INTRODUCTION

Insulin binding to the insulin receptor leads to the tyrosine phosphorylation of several insulin receptor substrates (IRS),¹ including IRS-1, IRS-2, IRS-3, Gab1, and Shc (1-3). These insulin receptor substrates then bind to a number of proteins in the cell via interaction of tyrosine phosphorylation motifs and specific domains of the target proteins termed SH2 (src homology 2) domains (4-6). For IRS-1 and IRS-2, the SH2 proteins include the lipid-modifying enzyme phosphatidylinositol (PI) 3-kinase, the cytoplasmic tyrosine kinase Fyn, the protein tyrosine phosphatase SHP2, and the adaptor molecules Nck and Grb2 (7-12). Exactly how IRS binding to SH2 domain proteins results in signal transduction is still not understood, but for PI 3-kinase and SHP2, binding of phosphorylated IRS-1 results in stimulation of enzymatic activity (13, 14). Grb2, on the other hand, has no intrinsic enzymatic activity, but links IRS proteins to the Ras pathway by activating the GTP exchange factor SOS thus stimulating Ras GTPase activity (15). Activated Ras then initiates a series of signaling events involving Raf, mitogen-activated protein kinase kinase, and mitogen-activated protein kinase which play a role in phosphorylation and dephosphorylation of some of the enzymes involved in insulin action, phosphorylation of several nuclear transcription factors, and some of the effects of insulin in cell growth and gene expression (16). Thus, the IRS proteins serve as an important point in the insulin action pathway at which the signal diverges to produce the multiplicity of final biological effects of the hormone.

Most studies of IRS proteins and SH2 domain proteins have been conducted in model cell systems. To identify novel IRS binding proteins in skeletal muscle, the predominant tissue involved in insulin-stimulated glucose disposal and the major site of insulin resistance in non-insulin-dependent diabetes mellitus, we screened a human skeletal muscle cDNA expression library with baculovirus-produced human-phosphorylated IRS-1. One of the new IRS-binding proteins identified by the method was the adult fast twitch skeletal muscle Ca²⁺-ATPase (SERCA1). This study reports the result of that screening and demonstrates that the interaction between different isoforms of the Ca²⁺-ATPase (SERCA1 and SERCA2) occurs with both IRS-1 and IRS-2 in intact cells and muscle, is dependent on tyrosine phosphorylation after stimulation with insulin, occurs through defined phosphorylation sites in IRS proteins, and may create a link between the tyrosine phosphorylation cascade and ion flux.

MATERIALS AND METHODS

Polymerase chain reaction was used to introduce a BamHI site at the 5-untranslated end of human IRS-1 (17) which was then subcloned into pBlueBac III (Invitrogen, San Diego, CA) in two fragments, using the BamHI and NcoI sites of the vector. The 5 fragment was from the BamHI site (directly before the initiation ATG at nucleotide 1021) to the SfiI site (nucleotide 1792), and the 3 fragment was from the SfiI site to the BspHI site (nucleotide 4840). The plasmid was introduced into baculovirus, and a baculovirus stock containing human IRS-1 was made according to manufacturer's protocol. Production of IRS-1 protein and purification over a Sephacryl 300 HR column was carried out as described for rat IRS-1 protein and verified by Western blot analysis using a C-terminal antibody to human IRS-1 (18).

A human skeletal muscle cDNA expression library was made in the EXlox system (Novagen, Madison, WI), and isopropylthiogalactopyranoside-induced proteins were bound to nitrocellulose filters (19). The IRS-1 protein was phosphorylated by wheat germ agglutinin-purified insulin receptor from Chinese hamster ovary cells overexpressing human insulin receptor in vitro using [-³²P]ATP (20). The phosphorylated IRS-1 was separated from free [-³²P]ATP by using a NAP-10 column (Pharmacia Biotech Inc.). Approximately 10⁵ plaques of the human skeletal muscle expression library were screened with the phosphorylated IRS-1 protein. The filters of the expressed proteins were washed in buffer 1 (10 mM Tris, pH 7.5, 150 mM NaCl, 0.05% Tween 20, and fresh 10 mM 2-mercaptoethanol) at room temperature and then placed in blocking buffer (10 mM Tris, pH 7.5, 250 mM NaCl, 5% dry milk, and fresh 10 mM 2-mercaptoethanol) for 4 to 5 h at 4 °C. This buffer was removed and replaced with blocking buffer containing the labeled IRS-1 protein probe, and the filters were incubated overnight at 4 °C. The filters were then washed in buffer 2 (10 mM Tris, pH 7.5, 150 mM NaCl, 0.01% Tween 20, and fresh 10 mM 2-mercaptoethanol) three times for 15 min each at room temperature, wrapped in plastic wrap, and subjected to autoradiography. Positive plaques were identified, isolated from the plates, and purified by additional screening. A plasmid (pEXlox) containing the insert was obtained by Cre-mediated excision according to Novagen's instructions and sequenced using a polymerase chain reaction-based automatic sequencing with fluorescent dyes and a model 373 automatic sequencer (Applied Biosystems, Foster City, CA).

C2C12 myoblasts were maintained in the Dulbecco's modified Eagle's medium (DMEM, 4.5 mg/ml glucose) supplemented with 10% fetal bovine serum in a 5% CO₂ environment. For differentiation of myocytes, confluent cells were fed with DMEM containing 1% calf serum and allowed to differentiate 6-7 days prior to harvesting. Primary smooth muscle cells were obtained from freshly prepared rat aortas and maintained in DMEM supplemented with 10% calf serum in a 10% CO₂ environment.

Male Sprague-Dawley rats (300-350 g) were maintained on standard rodent chow. Twelve to 14 h prior to the experiment, food was withdrawn. The rats were then anesthetized with sodium pentobarbital (100 mg/kg) injected intraperitoneally. Following loss of pedal and corneal reflexes, 5 units (0.2 mg) of regular insulin or diluent were injected into the portal vein. Quadriceps muscles and cardiac muscle were excised 5 min after injection and homogenized in ice-cold buffer (1% Triton X-100, 10% glycerol, 1% Nonidet P-40, 50 mM Hepes (pH 7.4), 100 mM sodium pyrophosphate, 100 mM sodium fluoride, 10 mM EDTA, 5 mM sodium vanadate, aprotinin (10 µg/ml), leupeptin (5 µg/ml), benzamidine (1.5 mg/ml), and 200 mM phenylmethylsulfonyl fluoride using a Polytron PTA 20S generator operated at maximum speed for 30 s (21). The insoluble material was removed by centrifugation at 55,000 rpm in a Ti 70 rotor for 60 min, and equal protein amounts of the supernatant were subjected to immunoprecipitation for 2 h using the indicated antibodies. Subsequently, protein A-Sepharose beads were added for another hour. The beads were washed three times, and proteins were solubilized in 2 × Laemmli sample buffer with boiling and separated on 7.5% polyacrylamide gels followed by Western immunoblotting using specific antisera and detection with ¹²⁵I-labeled protein A as described previously (22). SERCA1 and SERCA2 monoclonal antibodies were purchased from Affinity Bioreagents, Inc. Polyclonal antibody to IRS-2 was provided by Dr. M. F. White (Joslin Diabetes Center, Boston, MA).

The Ca²⁺-ATPase fragment (from nucleotide 894 to 3060) was generated by polymerase chain reaction using the plasmid from the library containing a clone that displayed the C-terminal end of SERCA1 as a template. The resulting cDNA was subcloned into BamHI-EcoRI, digested pGEX-2TK vector (Pharmacia Biotech Inc.). The recombinant plasmid was transformed into BL21 cells (Novagen), and expression of GST-Ca²⁺-ATPase fragment was induced with isopropyl -D-thiogalactone and purified with (GSH)-Sepharose beads (Pharmacia) (23). For the pull-down experiments, the tissue extracts were obtained as described above and incubated with GST-Ca²⁺-ATPase fragment or GST-Sepharose beads alone (20 µg) for 1 h at 4 °C followed by washing three times with the lysis buffer. The samples were processed for SDS-PAGE and Western blotting using the polyclonal C-terminal IRS-1 antibody and the polyclonal phosphotyrosine antibody.

IRS-1 was phosphorylated using wheat germ agglutinin-purified insulin receptor in vitro for 3 h at room temperature and then incubated for 1 h at 4 °C with the GST-Ca²⁺-ATPase fragment prepared as described above. The beads were washed three times, solubilized with Laemmli buffer, and the proteins resolved by 7.5% SDS-PAGE. The blots were probed with the polyclonal IRS-1 antibody.

The phosphopeptide library used contains peptides with the general sequence GAXXXpYXXXKKK (single letter amino acid designation), in which pY is phosphotyrosine and X indicates a position of degeneracy containing approximately equimolar amounts of all amino acids except cysteine and tryptophan. The synthesis and analysis of this library were performed as described previously (24). 150 µl of glutathione-agarose beads containing approximately 300 µg of the GST-SERCA1 fusion protein or GST protein alone were packed in a 1-ml syringe as an affinity column. The beads were washed with 2 ml of phosphate-buffered saline (PBS). Approximately 0.45 mg of the degenerate peptide mixture was loaded on the top of the column, and the column was allowed to sit at room temperature for 10 min without flow. The column was then quickly washed twice with 1 ml of ice-cold PBS (containing 10 mg of blue dextran per ml and 0.5% Nonidet P-40) and was washed once with 1 ml of cold PBS using a plunger to force the solution through. Peptides that remained bound were then eluted from the GST-agarose beads along with the GST-SERCA1 fusion protein using 200 µl of 30% acetic acid over 10 min at room temperature. The eluent was dried down overnight on a Savant Speed-Vac, resuspended in 40 µl of distilled water, and centrifuged at 10,000 × g for 2 min to remove the fusion protein that remains insoluble. The soluble phase, which contains the bound peptides, was then sequenced on an Applied Biosystems model 477A amino acid sequencer (25).

RESULTS

A human skeletal muscle cDNA expression library was created using the EXlox system and screened with purified, phosphorylated human IRS-1 protein as described under "Materials and Methods." Ten plaques producing IRS-1 binding proteins were identified in 100,000 clones, purified through two or three rounds of additional screening, and subjected to Cre-mediated excision (Fig. 1A). Sequence analysis revealed that one of these clones contained a 351-base pair cDNA with a high degree of homology to the rabbit fast twitch skeletal muscle Ca²⁺-ATPase (SERCA1). When the predicted amino acid sequence of the human muscle clone was compared with the known sequences of rabbit SERCA1 and SERCA2, there was an 88% homology at the amino acid level (Fig. 1B). In addition, the human clone contained an insert of 42 base pairs that encoded a proline-rich sequence near the C terminus of the molecule not present in the rabbit sequence. SERCAs consist of three globular cytoplasmic domains that are separated by a five-member stalk sector (S1-S5) from the transmembrane sector consisting of 10 helices (M1-M10) (Fig. 1C). The isolated clone contained the C-terminal end of SERCA1 including the cytoplasmic tail and a part of the transmembrane region 10 (M10).

To determine if IRS proteins interact with SERCA1 or SERCA2 in the context of the cellular environment, tissue lysates were prepared from rat muscle before and after insulin injection, immunoprecipitated with polyclonal antibodies to IRS-1 and IRS-2, and blotted with antibodies to SERCA1 and SERCA2. Immunoblotting of the precipitates from skeletal muscle with both anti-IRS-1 and anti-IRS-2 antibodies revealed co-precipitation of species of about 110 kDa representing SERCA1 (Fig. 2A). In both cases there was an insulin-dependent increase in association about 2-fold for IRS-1 and 6-fold for IRS-2.

To investigate the potential interaction of IRS-1 and IRS-2 with other SERCA isoforms, we tested extracts of rat cardiac muscle (which contains SERCA2a) before and after insulin stimulation and cultured primary aortic smooth muscle cells (which contains SERCA2b). Again, there was coprecipitation of IRS-1 and IRS-2 with the Ca²⁺-ATPase from rat cardiac muscle SERCA2 (Fig. 2B) which increased following insulin stimulation. With the smooth muscle cells in culture, insulin also stimulated association of IRS-2 with SERCA2 which was maximal after 5 min of stimulation (Fig. 3). The dose dependence of the effect was investigated in cultured C2C12 myotubes that express SERCA1 and primary smooth muscle cell cultures that express SERCA2b and indicated a maximum induction of association at 100 nM insulin (Fig. 3, A and B).

The most highly conserved region in SH2 domain proteins is the FLVRES sequence, which has been implicated in phosphotyrosine binding (25). Although the SERCA proteins do not contain a classic FLVRES sequence, they do contain a similar sequence in the C-terminal cytoplasmic domain (indicated by the boxed region in Fig. 1B). Therefore, a 10-amino acid peptide (LKFVARNYLE) was synthesized using the sequence of the region in the Ca²⁺-ATPase fragment that has the highest similarity with a typical SH2 domain and incubated at concentrations of 0 to 100 µM with lysates from insulin-stimulated rat skeletal muscle during immunoprecipitation with the polyclonal IRS-1 antibody. Western blots of these immunoprecipitates with a monoclonal SERCA1 antibody showed a concentration-dependent inhibition of the binding of IRS-1 to SERCA1 by this peptide, with almost complete inhibition at 1 µM (Fig. 4).

To explore the possible motif in IRS-1 or IRS-2 involved in binding to SERCA1 the Ca²⁺-ATPase C-terminal fragment expressed as a GST fusion protein was used to affinity purify a subgroup of phosphopeptides from a synthetic phosphopeptide library as described previously (24). The mixture was sequenced, and the relative abundance of the 18 amino acids present at each of the degenerate positions was determined and compared with the abundance in a control experiment. The results are presented in Fig. 5. The Ca²⁺-ATPase C-terminal fragment selectively bound phosphopeptides that have glycine at residue 4 (first residue after Tyr(P)), a serine at residue 5 (second residue after Tyr(P)), and a serine at residue 6 (third residue after Tyr(P)). Thus, a consensus sequence for SERCA binding is pYGSS. This is identical to potential tyrosine phosphorylation sites of human IRS-1 at position 431 (DEYGSSP) and human IRS-2 at position 500 (DEYGSSP).

To confirm the IRS-SERCA interaction, the Ca²⁺-ATPase C-terminal fragment corresponding to the human SERCA isoform identified by library screening was expressed as a GST fusion protein and tested for its ability to bind to and pull down phosphorylated IRS-1 in skeletal muscle lysates from control and insulin-stimulated rats. As shown in Fig. 6, the GST-Ca²⁺-ATPase fragment fusion protein was capable of pulling down IRS-1 in an insulin-dependent manner. The signal was similar to that obtained with pull-down experiments using a GST fusion protein of the N-terminal SH2 domain of the p85 subunit of PI3-kinase. GST alone as a control did not interact with IRS-1.

To determine if the SERCA-IRS interaction is direct and does not require other proteins in the complex, and to show that the interaction is dependent on tyrosine phosphorylation, we utilized a reconstitution assay with purified IRS-1 that had been phosphorylated using the insulin receptor in vitro. The GST-SERCA fusion protein was then incubated with phosphorylated and non-phosphorylated IRS-1 for 1 h at 4 °C, washed, and subjected to SDS-PAGE and Western immunoblotting with the IRS-1 antibody. Using equal amounts of IRS-1 protein, as determined by immunoblotting with an anti-insulin, stimulated both autophosphorylation of the insulin receptor and tyrosine phosphorylation of IRS-1 (Fig. 7A). Incubation of the non-phosphorylated and phosphorylated IRS-1 with GST alone showed no in vitro association (Fig. 7B, lanes 1 and 2). By contrast, the GST-Ca²⁺-ATPase fragment associated with IRS-1 in a phosphorylation-dependent manner (Fig. 7B, lanes 3 and 4). Thus, there is direct binding of IRS-1 and the Ca²⁺-ATPase fragment, and this is dependent on IRS-1 tyrosine phosphorylation.

Previous studies have indicated that the phosphorylation of IRS-1 and IRS-2 and their ability to bind to and activate SH2 domain proteins is altered in diabetic states. To determine if there is any alteration in the interaction of IRS-1 or IRS-2 with SERCA1 or SERCA2 in diabetes, we performed immunoprecipitations from control and insulin-stimulated skeletal and cardiac muscle of control and streptozotocin diabetic rats. As previously demonstrated, insulin stimulates rapid association between skeletal muscle SERCA1 and IRS-1 and -2 (Fig. 8, A and B). Likewise, in the control group, there was a large increase in the association of IRS-2 and SERCA2 after insulin stimulation (Fig. 8C). By comparison, there was no increase in the association of either IRS-1 or IRS-2 and SERCA1 in skeletal muscle or SERCA2 in cardiac muscle after insulin stimulation in streptozotocin diabetic rats. Indeed, in some cases, there was a tendency to a lower than basal association in the tissues of the insulin-treated diabetic rats, although this was not statistically significant.

DISCUSSION

To identify components that might be involved in the skeletal muscle-specific responses to insulin, we created and screened a human skeletal muscle cDNA expression library with baculovirus produced human IRS-1 that had been tyrosine-phosphorylated with the insulin receptor in vitro. Using this approach, we have previously identified two novel isoforms of PI 3-kinase that bind IRS-1, as well as other skeletal muscle-specific proteins, such as titan (26). In the present study, we have shown that a 351-base pair clone that represents the C terminus of the human adult fast twitch skeletal muscle Ca²⁺-ATPase (SERCA1) is also an IRS-binding protein. This interaction can involve both IRS-1 and IRS-2, as well as both SERCA1 (which is expressed primarily in skeletal muscle) and SERCA2 (which is expressed primarily in cardiac muscle), and can be detected in skeletal and cardiac muscle of the rat and cultured cell lines, such as C2C12 muscle cells and primary rat aortic smooth muscle cells. This interaction is insulin-stimulated in a time- and concentration-dependent manner. To better understand the binding of SERCA1 and SERCA2 with the IRS proteins, we attempted to recreate the interaction in vitro. The pull-down experiments with the Ca²⁺-ATPase fragment expressed as a GST fusion protein confirm the interaction in an insulin-dependent manner and, in particular, show a direct interaction depending on tyrosine phosphorylation.

The Ca²⁺-ATPase protein consists of a simple polypeptide chain with a molecular mass of 110 kDa. The cloning and sequencing of cDNA encoding the CA²⁺-ATPase of rabbit muscle sarcoplasmic reticulum have resulted in a proposed structural model for the enzyme in which two cytoplasmic regions are joined to 10 mediated transmembrane -helices (M1-M10) by a pentahelical stalk region 1-5 (27, 28). Each Ca²⁺-ATPase monomer binds two calcium ions to two high affinity sites on the cytoplasmic face of the sarcoplasmic reticulum membrane. Ca²⁺ translocation seems to be accomplished through conformational changes in the enzyme whereby the Ca²⁺ binding sites lose their high affinity and the calcium ions gain access to the terminal membrane surface (29, 30).

Cytoplasmic calcium acts as a ubiquitous messenger, controlling many different aspects of cellular physiology. In striated muscle, calcium is stored principally within the terminal cisternae of the sarcoplasmic reticulum, is released through the calcium release channel (ryanodine receptor) to initiate contraction, and then is taken back up into the sarcoplasmic reticulum via the Ca²⁺-ATPase pump thus allowing relaxation (31). The Ca²⁺-ATPase of sarcoplasmic reticulum regulates intracellular Ca²⁺ concentrations. The Ca²⁺-ATPase utilizes the energy derived from the hydrolysis of ATP to transport Ca²⁺ ions across the membrane against a gradient (29, 30, 32). At least three Ca²⁺-ATPase isoforms are expressed in tissue-specific and developmentally regulated patterns (33, 34). The fast twitch muscle isoform (SERCA1) is expressed at high levels in striated skeletal muscle. The cardiac/slow-twitch muscle isoform (SERCA2) is expressed at high levels in cardiac cells and, in an alternatively spliced form, in non-muscle cells. SERCA3, the non-muscle isoform, is widely expressed (31).

The binding of a specific subset of SH2 domain-containing proteins to IRS-1 and IRS-2 creates a unique signaling complex with the potential for a diverse range of downstream effects. Binding of phosphorylated IRS-1 or the respective IRS-1-derived phosphotyrosine-containing peptides to SH2 domain containing enzymes such as PI 3-kinase (4), Grb2 (12, 16), and SHP2 in vitro results in an increase in their specific activity (4). The binding of a specific subset of SH2 domain-containing proteins to IRS-1 defines a unique signaling molecule with the potential for a diverse range of downstream effects. SH2 domains are short (approximately 100 amino acids), non-catalytic regions of proteins that facilitate recognition by binding to phosphotyrosine containing proteins (35). Sequence alignments of SH2 domain proteins have demonstrated a relatively highly conserved region corresponding to the FLVRES sequence in src, which is implicated in phosphotyrosine binding (36). Although the SERCA1 and -2 do not contain a classical SH2 domain, we identified a conserved region near the C-terminal end of SERCA proteins that has similarity to the FLVRES sequence. Using a 10 amino acid peptide which included this sequence, we were able to compete the binding of IRS-1 to SERCA-1.

To identify the potential phosphorylation sites on IRS-1 and IRS-2 which specially bind to the Ca²⁺-ATPase, we utilized a peptide library screening technique (25). This revealed a consensus sequence of pYGSS. This is identical to the motifs in IRS-1 at position 431 and in IRS-2 at position 500 (17). In preliminary experiments, a GST-fusion protein containing residues 280 to 511 of IRS-1 bound to SERCA1 in vitro confirming this as the likely site of interaction.

Several changes have been demonstrated in cardiac and skeletal muscle in diabetes that reflect possible alterations in the contractile proteins and the membrane systems controlling intracellular Ca²⁺ kinetics. On one hand, it has been shown that the sarcolemmal Ca²⁺ transport activity is increased in skeletal muscle of diabetic rats (37). Treatment of the diabetic animals with insulin reverses the changes in the Ca²⁺ transport activity toward the control levels. A similar effect was found in the heart. Thus, the sarcolemmal Ca²⁺-ATPase activity may be of critical importance in the hypersensitivity of diabetic hearts and cardiomyopathy (38, 39). On the other hand, there is evidence that Ca²⁺-ATPase activity is impaired in diabetic animals (40). This could result in an increase in the cytoplasmic Ca²⁺ concentration and play a role in reduced muscle relaxation and development of hypertension (40).

In summary, expression library screening with IRS-1 protein has revealed novel interactions between IRS-1, IRS-2, and the sarcoplasmic Ca²⁺-ATPase. This interaction can be demonstrated in skeletal and cardiac muscle in vitro and in vivo, is insulin-regulated, dependent on tyrosine phosphorylation, involves specific domains of IRS-1, IRS-2, and SERCA1, and is altered in diabetes. This interaction creates a potential link between the tyrosine phosphorylation cascade and effects of insulin on calcium and thus may be important in understanding insulin signal transduction in one of the predominant insulin responsive tissues.