed by addition of an equal volume of a 50% (v/v) mixture of ethanol and 0.1 M monobasic sodium phosphate followed by incubation for 1 h at room temperature. After centrifugation, an equal volume of each supernatant was transferred to 96-well plates. The absorbance of supernatants at 550 nm was determined by using an EAR 400AT microplate spectrophotometer (SLT-Lab Instruments, Austria).
Statistics-All experimental results were expressed as mean Ϯ S.E. All statistical analyses were performed using unpaired two-tailed t tests. Results were considered significant at p Ͻ 0.05.

Identification of hic-5 as an hsp27
Binding Protein-To identify hsp27 binding proteins, ϳ1.3 ϫ 10 6 independent cDNA clones were screened from a normal rat kidney glomerular two-hybrid cDNA library, and a total of 32 positive yeast colonies were identified and sequenced. A homology search in Gen-Bank using the BLAST program revealed that 24 out of 32 plasmids encoded hsp27. One of the remaining positive plasmids contained a cDNA with high sequence similarity to the C-terminal sequences of mouse and human hic-5 cDNA. On this basis, the isolated cDNA (GenBank accession number AF314960) was assumed to code for rat hic-5. Fig. 1 shows the nucleotide sequence of the plasmid isolated from the rat glomerular library, and the amino acid sequence match between this partial rat hic-5 sequence and the full-length sequence of mouse hic-5. To confirm the interaction between hsp27 and hic-5 in the two-hybrid system, full-length murine hic-5 cDNA and full-length rat hsp27 cDNA were co-transfected into Y190 yeast cells. All of the resulting yeast colonies grew on the tryptophan-, leucine-, and histidine-deficient plates, whereas yeast cells transfected with either cDNA alone failed to grow. All double-transfected yeast colonies also turned dark blue in the ␤-galactosidase lift assay. The full-length hic-5 was negative when tested for interaction with p53, pLamin C, and pBD-GAL4 Cam in the yeast two-hybrid system. In addition, when mouse hic-5 was used as the bait to screen a yeast two-hybrid cDNA from C2C12 myoblasts, multiple positive colonies containing cDNA encoding hsp27 were identified. 3 Together these results confirmed a true interaction between rat/mouse hic-5 and rat hsp27 in the yeast two-hybrid system.
Biochemical Confirmation of hic-5 as a hsp27 Binding Protein-Because false-positive colonies could arise from the yeast two-hybrid screening, the interaction between hsp27 and hic-5 was further examined by co-immunoprecipitation. Extracts from co-transfected 293T cells were immunoprecipitated, and the immune complexes were analyzed by Western blotting probed with anti-FLAG, anti-Myc, or anti-paxillin antibodies. Because the molecular mass of hsp27 (about 25 kDa) is close to that of IgG light chain, and hic-5 (about 51 kDa) is close to that of IgG heavy chain, to avoid obscuring the visualization of these proteins on Western blots, the membranes were cut in half between the 45-and 30-kDa molecular mass markers and probed separately with secondary antibodies that recognized only the heavy or light chains of the immunoprecipitated antibody. Fig. 2 shows the results of an immunoprecipitated analysis of extracts from 293T cells transfected with vector (pcDNA3.1) alone, the FLAG-hic-5 or hsp27-Myc constructs, or co-transfected with both constructs. This analysis clearly shows that FLAG-hic-5 and hsp27-Myc are immunoprecipitated by the respective anti-FLAG and anti-Myc antibodies from extracts of cells transfected with the appropriate FLAGhic-5 or hsp27-Myc constructs (Fig. 2, lanes 5 and 8). No nonspecific immunoprecipitation of these proteins was observed using either the anti-FLAG or anti-Myc antibodies (Fig. 2,  lanes 6 and 7), or using normal mouse serum as control in cells co-transfected with both constructs (Fig. 2, lane 10). Furthermore, in 293T cells co-transfected with Lac-z-Myc and FLAGhic-5, no Lac-z-Myc was detected by Western blotting in the immunocomplexes precipitated by the anti-FLAG antibody (data not shown). The anti-FLAG and anti-Myc antibodies coimmunoprecipitated hsp27-Myc (Fig. 2, lane 3) and FLAG-hic-5 (Fig. 2, lane 4), respectively, from cells co-transfected with both constructs. Interestingly, anti-paxillin antibody failed to coimmunoprecipitate hsp27-Myc from cells transfected with this construct (Fig. 2, lane 9) despite the considerable sequence similarity between hic-5 and paxillin. Similarly, anti-Myc antibody did not co-immunoprecipitate paxillin from any transfected cell type (Fig. 2, lanes 4 and 8). These findings clearly demonstrated a biochemical interaction in vivo between hsp27 and hic-5 but not between hsp27 and paxillin.
To confirm that hsp27 and hic-5 interact in the absence of exogenous overexpression of these proteins, we chose a rat Sertoli cell line, ASC-17D, which has substantial endogenous expression of both hsp27 and hic-5, for a co-immunoprecipitation study. Extracts from ASC-17D cells were immunoprecipitated with an anti-hsp27 antibody (8A7) or a recently developed anti-hic-5 antibody (H-82420, Transduction Labs), which does not recognize paxillin on Western blots, and the resulting immune complexes were analyzed by Western blotting. Both the anti-hsp27 and anti-hic-5 antibodies were able to co-immunoprecipitate hsp27 and hic-5 from extracts of ASC-17D cells (Fig. 3, lanes 1 and 2). No nonspecific immunoprecipitation of these proteins was observed when normal mouse serum was used in place of the specific mouse anti-hsp27 or hic-5 antibodies (Fig. 3, lane 3).
Expression Pattern of hic-5 in Rat Tissues-To determine the expression of hic-5 in adult rat tissues, protein extracts were prepared and analyzed by Western blotting with a polyclonal hic-5 antibody (31) and a monoclonal anti-paxillin antibody (P13520, Transduction Labs). The distribution of hic-5 (Fig. 4, lower panel) in adult rat tissues was similar to but not identical with the distribution of paxillin (Fig. 4, upper panel) in these tissues. As shown in the lower panel of Fig. 4, hic-5 was strongly expressed in rat large intestine, lung, spleen, testis, and uterus; moderately expressed in brain, kidney, and liver; and undetectable in heart, pancreas, small intestine, and skeletal muscle. Tissues that contained hic-5 generally contained paxillin and vice versa, but the relative amounts of these proteins varied within these tissues. For example, although both proteins were present in liver and large intestine, more paxillin is present in liver than in large intestine, while there were  hic-5/ARA55 is an hsp27 Binding Protein vastly greater amounts of hic-5 in large intestine than in liver (Fig. 4).
Mapping of the Interaction Regions between hsp27 and hic-5-Both hsp27 and hic-5 contain multiple domains that have been shown to be responsible for protein⅐protein interactions. For example, the C-terminal region of the ␣B-crystallin domain of hsp27 is required for the formation of oligomers of hsp27 and other small heat shock proteins and for binding to the recently identified hsp27-BP, PASS 1 (25). The N-terminal LD repeats of hic-5 have been reported to interact with FAK and CAK (32)(33)(34), while the C-terminal LIM domains have been reported to bind to the tyrosine protein-phosphatase PEST (35) and to steroid receptors (36,37). In addition, the LIM domains of hic-5 also bind to specific DNA sequences with high GϩA content and/or a long A/T tract (38).
To determine the critical regions in both hsp27 and hic-5 for their interaction, several C-terminal Myc-tagged hsp27 and N-terminal FLAG-tagged hic-5 truncation-mutant constructs were designed and prepared (Fig. 5). We were unable to express the truncation-mutant proteins consisting of either the N-terminal 74 or the C-terminal 66 amino acids of hsp27, or the hic-5 truncation lacking three of the four C-terminal LIM domains in 293T cells. We chose to use 293T cells for our studies both because they are a line of kidney epithelial cells that can be transfected with high efficiency and because there is no detectable basal expression of hic-5 (Fig. 4B, lane 1) or hsp27 (Fig. 7B,  lane 1), although paxillin is constitutively expressed in these cells (Fig. 4A, lane 1). Therefore, 293T cells permitted manipulation of the interaction between hsp27 and hic-5 using truncation mutants without the interference of endogenous proteins. Using the truncation constructs that could be expressed at comparable levels as analyzed by Western blotting (Fig. 6B,  lanes 5-8 and Fig. 6C, lanes 2 and 3), we found that deletions from the C-terminal end of the protein of all four LIM domains (FLAG-hic-5-N), the last two LIM domains (FLAG-hic-5⌬LIM3/ 4), or only the last LIM domain (FLAG-hic-5⌬LIM4) abolished the ability of hic-5 to co-immunoprecipitate hsp27 (Fig. 6A,  lanes 5, 7, and 8). However, an hic-5 C-terminal mutant consisting of only the four LIM domains still interacted with hsp27 (Fig. 6A, lane 6). We also determined that a C-terminal region of hsp27 containing the entire ␣B-crystallin domain was sufficient for interaction with hic-5 (Fig. 6A, lane 3). Because the smaller hsp27 truncation mutants could not be expressed in 293T cells, we were unable to further define the portion of this region responsible for interaction with hic-5. These observations were consistent with our yeast two-hybrid result identi-fying the C-terminal half of hic-5 as a hsp27-binding peptide and provided confirmation that the hic-5 LIM domains, particularly the fourth LIM domain, and the C-terminal region of hsp27 containing the entire ␣B-crystallin domain are both necessary and sufficient for the interaction of hic-5 with hsp27.
Effect of hic-5 on hsp27 Cellular Stress Protection Properties-To determine whether hsp27⅐hic-5 interaction plays a role  A, results of immunoprecipitation with anti-FLAG antibody using protein extracts of 293T kidney epithelial cells co-transfected with full-length FLAG-hic-5 and various hsp27-myc truncation mutant constructs or co-transfected with full-length hsp27-Myc and various FLAG-hic-5 truncation mutant constructs. This analysis indicates that the C-terminal region of hsp27 (including the entire ␣B-crystallin domain) and the C-terminal region of hic-5 (including all four LIM domains) are necessary and sufficient for hic-5-hsp27 interaction. B and C, Western blots of the cell extracts used for the immunoprecipitations in A to demonstrate that the various transfected genes were expressed at comparable levels. Transfection constructs are indicated at the top, and the position of migration of various full-length or truncation mutant proteins is indicated at the right. hic-5/ARA55 is an hsp27 Binding Protein in the function of hsp27, we analyzed the effect of hic-5 overexpression on the known protection against thermal stress provided by hsp27 overexpression. 293T kidney epithelial cells were transiently transfected with vector pcDNA3.1 as a control, or with hsp27-Myc, FLAG-hic-5, or both hsp27-Myc and FLAG-hic-5 constructs. As expected, cells transfected with the hsp27-Myc construct accumulated a substantial quantity of this protein by 44 h after transfection (Fig. 7B, lane 2), and these cells exhibited a significantly increased tolerance to a subsequent heat shock (Fig. 7A, see survival of hsp27 transfectants at 45 and 46°C). Cells transfected with the FLAG-hic-5 construct also accumulated significant quantities of this protein (Fig. 7B, lane 3), but this transfection alone had no significant effect on cell survival after heat shock at 45 and 46°C. Cells co-transfected with both FLAG-hic-5 and hsp27-Myc constructs accumulated both proteins (Fig. 7B, lane 4) but displayed a significantly reduced ability to survive heat shock compared with cells transfected with the hsp27-Myc construct alone (p Ͻ 0.001). The neutral red dye uptake of cells heat shocked at 47°C for 30 min represents the background absorbance value, because Ͼ98% of these cells were dead as confirmed by trypan blue dye uptake as well as a lack of ability to proliferate in culture after the heat shock (data not shown). Western blot analyses of cells at the time of neutral red assay (14 h after heat shock) demonstrated that both FLAG-hic-5 and hsp27-Myc continued to be present in amounts comparable to preheat shock values in these cells (Fig. 7B, lanes 6 -8, compare with Fig. 7B, lanes 2-4) and exhibited an expected increase in the amount of endogenous hsp27 present in cells after heat shock (Fig. 7B, lanes 5-8).
To further confirm that the binding of hic-5 to hsp27 inhibits hsp27 thermo-protection, a similar experiment was performed with the substitution of the hic-5 truncation mutant lacking the last LIM domain (FLAG-hic-5⌬LIM4) for FLAG-hic-5. As demonstrated in the domain mapping studies, deletion of LIM4 abolished the interaction of hic-5 with hsp27 as assessed by co-immunoprecipitation. FLAG-hic-5⌬LIM4 accumulated in transfected cells (Fig. 8B, lanes 3 and 4), but, in contrast to the results using FLAG-hic-5, cells co-transfected with hsp27-Myc and FLAG-hic-5⌬LIM4 constructs exhibited the same increased resistance to thermal stress as cells transfected only with the hsp27-Myc construct (Fig. 8A, 45 and 46°C). Trans-fection with the FLAG-hic-5⌬LIM4 alone had no significant effect on cell survival after heat shock (Fig. 8A, 45 and 46°C). DISCUSSION In the present study, we identified hic-5, a protein related to the focal adhesion protein paxillin, as a novel hsp27-BP. hic-5 was originally isolated as a senescence-inducing gene from mouse osteoblastic cells treated with transforming growth factor ␤1 and hydrogen peroxide (31). Later, hic-5 was identified independently by several other groups with different research interests and has been described as a focal adhesion protein (34,39), an androgen and glucocorticoid receptor co-activator (36,37), and a negative regulator of muscle differentiation (40). Studies from several groups, including ours, have previously demonstrated that hsp27 binds to proteins in addition to itself or other small hsps. We previously reported that hsp27 binds specifically to a novel protein, PASS1, which is highly expressed in Sertoli cells (25), and to hsp22, a new member of the small heat shock protein superfamily (41). hsp27 has also been shown to interact with platelet transglutaminase (23) and to bind to and activate PKB/Akt (42). In addition, Drosophila hsp23, an hsp27 homolog, binds to ubiquitin-conjugating enzyme (DmUbc9) (24). These results together suggest that both hsp27 and hic-5 play important roles in fundamental biological processes through interaction with specific binding partners and suggest that the interaction between hsp27 and hic-5 may define new functional roles for one or the other or both proteins.
Despite the significant degree of sequence identity (42%) between hic-5 and the focal adhesion adaptor protein paxillin, our studies failed to indicate an interaction between hsp27 and paxillin (Fig. 2, lane 9). These results suggest that the hsp27⅐hic-5 interaction has a significant degree of specificity and that hic-5 has cellular functions that are distinct from that of paxillin. This potential functional distinction between hic-5 and paxillin was further supported by our finding of overlapping but distinct tissue expression patterns for hic-5 and paxillin in normal rats (Fig. 4).
To better understand the molecular basis for the hsp27⅐hic-5 interaction, we developed a series of truncation mutants of both genes and transfected 293T cells to study the domains required for protein⅐protein interaction in co-immunoprecipitation studies. These studies identified the C-terminal ␣B-crystallin do-FIG. 7. Overexpression of hsp27 but not hic-5 protects 293T cells against heat shock, while co-transfection inhibits hsp27 thermo-protection. A, neutral red dye uptake assay was used to determine cell survival of 293T cells transfected with various constructs after no treatment or heat shock at 45, 46,  main of hsp27 and the C-terminal LIM4 domain of hic-5 as the critical interaction domains. These results are consistent with the known importance of the ␣B-crystallin domain of hsp27 (27) and the LIM domains (43,44) in protein⅐protein interactions.
Identification of hic-5 as an hsp27-BP has potentially broad biological implications, given the similar cellular activities for which both hic-5 and hsp27 have been reported to have roles. hsp27 has been reported to regulate actin polymerization, particularly under conditions of cellular stress known to induce actin filament disruption (7,15), whereas hic-5 has been localized to focal adhesions where actin filaments are anchored to the cell membrane (32)(33)(34)39). hic-5, like paxillin, has been reported to bind to both focal adhesion structural proteins such as vinculin (34) and signaling proteins such as FAK, CAK␤, and p95PKL/PIX/PAK (32)(33)(34)45), suggesting hic-5 may have an important role as an adaptor protein in the regulation of integrin-initiated actin filament reorganization. Further support for this hypothesis was recently reported in a publication showing that hic-5 overexpression can reduce the rate of integrin-mediated cell spreading, presumably by competing with paxillin for the sites on paxillin binding proteins and by inhibiting paxillin tyrosine phosphorylation (46). In light of these results and the present data demonstrating that hic-5 binding to hsp27 can alter hsp27 cytoprotective function, we speculate that the interaction of hsp27 and hic-5 may have important functional consequences to the regulation of focal adhesion dynamics and cell-matrix interactions, as well as regulation of androgen and glucocorticoid-dependent gene expression.
In contrast to these similarities, hsp27 and hic-5 may have different roles during tumorigenesis and cell differentiation. Elevated hsp27 expression has been reported in human cancer tissues and malignant cells of human origin (47)(48)(49), whereas hic-5 is markedly down-regulated in immortalized cells and cultured cell lines derived from human tumors (31,50,51). In addition, hic-5 overexpression results in development of a senescence-like phenotype in cultured cells (31,50) and increased apoptosis in cells of muscle origin (40), whereas hsp27 expression prevents apoptosis and enhances cell survival (2, 52) as well as altering cell proliferation and differentiation in a cell type-dependent manner (53)(54)(55)(56). Based on our demonstration of the ability of hic-5 to alter hsp27 function, we speculate that the hic-5-hsp27 interaction could also play an important role in regulating the balance between cell proliferation and survival and cell death.
Although interaction of hic-5 with hsp27 can clearly alter hsp27 cytoprotective function, the precise mechanism is not yet clear. Our data clearly indicate that the inhibition of hsp27 thermo-protection by hic-5 is dependent on the biochemical interaction between hic-5 and hsp27. In this context, the inhibition of hsp27 function may have resulted from a hic-5-induced conformational change in hsp27 that impaired its putative chaperoning function. Alternately, this modulation of cytoprotection may have been the result of an hic-5-induced impairment of hsp27 ability to co-localize with and regulate cellular actin polymerization. Preliminary data from our laboratory has provided early evidence suggestive of the latter possibility. We have found that, compared with vector-transfected control cells, overexpression of hsp27 and hic-5 have opposite effects on total cellular polymerized actin content. 4 These findings imply opposing effects of hsp27 and hic-5 on cellular actin dynamics and suggest a potential mechanism by which hic-5 may regulate hsp27 function.
It is also possible that the hic-5-hsp27 interaction would result in alterations in hic-5 function in addition to the alterations in a known function of hsp27 reported here. One of the most likely areas within the cell that such an interaction might have biologic relevance is at focal adhesions, where hic-5 has been localized (32)(33)(34)39). One possibility is that hic-5 might serve to target hsp27 to focal adhesions where it could participate in the regulation of actin polymerization and anchoring of focal adhesions. Recruitment of hsp27 to the focal adhesions could also result in alterations (either positive or negative) in hic-5 adaptor protein function in this location. Because hic-5 has also been reported to localize to the nucleus and to function as a steroid receptor co-activator, it is possible that hsp27 might also alter hic-5 function in transmitting extracellular signals to the nucleus to stimulate changes in gene expression. Although in the present study hic-5 was originally identified 4 Y. Jia, unpublished results.

FIG. 8. Overexpression of hsp27 and hic-5DLIM4 in 293T cells does not inhibit hsp27 thermo-protection.
A, neutral red dye uptake assay was used to determine cell survival of 293T cells transfected with various constructs after no treatment or heat shock at 45, 46, or 47°C for 30 min and recovery for 14 h under normal culture conditions. Bars represent the mean of 14 independent experiments. *, p Ͻ 0.05 versus vector control cells by two-tailed, unpaired t test. B, Western blot detection of the quantity of endogenous hsp27, hsp27-Myc, and FLAG-hic-5⌬LIM4 in 293T cells transfected with the vector alone (lanes 1, 5), with hsp27- Myc (lanes 2, 6), with FLAGhic-5⌬LIM4 (lanes 3, 7), or with both hsp27-Myc and FLAG-hic-5⌬LIM4 (lanes 4, 8) constructs. Lanes 1-4 are extracts from untreated cells, lanes 5-8 are from cells heat-shocked at 46°C for 30 min and allowed to recover for a further 14 h. Position of migration of FLAG-tagged hic-5⌬LIM4, Myc-tagged hsp27, or endogenous hsp27 proteins are indicated at the left.
hic-5/ARA55 is an hsp27 Binding Protein from a rat glomerular cDNA library, based on the variable tissue distribution of hic-5 such interactions and functional alterations might result in either tissue-or even cell-specific regulation of hsp27 function.
In conclusion, we have identified the focal adhesion protein/ steroid receptor co-activator, hic-5, as a novel binding partner for the small heat shock protein, hsp27. hic-5 was identified as a putative hsp27 binding protein by yeast two-hybrid screening, and we confirmed a biochemical interaction by co-immunoprecipitation of hic-5 and hsp27 from transfected 293T cells overexpressing both proteins and from ASC-17D cells, which have significant endogenous expression of hic-5 and hsp27. We have also mapped the critical molecular domains required for this interaction to the C-terminal LIM4 domain of hic-5 and the C-terminal ␣B-crystallin domain of hsp27. In addition, evidence was provided of differences in tissue expression and hsp27 binding between hic-5 and the related protein, paxillin, suggesting these proteins have distinctive functions and may be distinctly regulated within cells and/or tissues. Finally, a functional analysis of the hsp27⅐hic-5 interaction revealed that hic-5 has a direct inhibitory effect on the cytoprotective function of hsp27, and that this inhibitory effect is dependent on the biochemical interaction of hic-5 with hsp27. These findings together suggest that hic-5 may have an important biologic role in regulating hsp27 function and possibly in targeting hsp27 to focal adhesions where it could participate in the regulation of focal adhesion dynamics.