ABSTRACT
Src family kinases are implicated in cellular proliferation and transformation. Terminally differentiated myocytes have lost the ability to proliferate, indicating the existence of a down-regulatory mechanism(s) for these mitogenic kinases. Here we show that feline cardiomyocyte lysate contains thermostable components that inhibit c-Src kinase in vitro. This inhibitory activity, present predominantly in heart tissue, involves two components acting combinatorially. After purification by sequential chromatography, one component was identified by mass and nuclear magnetic resonance spectroscopies as 5′-AMP, while the other was identified by peptide sequencing as a small heat shock protein (sHSP). 5′-AMP and to a lesser extent 5′-ADP inhibit c-Src when combined with either HSP-27 or HSP-32. Other HSPs, including αB-crystallin, HSP-70, and HSP-90, did not exhibit this effect. The inhibition, observed preferentially on Src family kinases and independent of the Src tyrosine phosphorylation state, occurs via a direct interaction of the c-Src catalytic domain with the inhibitory components. Our study indicates that sHSPs increase the affinity of 5′-AMP for the c-Src ATP binding site, thereby facilitating the inhibition. In vivo, elevation of ATP levels in the cardiomyocytes results in the tyrosine phosphorylation of cellular proteins including c-Src at the activatory site, and this effect is blocked when the 5′-AMP concentration is raised. Thus, this study reveals a novel role for sHSPs and 5′-AMP in the regulation of Src family kinases, presumably for the maintenance of the terminally differentiated state.
Terminally differentiated cells exit the cell cycle and do not reenter the cell cycle, even in the face of growth stimulation (reviewed in reference 42). Adult cardiomyocytes (cardiocytes) have not only lost the ability to proliferate but also become resistant to neoplastic transformation. The mechanism responsible for maintaining the terminally differentiated state in cells such as cardiocytes, which lack the ability to compensate for cell loss in disease states such as myocardial infarction, is not well understood. As a first step toward identifying this mechanism, we investigated whether adult cardiocytes contain novel factors that can suppress mitogenic kinases such as the Src family kinases, which are known to play important roles in both cellular proliferation and transformation.
The Src family of nonreceptor tyrosine kinases was initially identified and studied for its role in cellular transformation (37, 57), and elevated kinase activity of v-Src has been positively correlated with cell transformation (27). Further investigations have implicated Src kinases in regulating several vital cellular functions including proliferation, differentiation, transformation, and morphologic alterations, though the responsible mechanisms are not entirely known (reviewed in reference5). In studies of proliferation, higher levels of active c-Src were found in growth factor-stimulated cells which were rapidly transiting the cell cycle (48), and specific members of this kinase family have been shown to be required for the G0/G1-to-S transition (48, 34) as well as the G2-to-M phase transition (10).
During cardiac and skeletal muscle development, the interlinked programs of decreased proliferative ability and onset of differentiation culminate in the expression of muscle-specific genes. Reactivation of v-Src represses the transcription of muscle-specific genes in postmitotic quail myotubes (20), resulting in disruption of the myofibrillar architecture, although these changes were not sufficient to induce proliferation (9). These studies indicate that activated c-Src cause loss of maintenance of the differentiated state, independent of its proliferative role. Therefore, the activity of c-Src must be curtailed in order to permit transcription of muscle-specific genes and to maintain the phenotypic characteristics of differentiated myotubes. Hence, it is possible that the Src family kinases are firmly regulated for the sarcomeric protection of muscle cells.
Src family kinases are regulated dynamically by phosphorylation and dephosphorylation events on specific tyrosine and serine/threonine residues. The functional significance of serine/threonine phosphorylation in the unique domain at Ser-12, Ser-17, Thr-34, Thr-46, and Ser-72 is not well defined (reviewed in reference5). However, the effects of tyrosine phosphorylation in the catalytic domain at Tyr-416 and in the C-terminal loop at Tyr-527 have been well characterized (50, 54). A major mechanism of Src regulation involves reciprocal phosphorylation of these two tyrosine residues, leading to a switch from the closed to the open configuration and vice versa (reviewed in reference17). Phosphorylation of c-Src by (C-terminal Src kinase) at the Tyr-527 site (43) results in the closed configuration (51). For kinase activation, cellular proteins interact with SH2 and SH3 domains of Src, resulting in displacement of the C-terminal tail and consequently an open configuration (reviewed in reference 17). In the open configuration, c-Src undergoes autophosphorylation in the kinase domain at Tyr-416, resulting in enhanced kinase activity (54). Interestingly, our recent study (33) of the adult feline heart shows an absence of tyrosine phosphorylation for c-Src, indicating that this kinase is present in the inactive form via a Tyr-527 phosphorylation-independent mechanism. However, in the actively hypertrophying myocardium, we found that c-Src is recruited to the cytoskeleton, where it is present in the active (Tyr-416-phosphorylated) form. These observations suggest that in the normal adult heart a Tyr-527 phosphorylation-independent mechanism(s) may also control c-Src activity. In other phosphorylation-independent mechanisms of c-Src, the molecular chaperones HSP-90 (90-kDa heat shock protein) and pp50 have been shown to form a complex with nascent v-Src, resulting in altered functional activity (6, 7), and have been shown to play a role in oncogenic transformation (63). In addition, caveolin (38) and RACK1 (11) have been shown to inhibit c-Src activity.
Since the role and regulation of Src family members are distinctly different in terminally differentiated cells, we examined whether cardiocytes contain factors that could suppress Src family kinases by a novel mechanism. We report here that (i) adult feline cardiac tissue and cell lysates contain thermostable components that preferentially inhibit Src family nonreceptor kinases, (ii) the inhibitory activity is higher in heart than in other tissues, (iii) this inhibition on c-Src involves a Tyr-527 phosphorylation-independent mechanism, (iv) the inhibitory activity is mediated by two components acting in combination that were identified as 5′-AMP and a specific member of the small HSP (sHSP) family including HSP-27 and HSP-32, (v) the catalytic domain of c-Src interacts directly with the inhibitory components, and (vi) elevated ATP levels promote Src activation that can be blocked by 5′-AMP in vivo. Thus, this study shows for the first time a new role for sHSPs in suppressing the nonreceptor tyrosine kinases, especially the Src family kinases, in a 5′-AMP/5′-ADP-dependent manner. We propose that this mechanism may have a potential role in the maintenance of the differentiated state of cardiocytes.
MATERIALS AND METHODS
Chemicals.Aprotinin, 1-4-dithiothreitol (DTT), E-64 [trans-epoxysuccinyl-l-leucylamido-(4-guanidino)butane], and phenylmethylsulfonyl fluoride (PMSF) were purchased from Boehringer Mannheim, Mannheim, Germany. p-Aminobenzamidine, enolase, leupeptin, EGTA, β-glycerophosphate, laminin, myelin basic protein (MBP), nucleotides, pepstatin, agarose-conjugated pronase, sodium dodecyl sulfate (SDS), sodium orthovanadate, subtilisin, Triton X-100, and fusion protein corresponding to the proline-rich sequence of middle T antigen were from Sigma Chemical Co. (St. Louis, Mo.); AMP and other nucleotide analogs were from either Sigma or ICN Pharmaceuticals Inc. (Costa Mesa, Calif.); HSPs were from Stressgen Corp. (Victoria, British Columbia, Canada); acetonitrile and trifluoroacetic acid (TFA) were from Pierce (Rockford, Ill.); [γ-32P]ATP was obtained from Du Pont (Boston, Mass.); and the baculovirus expression system was from Novagen, Inc. (Madison, Wis.).
Antibodies.c-Src-specific monoclonal antibody GD11 was purchased from Upstate Biotechnology, Inc. (Lake Placid, N.Y.), whereas anti-c-Src N-terminus antibody N-16 was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, Calif.). Rabbit polyclonal anti-phospho-Tyr-416 c-Src antibody was a generous gift from New England BioLabs, Inc. (Beverly, Mass.). Horseradish peroxidase-labeled antiphosphotyrosine antibody PY20 and agarose-coupled anti-v-Src antibody (monoclonal) were obtained from Transduction Laboratories (Lexington, Ky.) and Calbiochem (La Jolla, Calif.), respectively. Anti-HSPs were purchased either from Upstate Biotechnology or Stressgen. Horseradish peroxidase-conjugated secondary antibodies were purchased from Vector Laboratories (Burlingame, Calif.).
Cell lysate preparation.Cardiomyocytes were isolated from adult feline ventricles as described previously (15). For lysate preparation, ∼108 cardiocytes were homogenized by sonication five times intermittently, each for a duration of 30 s, in 10 ml of ice-cold NP-40 lysis buffer that contained (in final concentrations) 30 mM Tris (pH 7.4), 0.1% NP-40, 150 mM NaCl, 50 mM NaF, 1 mM Na3VO4, 0.5 mM PMSF, 10 μg of leupeptin per ml, 10 μg of aprotinin per ml, 2 μg of pepstatin A per ml, 1 mM DTT, 2 μM E-64, 200 μg ofp-aminobenzamidine per ml, and 1 mM EGTA. The homogenate was centrifuged at 150,000 × g for 30 min, and the supernatant was immediately heat inactivated at 100°C for 10 min. After repetition of the above centrifugation step, the heat-stable supernatant was dialyzed overnight in 10 mM Tris (pH 7.4) buffer at 4°C, using a 1-kDa-cutoff dialysis membrane. The dialyzed material (referred to hereafter as the cardiocyte lysate) was filtered through 10-kDa-cutoff Centriplus filters. The filtrate (referred to as the <10-kDa fraction) and the concentrated material (referred to as the >10-kDa fraction), which was washed twice with phosphate-buffered saline and once with 20 mM Tris (pH 7.4), were used for assaying c-Src inhibitory activity. For tissue studies, cerebrum, pectoral muscle, liver, spleen, kidney, small intestine, and cardiac ventricular tissues were obtained from an adult cat. Each tissue (5 g) was minced and homogenized with a Tekmar Tissumiser in 25 ml of lysis buffer, sonicated briefly, and processed to obtain dialyzed heat-stable lysates as described for cardiomyocytes. For the preparation of detergent-free lysates, 15 g of adult feline ventricular tissue was homogenized in a Waring blender with 100 ml of lysis buffer that contained no NP-40 detergent. The lysate was processed as detailed above, and the thermostable supernatant was filtered through a 10-kDa-cutoff Centriplus filter to obtain the >10-kDa fraction.
The protein contents of the heat-stable cardiac and other tissue lysates and their >10-kDa samples were quantitated by the Bradford assay (Pierce). The concentration of the <10-kDa fraction was determined by measuring UV absorbance at 259 nm. A concentration of 1 U was arbitrarily assigned for the <10-kDa fraction having 3.0 absorbance units (AU).
Measurement of kinase activity.The kinase activity of purified human recombinant and native forms of c-Src (purchased from Upstate Biotechnology and Calbiochem, respectively) was determined in terms of either autophosphorylation or external peptide/protein phosphorylation. The synthetic substrate peptide (KVEKIGEGTYGVVYK), based on the sequence of p34cdc2 kinase (12), was synthesized at the MUSC peptide synthesis facility, and the acid-denatured enolase protein substrate was prepared as described previously (16). Kinase activity was measured as described previously (32) in a total volume of 18 μl. The reaction mixture contained (in final concentrations) 10 mM Tris (pH 7.4), 50 mM β-glycerophosphate, 0.1% bovine serum albumin, 0.9 units (∼1 ng of protein) of c-Src kinase (human recombinant; Upstate Biotechnology), and 1 or 10 μM ATP {[γ-32P]ATP (3 × 106 to 6 × 106 cpm/assay)}. The tubes were incubated for 15 min at 30°C in either the absence (for autophosphorylation) or presence of 1.8 μg of acid-denatured enolase (external substrate phosphorylation). After the addition of SDS sample buffer to arrest the reaction, the samples were resolved on precast gels (Novex, San Diego, Calif.) and autoradiographed. In some experiments, the p34cdc2 peptide (final concentration, 2 mM) was used as an external substrate for measuring c-Src kinase activity. In this case, after the 15-min kinase reaction, the radioactivity associated with the peptide was counted by spotting the reaction mixture directly onto a square of Whatman P-81 paper as described previously (32). To assess the activities of Fyn, Lyn, epidermal growth factor receptor (EGF-R; Upstate Biotechnology), and v-Abl (Calbiochem), autophosphorylation was measured by as described for c-Src except that ERK-2 and p34cdc2 kinase (Upstate Biotechnology) activities were measured at 10 μM ATP by using MBP at a final concentration of 0.3 μg/μl as that substrate. The phosphorylated MBP was resolved by SDS-polyacrylamide gel electrophoresis (PAGE) and subjected to autoradiography. To study the effects of nucleotides and HSPs on kinase activity, indicated amounts of these materials were also included in the kinase reaction.
Mass and NMR spectral analyses.The purified <10-kDa sample and the nucleotides 5′-AMP, 5′-ADP, and ATP were analyzed by electrospray ionization (ESI) mass spectrometry in the positive ion mode with a Perkin-Elmer Sciex API 300 mass spectrometer (W. Alton Jones Cell Science Center, Lake Placid, N.Y.). 1H nuclear magnetic resonance (NMR) spectra of the purified <10-kDa sample and AMP isomers were analyzed on a Nicolet NT 360 Spectrometer (Spectral Data Services, Inc., Ill.).
Determination of amino acid sequence.Direct sequencing of the purified sample from Cibacron blue column indicated N-terminally blocked proteins. Therefore, the sample was treated for 12 h in the dark with 5% CNBr solution in 88% formic acid to cleave proteins after their methionine residues and to generate peptide fragments. The digested material was subjected to automated Edman sequencing using a Procise 494 protein sequencer (PE/Applied Biosystems) at the MUSC sequencing facility. The sequences thus obtained were compared with sequences in the protein database (ProteinInfo, Swiss-PROT) to determine the identity.
Expression and purification of c-Src mutant proteins.Mouse cDNA for the constitutively active form of c-Src (pUSE Src Y529F [obtained from Upstate Biotechnology] in which the Tyr-527 site, which corresponds to Tyr-529 in mouse c-Src, is replaced with phenylalanine) was used to transfect Cos-7 cells according to the manufacturer’s protocol. For this, Cos-7 cells grown in 100-mm-diameter tissue culture plates were treated with 100 μl of Lipofectin in 10 ml of medium with c-Src cDNA or empty vector (10 μg) and transfected for 5 h. Transfected cells were cultured for 2 days before harvesting for immunoprecipitation.
For baculoviral expression of the c-Src catalytic domain (amino acids 279 to 522), the cDNA for the constitutively active form of c-Src, pUSE Src Y529F, was PCR amplified by using appropriate primers. The fragment generated was subcloned into the baculovirus shuttle vector pBAC gus-1 and expressed along with a C-terminal hexahistidine tag in Sf9 cells as instructed by the manufacturer (Novagen). The expressed catalytic domain was purified by using a Ni-nitrilotriacetic acid spin column from Qiagen (Valencia, Calif.) and concentrated by passage through Amicon filters.
Western blotting and immunoprecipitation.Aliquots of protein samples (5 to 20 μg of cellular proteins or 1 ng of recombinant c-Src) were boiled with SDS sample buffer, resolved by SDS-PAGE, and transferred onto Immobilon-P membranes (Millipore Corp., Bedford, Mass.). The blots were blocked for 1 h with 10% nonfat dry milk in TBST (10 mM Tris [pH 7.4], 130 mM NaCl, 0.1% Tween). Anti-HSPs were incubated for 2 to 4 h at room temperature or for 12 h at 4°C, washed five times, each for 5 min, in TBST, and then incubated with the appropriate horseradish peroxidase-conjugated secondary antibody. After the wash with TBST, proteins were detected by enhanced chemiluminescence reagent obtained from New England Nuclear (Boston, Mass.).
For immunoprecipitation-coupled kinase activity, Cos-7 cells expressing either empty vector or constitutively active c-Src were extracted with NP-40 lysis buffer as described above. The extract (1 mg of protein) was mixed with 20 μl of agarose-coupled monoclonal anti-v-Src antibody (reacts with both c-Src and v-Src) for 4 h at 4°C, washed three times with lysis buffer and twice with the kinase assay buffer (30 mM Tris [pH 7.4], 10 mM β-glycerophosphate, 1 mM DTT, 1 mM EGTA), and used for measuring autophosphorylation.
Electroporation.For each condition, 2 × 105 adult feline cardiocytes were cultured on laminin-coated 35-mm-diameter dishes for 18 h, using medium 199 (Life Technologies, Grand Island, N.Y.) containing 0.1% bovine serum albumin and no serum. After replacement of the medium with electroporation medium (62) containing or not containing nucleotides, the cardiocytes were electroporated in a microporator (59) at three independent areas in each plate for 30 ms at 200 V. After a recovery period of 20 min at 4°C, the cells were harvested, washed in phosphate-buffered saline, and then used for the cytoskeletal preparation as described previously (33).
Protease treatment.Both >10-kDa and <10-kDa fractions were lyophilized and reconstituted in 230 μl of 10 mM sodium acetate buffer (pH 7.5) containing 5 mM calcium chloride to obtain final concentrations of 1.3 μg of protein per μl and 1.6 U of solution, respectively. Either 25 enzyme units of subtilisin or 1.5 enzyme units of agarose-conjugated pronase was added, and the mixture was incubated for 30 min at 30°C. After removal of the pronase beads by centrifugation, the protease-treated samples were incubated at 100°C for 10 min in order to eliminate the protease activity. To arrest any residual protease activity, a cocktail of protease inhibitors (in final concentrations, leupeptin [10 μg/ml], aprotinin [10 μg/ml], pepstatin A [2 μg/ml], EGTA [1 mM], and PMSF [0.5 mM]) was added before the samples were used to assay the inhibitory activity on c-Src.
Purification and identification of inhibitory components.Purification of the inhibitory components was performed with a Bio-Rad Biologic FPLC (fast protein liquid chromatography) system. In the case of the <10-kDa fraction, the lyophilized sample obtained from cardiocytes was reconstituted in 0.1% TFA and applied to a reverse-phase column (Pep-RPC HR 5.5; Pharmacia Biotech, Piscataway, N.J.) equilibrated in 0.1% TFA. The adsorbed material was eluted with a 100-ml linear gradient from 0 to 50% acetonitrile in 0.1% TFA. Column fractions (1 ml) were lyophilized, dissolved in H2O, and assayed for c-Src inhibitory activity in the presence or absence of cardiocyte >10-kDa fraction. Fractions showing combinatorial inhibition of c-Src were pooled, concentrated to 200 μl, and applied to a molecular sizing column (Superdex; Pharmacia) equilibrated in 0.1% TFA. The column was precalibrated with known molecular mass standards: adenosine (267 Da), GMP (407 Da), ATP (551 Da), atriopeptin (2,083 Da), and hypercalcemia of malignancy factor (4,016 Da). Fractions were lyophilized and reconstituted in H2O to measure c-Src inhibitory activity in the presence and absence of the >10-kDa fraction. Based on the peak fractions showing combinatorial inhibitory activity, molecular size was estimated. The above chromatographic steps were repeated using feline ventricular <10-kDa fraction in order to obtain sufficient material for mass and NMR spectral analyses.
For purification of the >10-kDa inhibitory component(s), detergent-free lysate obtained from 12 g of cardiac tissue was heat inactivated to prepare thermostable lysate, and the >10-kDa fraction (50 mg of protein) was prepared as described for cardiocytes. For molecular size determination, the >10-kDa fraction (∼1 mg of protein) was chromatographed on an anion-exchange column (Mono Q; Pharmacia). Flowthrough fractions containing the inhibitory protein component were pooled, dialyzed against 20 mM Tris (pH 7.4) containing 50 mM NaCl and 1 mM DTT, and chromatographed on a molecular sizing column (Superose 12; Pharmacia). The column was precalibrated by using the molecular mass standards (Bio-Rad) thyroglobulin (670 kDa), gamma globulin (158 kDa), ovalbumin (44 kDa), myoglobin (17 kDa), and vitamin B12 (1.35 kDa).
For identification, part of the >10-kDa fraction (∼40 mg of protein) was applied to a gel filtration column (Superose 6; Pharmacia) equilibrated with 3 mM sodium phosphate buffer (pH 6.3). Peak fractions showing combinatorial inhibitory activity with cardiocyte <10-kDa fraction were pooled, concentrated, and loaded onto a hydroxyapatite column (CHT10-I; Bio-Rad) equilibrated with 3 mM sodium phosphate buffer (pH 6.3). The column was washed with 6 ml of 3 mM sodium phosphate buffer (pH 6.3), and bound proteins were eluted with a 50-ml linear gradient from 3 mM sodium phosphate (pH 6.3) to 150 mM sodium phosphate (pH 8.5). Peak fractions showing combinatorial inhibitory activity were eluted approximately at 30 mM sodium phosphate (pH 7.0). The inhibitory fractions were pooled, concentrated to 1 ml in Centriplus concentrators, and applied to a reverse-phase column (Pro-RPC HR5/5; Pharmacia) equilibrated with 0.05% TFA in H2O. The column was washed with 2 ml of 0.05% TFA and then eluted with a 40-ml linear gradient from 0 to 70% acetonitrile in 0.05% TFA. Peak fractions showing combinatorial inhibitory activity eluted at 40% acetonitrile were pooled and reconstituted in 3 mM sodium phosphate buffer (pH 6.3), using Centriplus filters, before loading to a dye affinity column (Cibacron blue 3G-A; Pharmacia). The column was washed with 3 ml of 3 mM sodium phosphate buffer, pH 6.3, and bound proteins were eluted with a 12-ml linear gradient from 0 to 1 M NaCl in 3 mM sodium phosphate buffer (pH 7.5). Proteins (∼4 μg) demonstrating combinatorial inhibitory activity that were eluted approximately at 700 mM NaCl were pooled and used for amino acid sequence analysis. For details on the amount of proteins loaded to and eluted from each column, see Fig. 4B.
RESULTS
c-Src is inhibited by a combinatorial action of at least two components in the cardiocyte lysate.To determine whether the inactivation of Src family members in terminally differentiated cells is mediated by any novel means, we used thermostable lysate prepared from adult feline cardiac muscle cells (cardiocytes). Thermostable inhibitory proteins have been reported for cyclin-dependent kinases (47). Kinase activity was measured both as autophosphorylation and as substrate phosphorylation in the presence or absence of the cardiocyte lysate. In the absence of dialyzed cardiocyte lysate, c-Src undergoes autophosphorylation (Fig.1A) and phosphorylates acid-denatured enolase. However, both autophosphorylation and enolase substrate phosphorylation were markedly reduced in the presence of 0.1 μg of cardiocyte lysate per μl. Similar results were also seen with purified native c-Src prepared from platelets (data not shown). To demonstrate that c-Src inhibition is dose dependent, a synthetic peptide substrate bearing sequence homology with p34cdc2 kinase (12) was incubated with various concentrations of the cardiocyte lysate, and kinase activity was determined (Fig. 1B). The inhibition was dose dependent, and 0.1 μg of protein per μl was sufficient to cause >85% inhibition. These experiments suggest that the lysate prepared from adult cardiocytes contains a potent c-Src kinase inhibitory activity. To demonstrate that the level of c-Src used in the kinase reaction is unaltered during incubation with the lysate, the samples after the reaction were resolved by SDS-PAGE, transferred to nitrocellulose membranes, and probed with anti-c-Src monoclonal antibody (Fig. 1C). The amount of c-Src (1 ng) used in the kinase reaction remains unaltered even after the incubation with cardiocyte lysate, indicating that c-Src is not hydrolyzed by proteases present (if any) in the lysate.
Demonstration of c-Src inhibitory activity in the cardiocyte lysate. (A) c-Src autophosphorylation and enolase substrate phosphorylation measured in the absence (−) or presence (+) of the cardiocyte lysate (0.1 μg/μl). The positions of c-Src (60 kDa) and enolase (45 kDa) are indicated. (B) Dose response of the effects of cardiac lysate on c-Src activity determined with p34cdc2 synthetic peptide as the substrate. (C) Activity and amount of c-Src in the kinase reaction determined by autoradiography and Western blotting (W. Blot) with c-Src monoclonal antibody GD11, respectively. (D) Cardiocyte lysate was split into <10- and >10-kDa fractions as described in Materials and Methods. Top, c-Src autophosphorylation in the absence or presence of cardiocyte lysate, the <10-kDa fraction, and/or the >10-kDa fraction; bottom, similar experiment performed with c-Src preautophosphorylated with [γ-32P]ATP, where the effect of cardiocyte fractions was measured subsequently with unlabeled ATP. (E) The <10-kDa fraction was treated with buffer, subtilisin, or pronase and then used to study their effects on c-Src autophosphorylation in the absence or presence of the untreated >10-kDa (left) or <10-kDa (right) fraction.
In our initial purification and characterization studies, a complete loss of the inhibitory activity was observed upon fractionation of cardiocyte lysate by either ion-exchange or sizing column chromatography. These results led us to suspect the involvement of more than one component for c-Src inhibition. To test this possibility, the cardiocyte lysate was processed to obtain <10- and >10-kDa fractions. These two fractions demonstrated no inhibitory activity on c-Src when assayed separately (Fig. 1D, upper panel). However, when the two were combined, the inhibitory activity was restored. These data suggest that a component(s) present in the <10-kDa fraction cooperates with a component(s) in the >10-kDa fraction to inhibit c-Src kinase activity.
Although both phosphatase and protease inhibitors were included during the preparation of cardiocyte lysate, the loss of c-Src autophosphorylation with cardiocyte fractions could be due to low-level activities of phosphatase and/or protease. To rule out such a possibility, we repeated the experiment with preautophosphorylated c-Src. For this, c-Src was autophosphorylated with [γ-32P]ATP, free ATP was removed by washing through Centriplus filters, and the phosphorylated c-Src was then used in the kinase reaction with unlabeled ATP (Fig. 1D, lower panel). Neither the cardiocyte lysate nor its fractions, separately or in combination, demonstrated any dephosphorylation or degradation of preautophosphorylated c-Src, indicating that the loss of c-Src activity is not due to protease or phosphatase activity.
To determine the nature of inhibitory components, the <10-kDa and >10-kDa fractions were subjected to protease treatment with either pronase or subtilisin and then assayed for c-Src inhibitory activity (Fig. 1E). Treatment of the <10-kDa fraction with either protease did not decrease its inhibitory activity, demonstrating that the inhibitory component in the <10-kDa fraction is not a protein. However, the combinatorial inhibitory effect of the >10-kDa fraction was completely lost upon digestion with either protease, suggesting that the inhibitory component present in the >10-kDa fraction is a protein. Taken together, these experiments indicate that adult cardiocyte lysates contain a potent c-Src inhibitory activity that is mediated by the combined action of a <10-kDa nonprotein factor(s) and a >10-kDa protein factor(s).
Components required for the inhibitory activity are present predominantly in the heart and preferentially inhibit Src family kinases.Before undertaking purification of the inhibitory components, we performed studies to (i) determine the tissue distribution of inhibitory activity and (ii) determine if the inhibitory activity is specific for Src family kinases. Thermostable lysates (>1-kDa fractions) were prepared from various feline tissue samples as detailed in Materials and Methods. When the tissue lysates (>1-kDa fractions) were assayed for c-Src kinase inhibitory activity, inhibition was predominantly observed with heart lysate (Fig. 2A, top). The absence or lower levels of c-Src inhibitory activity in other tissue lysates might be due to the absence of either the <10-kDa nonprotein component, the >10-kDa protein component, or both. To test these possibilities, the <10-kDa fraction prepared from cardiocytes was included with the tissue lysates. In the presence of the <10-kDa fraction of cardiocytes, muscle lysate demonstrated inhibitory activity; this activity was even stronger than that of heart lysate (Fig. 2A, middle). However, other tissue lysates did not show appreciable inhibitory activity. A similar assay of the tissue lysates with the addition of >10-kDa cardiocyte fraction was also performed (Fig. 2A, bottom). However, under these conditions, none of the tissue lysates demonstrated significant inhibitory activity. Overall, these experiments indicate that heart lysate contains both >10-kDa and <10-kDa inhibitory components and the muscle lysate has at least the >10-kDa protein component, and the data therefore account for the presence of inhibitory activity only in cardiac lysate (Fig. 2A, top). However, it should be noted that these experiments reflect levels of inhibitory components present in the >1-kDa heat-stable tissue lysate preparations and do not necessarily reflect their in vivo levels.
Inhibitory components are present predominantly in the heart tissue and preferentially inhibit Src family kinases. (A) Adult feline tissue samples were processed to obtain heat-stable lysates. The effect of each lysate (final concentration, 0.1 μg of protein/μl) on c-Src autophosphorylation was measured in the absence (top) or presence of cardiocyte <10-kDa fraction (final concentration, 0.16 U; middle) or >10-kDa fraction (final concentration, 0.1 μg of protein/μl; bottom). c-Src activity in the absence or presence of cardiocyte fractions is shown in the buffer lane.32P-labeled c-Src was detected by autoradiography. (B) Combinatorial effects of <10- and >10-kDa components on the autophosphorylation of c-Src, Lyn, Fyn, and EGF-R and MBP substrate phosphorylation by ERK-2 and p34cdc2 kinase. The concentrations of <10- and >10-kDa fractions were determined as described in Materials and Methods. Left, effects of increasing amounts of the >10-kDa fraction in the presence of a constant amount of the <10-kDa fraction (final concentration, 0.16 U; 1 U = 3.0 AU at 259 nm); right, effects of increasing amounts of the <10-kDa fraction in the presence of a constant amount of the >10-kDa fraction (final concentration, 0.1 μg of protein/μl). Kinase activity in the absence of both cardiocyte fractions is shown in buffer lanes. c-Src autophosphorylation was measured at either 1 (uppermost panel) or 10 (lowermost panel) μM ATP. Autophosphorylation of other tyrosine kinases was determined at 1 μM ATP, whereas the MBP phosphorylation by p34cdc2 and ERK-2 was measured at 10 μM ATP due to their higher Km for ATP. The32P-labeled tyrosine kinases and MBP were detected by autoradiography. (C) Activities of c-Src and v-Abl measured in the absence or presence of either <10-kDa fraction (final concentration, 0.16 U), >10-kDa fraction (final concentration, 0.05 μg of protein/μl), or both. After autoradiography, the bands were cut for Cerenkov counting.
To determine if the observed inhibition was specific for Src family kinases, the effects of cardiocyte inhibitory fractions on autophosphorylation of the tyrosine kinases, c-Src, c-Fyn, c-Lyn, and EGF-R, and on MBP substrate phosphorylation by the serine/threonine kinases, mitogen-activated protein kinase (MAPK) (ERK-2) and p34cdc2 kinase, were analyzed and compared with the effect on c-Src autophosphorylation (Fig. 2B). Kinase activities were measured in the absence or presence of constant amounts of <10-kDa fraction with increasing amounts of >10-kDa fraction and vice versa. In the absence of both inhibitory fractions, all of these kinases demonstrated normal phosphorylation activities, and addition of the <10-kDa or >10-kDa fraction alone did not cause inhibition. A dose-dependent inhibition of c-Src kinase activity was noticed for the >10-kDa fraction in the presence of the <10-kDa fraction (Fig. 2B, left). Such an inhibitory dose response was also seen for c-Lyn, although it required a slightly higher concentration of the >10-kDa fraction. c-Fyn demonstrated only partial inhibition at concentrations that were sufficient to almost completely inhibit c-Src. The inhibition was found to be lower for EGF-R kinase. However, the activities of both p34cdc2 kinase and ERK-2 remained unaffected at concentrations that showed strong inhibition of c-Src activity. To determine the dose response for the <10-kDa fraction, a constant amount of the >10-kDa fraction was added with increasing concentrations of the <10-kDa fraction. These assays demonstrated inhibitory dose-response patterns similar to those observed for the >10-kDa fraction. To study the combinatorial inhibitory effect on other nonreceptor tyrosine kinases, we used v-Abl (Calbiochem) and found it to be partially inhibited (Fig. 2C); i.e., compared to buffer controls, both components together caused about 73% inhibition of c-Src but only 37% inhibition of v-Abl. Since the total activities of these kinases in the assays were maintained at nearly the same level (0.9 U), the changes in the extent of kinase inhibition could not be due to a difference in their total activities. Furthermore, a fivefold-higher concentration of c-Src (4.5 U) resulted in no change in the inhibitory profile (data not shown). Thus, among the kinases that were examined, the inhibition was seen in the order c-Src > c-Lyn > c-Fyn > v-Abl > EGF-R for the tyrosine kinases and was almost absent for the serine/threonine kinases.
The <10-kDa nonprotein component is 5′-AMP.To identify the nonprotein component, the <10-kDa fraction prepared from cardiocytes was chromatographed over a reverse-phase column (Fig.3A; the chromatographic profile measured at 280 nm [top] and the combinatorial c-Src inhibitory activity [bottom] are shown). The <10-kDa inhibitory activity corresponded to a peak in the flowthrough, while the peaks eluted with the acetonitrile gradient had no detectable inhibitory activity (data not shown). This finding suggested that the inhibitory component might be a water-soluble polar molecule.
The <10-kDa nonprotein component was purified and identified as 5′-AMP. (A) The <10-kDa cardiocyte fraction was chromatographed on a Pharmacia PepRPC column (graph); fractions (Fr.) showing inhibitory activity (fractions 9 and 10) on c-Src autophosphorylation in the presence of the >10-kDa fraction are shown in the autoradiograph. (B) PepRPC inhibitory fractions were pooled and chromatographed on a Pharmacia Superdex column (graph). A major peak fraction (fraction 42) exhibiting combinatorial inhibitory effect on c-Src autophosphorylation (autoradiograph) corresponds to a molecular mass of 350 Da. The column was precalibrated as described in Materials and Methods. (C) Molecular mass determination of the purified fraction from a Superdex column (top) and 5′-AMP with purity of >99% (bottom) was performed by ESI-mass spectroscopy. The molecular mass of the purified sample was determined to be 347.8 Da, matching that of 5′-AMP. (D) 1H NMR spectra of the purified fraction from the Superdex column (top) and 5′-AMP (bottom) were determined as described in Materials and Methods. In the NMR profiles, dimethyl sulfoxide (DMSO) and H2O lines are marked, and an unmatched minor line in the purified sample is indicated by an asterisk. (E) Combinatorial c-Src inhibitory activity due to the addition of increasing amounts of either the purified Superdex column fraction (sample) or 5′-AMP, measured in the presence of cardiocyte >10-kDa fraction (0.1 μg/μl, final concentration). Nucleotide concentrations (Conc) in the purified <10-kDa fraction and in 5′-AMP samples were adjusted on the basis of absorbance at 259 nm. c-Src autophosphorylation was quantitated by Cerenkov counting (graph). (F) Inhibitory effect by the combined action of 5′-AMP and the >10-kDa fraction in assays using other nucleotide analogs. c-Src autophosphorylation was measured with increasing concentrations of AMP isomers and related nucleotides. In the absence (−) of the >10-kDa fraction, the effect on c-Src activity was measured with two higher concentrations (0.08 and 0.16 μg/μl) of nucleotides. 32P-labeled c-Src was detected by autoradiography.
To obtain further purification and determine the approximate molecular size, the fraction demonstrating peak inhibitory activity from the reverse-phase chromatography was applied to a molecular sizing column. A single prominent peak that corresponded to a molecular mass of approximately 350 Da showed c-Src inhibition only in the presence of the >10-kDa fraction (Fig. 3B). It is interesting that this small-molecular-weight component was present in significant amounts even though the total heat-stable extract had been dialyzed during early extraction procedure through a 1-kDa-cutoff dialysis membrane. Based on the molecular size, we suspected that the lower-molecular-weight component(s) was a nucleotides. To precisely identify the molecule, both mass and NMR spectroscopy analyses were performed. For this, we scaled up the entire purification procedure using feline heart tissue. The chromatography profiles obtained for the heart lysate were identical to that of cardiocyte lysate (data not shown). The major peak showing a molecular size of ∼350 Da on the molecular sizing column was then pooled, lyophilized, and used for spectroscopic analysis.
Mass spectrometry of the purified <10-kDa sample (Fig. 3C, top) indicated the presence of a major component with a molecular mass of 347.8 Da. Since this molecular mass precisely matches the size of AMP isomers, a similar analysis was performed for commercially obtained 5′-AMP. 5′-AMP demonstrated the predicted molecular mass of 347.8 Da (Fig. 3C, bottom). The minor peak corresponding to the mass of 370.2 Da is probably due to a sodium adduct of 5′-AMP. These data suggested that the inhibitory component present in the <10-kDa fraction might be 5′-AMP or its isomer. To precisely identify the low-molecular-weight inhibitory partner, NMR spectral analysis was performed (Fig. 3D, top) and compared with NMR spectra of 5′-AMP (Fig. 3D, bottom) as well as 2′-AMP and 3′-AMP (data not shown). The 1H NMR spectrum of the purified fraction was found to be similar to that of 5′-AMP (excluding the dimethyl sulfoxide and water line), whereas the profile did not match with 2′-AMP and 3′-AMP (data not shown). A minor line at 7.7 ppm in the purified sample represents low levels of some other unidentified compound(s). Nevertheless, both the mass and NMR spectra clearly indicate that the major component present in the <10-kDa fraction is 5′-AMP.
To confirm that 5′-AMP is the <10-kDa inhibitory component, we performed a dose-response assay for c-Src inhibition (Fig. 3E), using either the purified <10-kDa sample (upper autoradiogram) or commercial 5′-AMP (lower autoradiogram) in the presence of >10-kDa component. Cerenkov counting of the autophosphorylated c-Src was used to plot the dose-response curves (graph). 5′-AMP at 100 μM was sufficient to completely inhibit c-Src activity in the presence of a >10-kDa inhibitory partner at these assay conditions, and the 50% inhibitory concentration was found to be ∼8 μM. As the inhibition profile of 5′-AMP matches that of the purified sample, these experiments confirm that 5′-AMP is indeed the <10-kDa inhibitory component that cooperates with the >10-kDa fraction.
To test whether the combinatorial inhibition was mediated specifically by 5′-AMP, the effects of several commercially available nucleotide analogs were tested by c-Src kinase assay in the absence and presence of the >10-kDa fraction (Fig. 3F). 5′-AMP caused a dose-dependent inhibition of c-Src autophosphorylation only in the presence of the >10-kDa fraction; it showed no effect when added alone. Interestingly, other structural analogs (2′-AMP, 3′-AMP, 5′-IMP, and 3′,5′-cyclic AMP) did not exhibit inhibition either alone or in combination with the >10-kDa fraction. However, 5′-ADP alone was able to inhibit c-Src activity to some degree, and inclusion of the >10-kDa fraction enhanced the inhibition. Moreover, adenosine, 5′-GMP, 5′-GDP, 5′-TMP, 5′-CMP, and β-NADP did not show any inhibition either alone or in combination at similar concentrations (data not shown). These data suggest that a great degree of nucleotide specificity is involved in the inhibition of c-Src during the combined action of nucleotides with the >10-kDa fraction.
The >10-kDa protein component is a member of the sHSP family.To purify the >10-kDa component, we used a detergent-free lysis buffer to minimize membrane proteins in the preparation. Initial studies demonstrated comparable inhibitory activities in the cardiocyte lysates prepared with and without detergent (data not shown). The heating step used to obtain thermostable cardiac lysate not only reduced the protease activity significantly but also removed >80% of the heat-labile proteins. To determine the approximate molecular weight of the inhibiting protein, the >10-kDa fraction was partially purified on an anion-exchange column and passed through a precalibrated molecular sizing column (Fig. 4A). A peak fraction showing combinatorial inhibition with the <10-kDa fraction corresponded to a molecular mass of ∼18 kDa (Fig. 4A, bottom).
Purification and identification of the >10-kDa component as an sHSP. (A) The >10-kDa fraction (∼1 mg), prepared as described in Materials and Methods, was chromatographed (data not shown) on an anion-exchange column (Mono Q; Pharmacia). Flowthrough fractions (Fr.) containing the inhibitory protein component were pooled, dialyzed, and chromatographed (graph) on a molecular sizing column (Superose 12; Pharmacia). Proteins eluted at an approximate molecular size of 18 kDa showed inhibition of c-Src autophosphorylation in the presence of cardiocyte <10-kDa fraction (bottom). (B) Purification of the >10-kDa protein component by sequential chromatography. Amounts of protein loaded to and eluted from each column are indicated in boxes and brackets, respectively. Differences in the amounts of protein recovered from an earlier column and the amounts of protein loaded to the next column are due to material saved for analyses and/or losses during sample processing. Buffer conditions for each chromatographic step and elution conditions (shown in parentheses) are indicated. Briefly, the detergent-free >10-kDa fraction (40 mg) was applied to a Superose 6 column equilibrated in sodium phosphate buffer. Inhibitory proteins eluted in the molecular mass range of ∼10 to 100 kDa were applied to a hydroxyapatite column equilibrated in sodium phosphate buffer. The inhibitory fractions were pooled and applied to a Pro-RPC column equilibrated in 0.1% TFA solution. Fractions demonstrating inhibitory activity were pooled and applied to a Cibacron blue column equilibrated in sodium phosphate buffer. The inhibitory fractions were pooled and used for protein sequencing. For details, see Materials and Methods. (C) Edman sequencing of part of the purified sample from Cibacron blue chromatography performed after digestion of the purified sample with CNBr. The sequence of the digested sample shows strong homology with that of αA-crystallin and matches the conserved residues (indicated in boxes) of other sHSP family members. (D) Effects of various HSP family members on c-Src autophosphorylation in the presence (+) or absence (−) of 5′-AMP compared with that of the cardiocyte >10-kDa fraction. c-Src autophosphorylation was measured with increasing amounts of either >10-kDa fraction or HSPs. In the absence of 5′-AMP, the direct effects of these proteins on c-Src autophosphorylation were determined with two higher concentrations (0.08 and 0.16 μg/μl).32P-labeled c-Src was detected by autoradiography.
A purification scheme for the ∼18-kDa protein was then developed, and a typical purification protocol starting with 12 g of feline cardiac tissue is illustrated in Fig. 4B. We designed a series of four sequential FPLC chromatographic steps using gel filtration (Superose 6), hydroxyapatite (CHT10-I), reverse-phase (Pro-RPC), and dye affinity (Cibacron blue 3G-A) columns as detailed in Materials and Methods. The purified sample after CNBr digestion yielded a major sequence KVLGXFVE (Fig. 4C) and also a secondary sequence. Comparison with known protein sequences in the database revealed that the major sequence was homologous to αA-crystallin as well as to certain conserved residues of the sHSP-20 family members (illustrated in boxes in Fig. 4C). Although feline sHSP sequences were not available in the database for comparison, the significant sequence homology between various sHSPs led us to consider the possibility that the inhibiting protein is a feline muscle isoform of HSP-20, αB-crystallin, or HSP-27. Furthermore, Western blot analyses of the partially purified inhibitory fraction from the anion-exchange column were performed with four different antibodies against HSP-32, HSP-27, αB-crystallin, and HSP-20 (data not shown). Though these antibodies did not detect proteins of the corresponding molecular sizes in the >10-kDa fraction, antibodies for HSP-32, HSP-27, and αB-crystallin cross-reacted with a protein of ∼22 kDa. This matches approximately the molecular size of the inhibitory protein determined by the molecular sizing column (Fig. 4A). Therefore, based on the amino acid sequence and Western blot analysis, the inhibitory protein present in the partially purified sample appeared to be an sHSP. Prior sequencing attempts using partially purified hydroxyapatite fractions without CNBr digestion demonstrated the presence of FKBP-12, myoglobin, and superoxide dismutase, and analysis of these proteins (obtained commercially) showed no effects on c-Src autophosphorylation, either alone or in combination with 5′-AMP (data not shown).
In view of the significant sequence homology with HSP family members, we assayed different commercially available HSP isoforms for c-Src inhibitory activity in combination with 5′-AMP. The effects of varying the concentration (Fig. 4D) of HSPs on c-Src autophosphorylation activity were examined in the absence and presence of 5′-AMP. Neither HSP-10 (ubiquitin) nor the crystallins purified from bovine lens (αA and αB) demonstrated any inhibitory activity either alone or in combination with 5′-AMP. HSP-27 and HSP-25 (murine form of HSP-27) showed a dose-dependent inhibition of c-Src autophosphorylation only in the presence of 5′-AMP. Similarly, HSP-32 (heme oxygenase 1), a cytosolic stress-inducible sHSP, demonstrated a potent c-Src inhibitory activity only in combination with 5′-AMP. Inhibitory effects were not found for the larger HSPs, including HSP-47, HSP-70, and HSP-90, indicating that the combinatorial inhibition is restricted to a small group of structurally related sHSPs. These data demonstrate that specific members of the sHSP family can inhibit Src family kinases in the presence of 5′-AMP.
Mechanism of kinase inhibition.To explore the possible mechanism of this kinase inhibition, we first analyzed the effects of various concentrations of ATP on c-Src autophosphorylation at different concentrations of 5′-AMP as well as 5′-ADP in the absence or presence of a fixed concentration of the >10-kDa fraction (Fig.5A). Our previous experiments indicated the possibility that 5′-ADP, in addition to causing a direct inhibitory effect on c-Src, could also mediate the combinatorial inhibition in the presence of the >10-kDa fraction (Fig. 3F). At low concentrations of ATP (1 to 100 μM), the addition of increasing amounts of 5′-ADP alone resulted in decreased autophosphorylation of c-Src. This drop in autophosphorylation was more striking upon addition of the >10-kDa fraction. At a higher ATP concentration (1,000 μM), the direct inhibitory effect of 5′-ADP was abolished. However, in the presence of the >10-kDa fraction, a combinatorial inhibitory effect by 5′-ADP could still be observed. A similar experiment was performed to determine the effect of ATP concentration on the 5′-AMP-mediated combinatorial inhibition. Two distinct differences can be observed between 5′-AMP- and 5′-ADP-mediated c-Src inhibition: (i) in the absence of the >10-kDa fraction, 5′-AMP does not alter c-Src autophosphorylation even at low concentrations of ATP; and (ii) the combinatorial inhibitory effect mediated by 5′-AMP is significantly greater than that mediated by 5′-ADP. These experiments suggest that 5′-ADP-mediated direct inhibition of c-Src is not likely to occur at higher ATP concentrations and that both 5′-AMP and 5′-ADP could mediate the combinatorial inhibitory activity with HSPs by competing for the ATP binding site.
Mechanism of c-Src inhibition. (A) Various concentrations of ATP (1, 10, 100, and 1,000 μM) in the c-Src kinase reaction mixture were used to determine whether the 5′-AMP- or 5′-ADP-mediated combinatorial inhibitory activity on c-Src is competitive with ATP. The effects of these nucleotides (ADP and AMP) on c-Src autophosphorylation were analyzed either in the presence (+) or absence (−) of the cardiocyte >10-kDa fraction. c-Src autophosphorylation in the absence of nucleotides but in the presence or absence of the >10-kDa fraction is shown for controls. (B) Effect of c-Src inhibitory activity on the preautophosphorylated c-Src in cardiocyte lysate. A 15-min standard kinase assay was performed for c-Src autophosphorylation as well as α-casein substrate phosphorylation in the absence or presence of cardiocyte lysate. In part of the assay, performed in the absence of the cardiocyte lysate (last two lanes), we performed an additional 15-min kinase reaction after omitting or including the cardiocyte lysate.32P-labeled c-Src (top) and α-casein (doublet in the lower autoradiograph) were detected by autoradiography. (C) c-Src autophosphorylation time course determined under standard assay conditions with 10 μM ATP at 30°C (top) and kinase activity of both the unphosphorylated and 1-h preautophosphorylated c-Src in the absence or presence of HSP-27 and 5′-AMP for 15 min, using casein as the substrate (bottom). (D) Cos-7 cells were transfected with either the constitutively active form of c-Src, which has phenylalanine substitution at the Tyr-527 site, or an empty vector alone according to the protocol of the manufacturer (Upstate Biotechnology). c-Src was immunoprecipitated with agarose-conjugated v-Src antibody (Ab) and assayed in the presence or absence of HSP-27 and 5′-AMP. (E) Kinase activity of either c-Src (60 kDa) or a fusion protein bearing only the catalytic domain of c-Src (amino acids 279 to 522 in mouse c-Src; 28 kDa) determined in the absence or presence of 5′-AMP, HSP-27, and HSP-32, alone or in combination, as shown in the upper and lower autoradiographs. (F) c-Src autophosphorylation, performed in the absence or presence of 5′-AMP and HSP-27, alone or in combination, analyzed by Western blotting (WB) with either phosphotyrosine antibody or phospho-Tyr-416 c-Src antibody. Unphosphorylated c-Src and HSP-27 phosphorylation in the absence of c-Src (if any) are shown in the first and last lanes, respectively. Positions of the phosphorylated c-Src species with molecular masses of 60, 54, and 52 kDa are numbered 1, 2, and 3, respectively.
Since the phosphorylation at Tyr-416 of c-Src is critical for its kinase activity (54), inhibition by the combinatorial action of 5′-AMP and sHSPs could be mediated either by preventing the Tyr-416 residue from undergoing phosphorylation, by interfering directly with the catalytic activity, or by both. If the first possibility were true, then the c-Src prephosphorylated at Tyr-416 would not be susceptible to the combinatorial inhibition. To test this possibility, c-Src kinase activity was first measured for 15 min in terms of autophosphorylation as well as phosphorylation of α-casein (Fig. 5B). Inclusion of cardiac lysate results in the loss of both activities. In a second set, a 30-min kinase reaction was performed in the absence or in the presence of inhibitory components that were added after an initial 15-min reaction. Both autophosphorylation and substrate phosphorylation were more than doubled after the 30-min reaction compared to the 15-min kinase reaction. Interestingly, addition of the inhibitory components after 15 min of initial reaction arrested the kinase activity despite the fact that a portion of c-Src was Tyr-416 phosphorylated.
To further confirm that phosphorylated c-Src is equally susceptible to the inhibition, we prephosphorylated c-Src extensively and then tested its susceptibility to kinase inhibition. In the absence of any inhibitory components, a time course for c-Src autophosphorylation showed continued increase of phosphorylation for at least 1 h (Fig. 5C, top). Based on this, c-Src was preactivated for 1 h with 10 μM unlabeled ATP, and after removal of the ATP with Centricon filters, the prephosphorylated c-Src was used in the kinase assays (bottom). The extents of inhibition, measured in terms of α-casein phosphorylation by sHSP and 5′-AMP on both phosphorylated and unphosphorylated c-Src, were found to be similar: based on Cerenkov counting, 83% inhibition for unactivated c-Src and 75% for preactivated c-Src. Therefore, unphosphorylated and autophosphorylated c-Src are inhibited to similar extents, suggesting that the underlying mechanism is a direct inhibition of c-Src catalytic activity. Furthermore, we tested whether the constitutively active form of c-Src, in which the Tyr-527 residue has been substituted by phenylalanine and analogous to v-Src in terms of its transforming ability (8), is susceptible to the sHSP/5′-AMP-mediated inhibitory activity. For this, the cDNA was expressed in Cos-7 cells, and the protein after immunoprecipitation with agarose-conjugated v-Src antibody was assayed for kinase activity in the presence or absence of the inhibitory components (Fig. 5D). Compared to the activity from the control plasmid-expressing cells, constitutively active kinase showed enhanced kinase activity. This activity is not due to immunoglobulin G antibody, since the immunoprecipitation performed in the absence of cell lysate does not show any phosphorylation (Ab control lane). Importantly, these experiments clearly show that the constitutively active kinase is also susceptible to the combinatorial inhibition by HSP-27 and 5′-AMP.
Finally, to test whether the c-Src catalytic domain is directly involved in the sHSP/5′-AMP-mediated inhibition, we expressed this domain in a baculoviral system. Since in the previous experiment HSP-32 showed stronger inhibition than HSP-27 (Fig. 4D), we tested the combinatorial effect of HSP-32 and 5′-AMP on the catalytic domain in addition to HSP-27. Similar to what was observed for wild-type c-Src, kinase activity of c-Src catalytic domain was found to be inhibited significantly by the combinatorial action of HSP-27 plus 5′-AMP and HSP-32 plus 5′-AMP (Fig. 5E). Overall, these experiments demonstrate the possibility of a direct involvement of the catalytic domain with the inhibitory components although other domains either directly or indirectly might influence such interaction.
We extended these studies to explore the possible involvement of the c-Src unique domain in the kinase inhibition. In this context, it has already been shown that c-Src is often fragmented during extraction from cells and tissues, resulting in the generation of 54/52-kDa species due to N-terminal deletion of part of the unique domain (21). We tested for the presence of such fragments in the baculoviral recombinant c-Src (Upstate Biotechnology) used routinely for our experiments. c-Src was phosphorylated with unlabeled ATP in the presence or absence of HSP-27 and 5′-AMP, and the proteins were resolved on standard SDS-polyacrylamide gels for Western blot analysis (Fig. 5F) with either phosphotyrosine antibody or an antibody that specifically detects Tyr-416 phosphorylated c-Src. Whereas neither antibody detected the unphosphorylated c-Src (lane 1), both detected three proteins bands, a major 60-kDa band and two minor bands of 54 and 52 kDa, following the autophosphorylation. This result indicates that recombinant c-Src has at least three different species that can undergo autophosphorylation on the Tyr-416 site. Inclusion of either 5′-AMP or HSP-27 independently resulted in no appreciable change in the autophosphorylation. However, when both components were added, the levels of tyrosine phosphorylation for the 60- and 54-kDa c-Src species were markedly reduced. Interestingly, the 52-kDa fragment is significantly resistant to the HSP-27/5′-AMP-mediated inhibition, as evidenced by both antibodies. This 52-kDa c-Src species is detected by the anti-c-Src monoclonal antibody but not by the N-terminus-specific antibody (data not shown), indicating the N-terminal deletion in c-Src. Overall, these studies demonstrate that (i) the 52-kDa species, which probably lacks a considerable portion of the unique domain, is less susceptible to inhibition by the combinatorial action of 5′-AMP and HSP-27; (ii) the sHSP/5′-AMP-mediated inhibitory mechanism blocks the Tyr-416 phosphorylation of c-Src and complements all earlier experiments in which [γ-32P]ATP was used for detecting the phosphorylated c-Src; (iii) the antibody used for in vivo experiments (see below) is specific for the Tyr-416 phosphorylated c-Src, since it does not detect the unphosphorylated c-Src.
Changing the ATP/AMP ratio in vivo mimics the in vitro effects of c-Src inhibitory components.To demonstrate that this novel inhibitory mechanism plays a role in regulating Src kinase activity in vivo, we altered the levels of ATP and 5′-AMP in isolated cardiocytes and analyzed the overall changes in the tyrosine phosphorylation levels of proteins including c-Src. Adult feline cardiocytes were cultured on standard laminin-coated dishes and electroporated in the presence or absence of appropriate amounts of the nucleotides (see Materials and Methods for details). For these studies, we prepared and used cytoskeletal samples (Triton X-100-insoluble fraction) of cardiocytes based on our previous experiments that showed an increase in tyrosine phosphorylation of several cytoskeleton-bound proteins subsequent to c-Src’s cytoskeletal binding and activation during hypertrophic growth induction (33). To demonstrate the effect of an increase in ATP concentration, we performed electroporation after including 9 mM ATP (the normal ATP concentration in cardiocytes is reported to be in the range of 4 to 5 mM) to the culture medium. Several cytoskeletal proteins showed increased tyrosine phosphorylation under this condition (Fig. 6A). Importantly, this effect was not observed when electroporation was performed with 0.5 mM 5′-AMP either alone or in the presence of 9 mM ATP. Experiments using phospho-Tyr-416 antibody for c-Src (Fig. 6B) showed the presence of Tyr-416-phosphorylated c-Src subsequent to the increase in ATP concentration, and this effect was also significantly reduced when 5′-AMP was included. These changes were mostly observed in the Triton X-100-insoluble fraction compared to the soluble fraction (data not shown). All of these experiments demonstrate that increased levels of ATP can result in the tyrosine phosphorylation of cellular proteins accompanied by the activation (via Tyr-416 phosphorylation) of c-Src and that this effect can be blocked by 5′-AMP at a concentration approximately 1/20 of the concentration of ATP. Therefore, these studies clearly demonstrate that a change in the ratio of ATP to 5′-AMP in cardiocytes can alter c-Src activity and complements the changes observed with the in vitro experiments (Fig. 5A).
In vivo effect of changing nucleotide concentrations. (A) Isolated adult feline cardiocytes (2 × 105) were cultured for 24 h on laminin-coated 60-mm-diameter culture dishes. Part of the culture plates was used for electroporation in the absence or presence of ATP (9 mM), 5′-AMP (0.5 mM), or both as described in Materials and Methods. A Triton X-100-insoluble fraction was prepared, and the denatured protein samples were used for Western blot analysis using antiphosphotyrosine antibody. Arrows indicate the positions of proteins showing increased tyrosine phosphorylation following the ATP transfer; the solid arrow indicates the position of c-Src in the blot. Positions of molecular mass markers are shown in kilodaltons on the left. (B) Western blot analysis for the same samples was performed with anti-phospho-Tyr-416 c-Src antibody.
DISCUSSION
The progression of proliferating precursor cells to terminal differentiation via the expression of type-specific genes is governed by complex regulatory mechanisms that are not completely understood. For development of both cardiac and skeletal muscle, common mechanisms culminate in the expression of muscle-specific genes with distinct molecular differences for each tissue type, leading to terminally differentiated myocytes (reviewed in reference 42). The nonproliferating nature of the terminally differentiated cardiocytes prompted us to use them to identify novel factors that might control key kinases involved in mitogenic signaling. In this paper, we report the detection of an inhibitory activity that suppresses primarily Src family kinases. This activity is present in lysates prepared either from adult cardiocytes or from cardiac tissue and is constituted by a combinatorial action of two components, 5′-AMP and specific members of the sHSP family. Extensive analyses in vitro with the members of the various kinase families indicate that the nonreceptor tyrosine kinases such as the Src family are preferentially inhibited. Furthermore, only 5′-AMP and to a lesser extent 5′-ADP cooperate with a few members of sHSPs, such as HSP-27 or HSP-32, in causing the inhibition. Although Src family kinases are known to undergo inactivation via a phosphorylation-dependent mechanism, this novel inhibitory activity is independent of the tyrosine phosphorylation state of c-Src.
Several synthetic inhibitors have been reported for tyrosine kinases (reviewed in reference 52). These inhibitors, in particular those that inhibit the Src family members, include 4-amino - 5 - (4 - methylphenyl) - 7 - (t - butyl)pyrazolo[3,4 - d]pyrimidine (PP1) (25), quercetin (22), and 5′-adenylylimidodiphosphate (1). Both 5′-AMP, which we have identified as a component of the inhibitory activity, and 5′-ADP have structural similarities to these inhibitors. The inhibitory activities of sHSP and 5′-AMP and of PP1 (25) have several similarities: (i) they preferentially inhibit Src family kinases and exhibit low-level inhibition for EGF-R kinase; (ii) they show very low or no inhibition on serine/threonine kinases; (iii) among the other nonreceptor tyrosine kinases, both inhibitory mechanisms partially affect one or more kinase members (Fig. 2B and C); and (iv) 5′-AMP appears to interact directly with the ATP binding site as demonstrated recently for PP1 (49). In the case of 5′-ADP, previous studies (4) demonstrate that it can mediate a direct dephosphorylation of c-Src. However, our studies on c-Src autophosphorylation demonstrate that in the absence of HSPs, the higher ATP concentration, as found in vivo in cardiocytes (60, 64), may prevail over 5′-ADP, resulting in prevention of the dephosphorylation reaction. However, the presence of sHSPs facilitates both 5′-ADP- and 5′-AMP-mediated c-Src inhibition, apparently due to increased affinity of these nucleotides for c-Src.
The protein sequence obtained after purification shows significant homology to sHSP sequences. Studies with purified HSPs indicate that the 50% inhibitory concentration of HSP-27 is in the range of 100 to 200 nM, well below its cellular concentration (29). Both HSP-27 and αB-crystallin are expressed at high levels in cardiac and skeletal muscle (29). Importantly, the expression of sHSPs during development coincides with the onset of differentiation, suggesting that sHSPs may play a role in terminal differentiation (reviewed in reference 3). Therefore, based on the inhibitory concentrations of several HSPs in these in vitro studies, we anticipate that their inhibition on c-Src may prevail in vivo as well, especially in terminally differentiated cardiocytes. Furthermore, our data showing c-Src inhibition with HSP-25, the murine form of HSP-27, indicate that the inhibitory mechanism prevails in other species. Cardiac lysate prepared from murine hearts showed c-Src inhibition similar to that for the feline heart sample (data not shown). Our studies also indicate that HSP-32-mediated combinatorial inhibition of c-Src appears to be stronger than that of other sHSPs. The presence of high levels of HSP-32 in the cytoplasm of cardiac muscle and other tissues such as liver (39) suggests that this mechanism of c-Src inhibition may play a role in other cell types as well. Moreover, it has been well established that c-Src is involved in thrombin-induced platelet activation (14), and a recent study demonstrates that HSP-20 prevents this activation (41).
Our data indicate that the sHSP-mediated c-Src inhibition occurs within the physiological ratios of ATP to 5′-ADP and/or 5′-AMP. The concentrations of 5′-AMP and 5′-ADP in the heart appear to be around 10 and 800 μM, respectively (60, 64). Therefore, in the presence of HSP-27 and/or HSP-32, these two nucleotides can cause c-Src inhibition even though the ATP concentration in the heart is estimated as 5 mM. Furthermore, sHSPs do not appear to alter the affinity of ATP for c-Src, since in the absence of 5′-AMP or 5′-ADP, sHSPs do not affect c-Src kinase activity. Overall, these experiments suggest that in cells expressing sHSPs such as HSP-27, both 5′-AMP and 5′-ADP are likely to regulate the activity of Src family kinases. This view is further supported by our in vivo experiments wherein an increase in the cardiocyte ATP concentration resulted in increased tyrosine phosphorylation of several cytoskeleton-bound proteins including c-Src. Phospho-Tyr-416 antibody shows the activation of c-Src subsequent to the increase in ATP level. Importantly, this ATP-induced effect can be blocked by including 5′-AMP during electroporation. These experiments indicate that changing the nucleotide levels alters the cellular tyrosine kinase activity. Based on the observation that 5′-AMP does not inhibit c-Src independently in vitro and that the cardiocytes contain sHSPs such as HSP-27 and HSP-32, we propose that the sHSP/5′-AMP-mediated combinatorial inhibition prevails at least in the terminally differentiated cardiocytes.
Regulation of c-Src activity has been extensively studied since the discovery of its role in cellular transformation and proliferation. Earlier models for c-Src regulation, describing closed inactive and open active configurations (reviewed in reference17), were based on the findings that phosphorylation of either Tyr-416 or Tyr-527 activates or down regulates c-Src activity, respectively. Additional details are now known (reviewed in reference 58) based on the crystal structure studies (53, 65). In adult cardiocytes, c-Src and c-Fyn are the predominantly expressed Src family members (31, 33), whereas c-Yes is expressed only in neonatal cardiocytes. In most tissues and cells, c-Src has been shown to be phosphorylated at Tyr-527 by C-terminal Src kinase, which is present in almost all cells and tissues in amounts sufficient to phosphorylate and inactivate c-Src (reviewed in reference 5). However, the inhibition we report here is not mediated via Tyr-527 phosphorylation, since the phosphorylation of c-Src was completely blocked in all of our experiments upon addition of the inhibitory components in the kinase reaction. Furthermore, our recent studies in vivo demonstrate the presence of significant amount of unphosphorylated c-Src in the normal adult heart (33). Such observations indicate that an inactive form of c-Src is present in the heart via a Tyr-527 phosphorylation-independent mechanism. Taken together, our previous studies complement our present work, demonstrating a phosphorylation-independent mechanism in cardiocytes for suppressing Src family kinases. An absence of tyrosine-phosphorylated c-Src has been shown in Rous sarcoma virus-transformed chicken embryo fibroblasts, in which c-Src is complexed with HSP-90 and pp50 (7). However, in the present study, HSP-90 did not alter c-Src activity in the absence or presence of 5′-AMP. Our results also show that the inhibition is independent of the Tyr-416 phosphorylation status of c-Src or the associated activation process since the unphosphorylated and preautophosphorylated (active) forms of c-Src are inhibited to similar extents. Furthermore, a constitutively active form of c-Src, which lacks Tyr-527 site and is similar to v-Src, is also susceptible to this inhibitory effect. Therefore, this inhibitory mechanism appears to be different from the recently identified caveolin-mediated inhibition, which is selective for the inactive form of c-Src (38).
Our efforts to identify the c-Src domain required for the kinase inhibition reveal that the catalytic domain can directly interact with the inhibitory components. This observation is further supported by the fact that the nonreceptor tyrosine kinases that share homology in the conserved regions of the Src catalytic domain, such as v-Abl, show moderate susceptibility to the inhibitory activity. Therefore, both sHSP and 5′-AMP appear to interact directly with c-Src at distinct subdomains of the catalytic region of c-Src. Based on our studies, we propose that sHSPs facilitate a direct competition of 5′-AMP and/or 5′-ADP with ATP by interacting with the catalytic domain of Src, resulting in the loss of kinase activity. Since sHSPs have no nucleotide binding site, these studies also suggest a direct interaction of the c-Src catalytic domain both with the nucleotides (5′-AMP and 5′-ADP) and with sHSPs. Among the other functional domains of c-Src, the SH2 and SH3 domains do not appear to interact directly with the sHSP for the following reasons: (i) during the kinase assay, HSP-27 is not found to be tyrosine phosphorylated for the interaction with the SH2 domain of c-Src (data not shown); and (ii) a fusion protein corresponding to the proline-rich sequence of middle T antigen, which is known to interact with the SH3 domain of c-Src, does not interfere with the kinase inhibition (data not shown). However, it is possible that the cellular proteins that bind to the SH2 and SH3 domains influence the interaction between c-Src and the inhibitory complex and thus interfere with the inhibitory activity. For example, cytoskeleton-bound c-Src has been reported to be active in several cell types (24) including cardiocytes (33). Whereas the c-Src catalytic domain alone undergoes significant inhibition, our study also suggests the participation of the unique domain, since the 52-kDa fragment of c-Src, which is thought to lack a part of the unique domain (21), is less susceptible to the combinatorial inhibition, possibly due to a conformational change. This finding further clarifies why Fyn, which differs considerably in the unique domains from other Src family members, is less susceptible to the combinatorial inhibition relative to c-Src.
In terminally differentiated cells, such as cardiac muscle cells, two sets of earlier findings support the view that c-Src mediates a role opposite that of the protective effect of sHSPs and nucleotide analogs on the sarcomeric structure. First, expression of v-Src in both myoblasts and myotubes induced a vacuolation response caused by sarcomere disruption (19); this sarcomeric I-Z-I band disassembly by v-Src was blocked with 2-aminopurine (9), and under these conditions the MAPK-mediated mitogenic signaling was not affected. In addition, v-Src activation in myoblasts results in transformation, whereas down regulation leads to fusion of myoblasts to form myotubes (26). Second, overexpression of HSP-27 in neonatal and adult cardiocytes demonstrates protection of the heart against ischemic injury (40). Moreover, phosphorylated HSP-27 has been shown to mediate a cell-protective role via actin filament stabilization (35), and overexpression of HSP-32 in pulmonary epithelial cells results in both growth arrest and cytoprotection (36). Therefore, sHSPs and 2-aminopurine play a protective role in sarcomere structure, whereas activated Src plays a disruptive role. Since the inhibition that we report is preferential for c-Src rather than MAPK, and 2-aminopurine is structurally similar to 5′-AMP, these studies further support that the c-Src inhibitory mechanism, mediated by the combinatorial action of sHSPs and 5′-AMP/5′-ADP, is likely to prevail in vivo and promote both formation and maintenance of sarcomere structure in muscle. Although we propose a strong inhibitory mechanism, c-Src appears to be dynamically regulated in the heart; for example, during pressure overloading, activated c-Src was found to be present exclusively in the cytoskeletal fraction (33). Since the ATP concentration is not appreciably changed during this initial phase of hypertrophic growth initiation, the mechanism of c-Src activation could be due to either compartmentalization of c-Src or its association with the cytoskeletal proteins whereby the interaction of the inhibitory components with c-Src is prevented. Furthermore, presence of such strong inhibitory activity in the heart may allow c-Src to function mostly as a scaffolding molecule for actin bundling via a kinase-independent mechanism (28). Whereas this inhibition is preferentially observed for Src family kinases, a moderate level of inhibition was also observed for other nonreceptor tyrosine kinases such as v-Abl, suggesting that the sHSP/5′-AMP combination might play a broader role in the regulation of tyrosine phosphorylation of cellular proteins and control functional redundancy among tyrosine kinases.
The implications of the present findings may extend to other areas such as cancer research, where c-Src activation has been associated with neoplastic transformation (reviewed in reference45). HSP-90 was demonstrated to play a role in promoting oncogenic transformation (63) and positively correlated with cell proliferation (46). In contrast, certain sHSPs are known to decrease cell proliferation (30), and HSP-27 has been shown to be a marker for favorable prognosis in certain cancers, although a negative prognosis has been shown in other cancers (reviewed in reference 13). Furthermore, HSP-27 has been correlated with growth arrest in normal and neoplastic human B lymphocytes (56), and 5′-AMP-mediated antilymphoproliferative activity was demonstrated in a <1-kDa fraction prepared from brown adipose tissue (2).
In conclusion, we describe the identification of a novel inhibitory mechanism that preferentially inhibits Src family kinases by factors isolated from cardiocytes. This inhibitory mechanism involves sHSPs in conjunction with 5′-AMP or 5′-ADP and may serve as a potential mechanism to reduce Src kinase activity for maintenance of the terminally differentiated state.
ACKNOWLEDGMENTS
This study was supported by program project grant HL-48788 from the National Heart, Lung, and Blood Institute and by research funds from the American Heart Association (South Carolina Affiliate, Inc.).
We thank Bala Sundaravadivel for constructing baculoviral plasmid for the c-Src catalytic domain, Charlene Kerr for excellent technical assistance, George Cooper for overall support, Donald R. Menick and Kathryn E. Meier for careful reading of the manuscript and Kanefusa Kato for the polyclonal HSP-20 antibody.
FOOTNOTES
- Received 7 June 1999.
- Accepted 7 July 1999.
- Copyright © 1999 American Society for Microbiology
