In vivo phosphorylation of insulin receptor substrate 1 at serine 789 by a novel serine kinase in insulin-resistant rodents.

Insulin resistance is a key pathophysiologic feature of obesity and type 2 diabetes and is associated with other human diseases, including atherosclerosis, hypertension, hyperlipidemia, and polycystic ovarian disease. Yet, the specific cellular defects that cause insulin resistance are not precisely known. Insulin receptor substrate (IRS) proteins are important signaling molecules that mediate insulin action in insulin-sensitive cells. Recently, serine phosphorylation of IRS proteins has been implicated in attenuating insulin signaling and is thought to be a potential mechanism for insulin resistance. However, in vivo increased serine phosphorylation of IRS proteins in insulin-resistant animal models has not been reported before. In the present study, we have confirmed previous findings in both JCR:LA-cp and Zucker fatty rats, two genetically unrelated insulin-resistant rodent models, that an enhanced serine kinase activity in liver is associated with insulin resistance. The enhanced serine kinase specifically phosphorylates the conserved Ser(789) residue in IRS-1, which is in a sequence motif separate from the ones for MAPK, c-Jun N-terminal kinase, glycogen-synthase kinase 3 (GSK-3), Akt, phosphatidylinositol 3'-kinase, or casein kinase. It is similar to the phosphorylation motif for AMP-activated protein kinase, but the serine kinase in the insulin-resistant animals was shown not to be an AMP-activated protein kinase, suggesting a potential novel serine kinase. Using a specific antibody against Ser(P)(789) peptide of IRS-1, we then demonstrated for the first time a striking increase of Ser(789)-phosphorylated IRS-1 in livers of insulin-resistant rodent models, indicating enhanced serine kinase activity in vivo. Taken together, these data strongly suggest that unknown serine kinase activity and Ser(789) phosphorylation of IRS-1 may play an important role in attenuating insulin signaling in insulin-resistant animal models.

Insulin resistance, commonly defined as a decreased ability of insulin to stimulate glucose uptake/metabolism in peripheral tissues and to inhibit hepatic glucose output, is a major pathogenic problem in many human diseases, including obesity, type 2 diabetes, atherosclerosis, hypertension, hyperlipidemia, and polycystic ovarian disease (1)(2)(3). Although knowledge of the molecular mechanism of insulin action has been greatly enhanced, the molecular basis for insulin resistance remains unknown. Insulin receptor substrate (IRS) 1 proteins are key molecules of the insulin signaling cascade (4,5). They are tyrosyl-phosphorylated upon insulin stimulation, thereby triggering intracellular signaling through recruitment of proteins with the Src homology-2 domain, including PI3K, Grb-2, Nck, fyn, and Shp-2 among others (4, 6 -11). Studies of mice with a targeted disruption of IRS-1 or IRS-2 revealed insulin resistance (12)(13)(14). Consistent with this, a defect in tyrosyl phosphorylation of IRS proteins is associated with insulin resistance in human type 2 diabetes as well as in insulin-resistant animals and cultured cells (15)(16)(17)(18)(19)(20)(21), suggesting that the molecular basis for insulin resistance may reside at the level of IRS proteins. However, the specific molecular mechanism is not known.
A hypothesis has emerged recently that serine/threonine phosphorylation of IRS proteins (via enhanced serine/threonine kinase activity) decreases the ability of IRS proteins to be phosphorylated on tyrosine, thereby attenuating insulin signaling (20 -29). Several serine kinases have been reported to phosphorylate IRS-1 in vitro and/or in vivo, including MAPK (at Ser 612 ), glycogen-synthase kinase 3 (GSK-3), casein kinase (at Thr 502 /Ser 99 ), PI3K, mTor (at Ser 636,639 ), c-Jun N-terminal kinase (at Ser 307 ), and Akt (30 -39) and have been implicated in various manipulations that impair insulin signaling in cultured cells. However, whether any of these kinases and serine phosphorylation of IRS-1 are associated with insulin resistance in animal models has not been established.
We previously studied liver extracts from insulin-resistant rodents and showed enhanced serine kinase activity for IRS-1 in vitro (40). In this report, we have identified the specific serine phosphorylation site (Ser 789 ) on IRS-1 for the enhanced serine kinase activity. We then used a specific antibody that recognizes Ser(P) 789 to confirm the increased Ser 789 phosphorylation of IRS-1 in two genetically unrelated insulin-resistant rat models: Zucker fatty rats and JCR:LA-cp obese rats. The identity of the enhanced serine kinase is still not known; however, our study excluded all serine kinases known to phosphorylate IRS-1, indicating that a novel serine kinase exists that may underlie the molecular mechanism of insulin resistance.

EXPERIMENTAL PROCEDURES
Animals-JCR:LA-cp rats in this study were 6-month-old males, bred as previously described (kindly provided by Dr. Russell at the University of Alberta) (40). Three obese (cp/cp) and three lean male animals (ϩ/cp or ϩ/ϩ) were used in the study. Eight obese (fa/fa) Zucker and eight lean male control (Charles River) rats were used at 10 weeks of age (41). All rats were fasted for 4 h prior to experiment. All care and treatment of the animals were in accordance with the guidelines of the National Institute of Health and subjected to prior approval by the Institutional Animal Care and Use Committee of the University of Vermont.
Preparation of Liver Extracts-Livers were rapidly minced and homogenized in lysis buffer (10 mM Tris-HCl, pH 7.4, 250 mM sucrose, 100 mM NaF, 5 mM EDTA, 5 mM EGTA, 1 mM DTT, 0.5 mM PMSF, 0.2 mM Na 3 VO 4 , 10 g/ml leupeptin, and 10 g/ml aprotinin) (1 g of tissue/10 ml) with a Brinkmann Polytron homogenizer, followed by centrifugation at 10,000 ϫ g for 10 min (Sorvall RC-5B). The supernatants were centrifuged at 100,000 ϫ g for 30 min in a Beckman L8-M ultracentrifuge. The supernatants were precipitated with (NH 4 ) 2 SO 4 at 50% saturation, followed by centrifugation at 100,000 ϫ g for 30 min. (NH 4 ) 2 SO 4 precipitates were re-dissolved in lysis buffer followed by centrifuging at top speed in a Biofuge (Heraeus) for 15 min. The recovered supernatants were used as the source of kinase for the in vitro kinase assay (40).
Phosphorylation of IRS-1 by Liver Extracts or Recombinant Erk2-GST-IRS-1 fusion proteins or the ␣IRS-1 immune complexes were phosphorylated by 10 g of liver extract (50% (NH 4 ) 2 SO 4 precipitates) in vitro in a final volume of 40 l of kinase buffer (20 mM HEPES, pH 7.4, 1 mM DTT, 10 mM MgCl 2 , 100 g/ml BSA, 0.5 g/ml okadaic acid (Sigma Chemical Co.), 50 M cold ATP, with or without 5 Ci of [␥-32 P]ATP) at 30°C for 20 min (40). Phosphorylation of GST-IRS-1 fusion proteins or anti-IRS1 immune complex by recombinant Erk-2 (kindly provided by Drs. James Posada and Paul Vicki at University of Vermont) were carried out in a linked in vitro recombinant MAPK assay as described previously (40). Reactions were stopped by adding 10 l of 5ϫ Laemmli buffer containing 0.5 M DTT and boiled for 5 min. Proteins were separated on 10% SDS-PAGE for GST fusion proteins or 7.5% SDS-PAGE for full-length IRS-1, respectively, stained, and destained. 32 P-Labeled phosphorylated proteins were visualized by autoradiography. In some cases, protein bands were excised and counted in a scintillation counter (MINAXI Tri-CarB 4000). Unlabeled phosphorylated proteins were visualized by immunoblotting analysis.
AMPK Activity in Rat Liver Extracts-The liver extract (100 g of proteins diluted into 500 l of lysis buffer containing 1% Nonidet P-40) was incubated with either anti-␣1 subunit (␣1), anti-␣2 subunit of AMPK (␣2), or non-immune serum (NI) for 2 h followed by incubating with protein A-agarose (Invitrogen) for an additional 1 h. The immune complexes were washed with the same buffer three times and with kinase buffer twice before the kinase activity was measured. AMPK activity was measured by the in vitro kinase assay using GST-IRS1-(765-816) or SAMS peptide (HMRSAMSGLHLVKRR) (43) as a substrate in the presence or absence of 200 M AMP. Phosphorylated GST-IRS1-(765-816) was analyzed by 12% SDS-PAGE. Phosphorylated SAMS peptide was sported on Whatman P81 filter paper. The papers were extensively washed in 1% phosphoric acid, dried, and counted on a scintillation counter (MINAXI Tri-CarB 4000).
Reverse-phase HPLC Separation of Phosphopeptides-Phosphorylated proteins were extracted from SDS-PAGE gels as described previously (40,44), dissolved in 100 l of 50 mM ammonium bicarbonate (pH 7.6) containing 0.3 mg/ml TPCK-treated trypsin (Worthington Biochemical Corp.) or in 50 l of 0.1 M Tris-HCl, pH 7.8, and 0.01 M CaCl 2 containing 100 g/ml chymotrypsin (Sigma), and incubated at 37°C overnight. Digested phosphopeptides were dried in a Speedvac and redissolved in 0.055% trifluoroacetic acid (TFA). The tryptic or chymotryptic phosphopeptides were separated in a Rainin Dynamax HPLC system equipped with a Hi-Pore reverse-phase RP318 column (Bio-Rad) and eluted at a flow rate of 0.5 ml/min with 0.055% TFA modified with 75% acetonitrile-0.05% TFA. 32 P-Labeled phosphopeptides were monitored with a ␤-RAM on-line radioactivity detector (INSUS System) or by measuring Cerenkov radiation in 0.5-ml fractions in a scintillation counter (MINAXI Tri-CarB 4000).
Phosphopeptide Mapping by Tricine SDS-PAGE-Tryptic phosphopeptides derived from in vitro phosphorylation of GST fusion proteins or full-length IRS-1 were dissolved in Tricine sample buffer (Bio-Rad), resolved by 16.5% Tricine SDS-PAGE, and detected by autoradiography as described previously (40).
Manual Radioactive Amino Acid Sequencing-Tryptic or chymotryptic phosphopeptides isolated from HPLC were coupled to an acrylamine-Sequelon disc as described by the manufacturer (MilliGen/ Biosearch). Edman degradation of immobilized peptides was carried out in cycles consisting of the following steps in each cycle: (a) coupling with phenylisothiocyanate (Pierce) at 50°C for 10 min; (b) washing the disc five times with methanol; (c) removing the N-terminal amino acid derivative from the disc with TFA at 50°C for 6 min; (d) measuring the released radioactivity in TFA; and (e) washing the disc six times with methanol before the next cycle (10,45).
PCR-mediated Oligonucleotide-directed Mutagenesis-Serine 789 to glycine mutation in GST-IRS-1-(526 -859) (GST-IRS1-(526 -859)/ S 789 G) was generated by PCR-mediated oligonucleotide-directed mutagenesis (46,47) using pGEX-2T/GST-IRS-1-(526 -859) as a template and 5Ј-CGTCTCTCTTCAGGCTCTGGACTCC-3Ј and 5Ј-GGAGTCCA-GAGCCTGAAGAGAGACG-3Ј as mutagenic primers. The pGEX-2T/G-ST-IRS-1-(526 -859) and mutant PCR product were digested with BamHI and EcoRI, and the mutant PCR fragment was inserted in place of the wild type sequence. Serine 789 to alanine mutation in full-length IRS-1 (IRS1 S789A ) was generated by PCR-mediated oligonucleotide-directed mutagenesis with pBluescript/IRS-1 as a template and 5Ј-CG-TCTCTCTTCAGCCTCTGGACTCC-3Ј and 5Ј-GGAGTCCAGAGGCTG-AAGAGAGACG-3Ј as mutagenic primers. The mutant PCR product was digested with BstEII and BamHI and was inserted in place of the wild type sequence. The presence of the desired mutations was confirmed by sequencing the recombinant molecules on a 373 XL DNA sequencer in the Vermont Cancer Center DNA Analysis Facility at the University of Vermont.
Expression of Wild Type and Mutant IRS-1 in CHO/IR Cells-The wild type IRS-1 and IRS-1 S789A were subcloned into pCMV/his expression vector at SacI and HindIII sites. The pCMV/his expression vector confers histidinol resistance to CHO cells (42). CHO/IR cells were transfected with pCMV/his/IRS-1 or pCMV/his/IRS-1 S789A (10 g) by calcium phosphate-mediated transfection (42). The transfected cells were selected by resistance to 10 mM histidinol (Sigma) and used as sources for preparation of full-length wild type IRS-1 or IRS-1 S789A mutants.
Preparation of Phosphoserine-specific Antibody-Specific antibody against Ser(P) 789 of IRS-1 (␣IRS1 PS789 ) was prepared in rabbits immunized with the synthetic phosphopeptide (CLRLSS(pS)SGRLRY-amide) coupled to keyhole limpet hemocyanin (Quality Controlled Biochemicals, Inc.). Antiserum was tested by enzyme-linked immunosorbent assay showing a 25,000 higher titer for the phosphopeptide than for the non-phosphopeptide. The specificity of the antiserum was also characterized by immunoprecipitation and immunoblotting with phosphorylated and unphosphorylated IRS-1 proteins (see under "Results").
In Situ Immunofluorescent Detection of Serine 789 Phosphorylation of IRS-1 in Rat Livers-Liver tissues were briefly washed in ice-cold PBS and then fixed in 4.0% paraformaldehyde/PBS for 12 h at 4°C. After extensively washing in PBS, tissues were equilibrated in 30% sucrose/PBS overnight. Sucrose-infiltrated liver was embedded in OCT cryoembedding medium (Miles Scientific) and sectioned at 5 m in a cryostat. Frozen sections were mounted onto charged microslides (Charge-Plus, Fisher Scientific), hydrated in PBS, and permeabilized 20 min in PBS with 0.1% Triton X-100. Sections were blocked in PBS supplemented with 5% normal donkey serum and 1% bovine serum albumin (BSA) for 1 h at room temperature and incubated for 12 h at 4°C with ␣IRS1 (1:1000 dilution), ␣IRS1 PS789 (1:1000), or their corresponding pre-immune serum diluted in PBS containing 0.1% Triton and 1% BSA. Following three washes, sections were incubated with the secondary antibody (ML-grade donkey anti-rabbit IgG-CY3, Jackson ImmunoResearch, 1:2000). Nuclei were counterstained for reference with Yo-Pro1 (Molecular Probes), which was added to the diluted secondary antibody solution at 0.5 g/ml. After washing, liver sections were mounted in Aqua-PolyMount (Polysciences).
Semiquantitative Assessment of IRS-1 and IRS-1 PS789 in Hepatocytes in Situ-A comparative semiquantitative assessment of both total IRS-1 and IRS-1 PS789 immunoreactivity was accomplished by batch staining and sampling by confocal microscopy. Samples were imaged with a laser-scanning confocal microscope (Bio-Rad MRC 1024) (University of Vermont College of Medicine Cell Imaging Facility) using a 60ϫ PlanApo objective lens (numerical aperture ϭ 1.4) and the 568-nm excitation line of an argon/krypton laser. For each field the microscope was focused to maximize the number of cells optically sectioned through the middle of the nucleus. All confocal imaging parameters were identical for each imaged field and a minimum of four non-overlapping fields of ϳ190 ϫ 190 m each were captured for each specimen thereby yielding ample numbers of cells for statistical analyses. Grayscale images (512 ϫ 512 pixels) were transferred to a Power Macintosh G3 computer running IMAGE (version 1.62, National Institutes of Health) for image analysis.
Individual hepatocytes suitable for quantitation were identified and numbered. For each cell studied, an 18-pixel diameter circle (256 pixels total), the width of which is smaller than the cytoplasmic area between the surface and nucleus at this magnification, was used to measure mean pixel intensities (range ϭ 0 -255 grayscale levels) of the corresponding immunofluorescence signal. For uniformity in sampling, only a single area per cell was recorded, and only cells with their maximum widths in the confocal section were scored. Between 34 and 70 cells for each animal and antibody staining were analyzed and imaged under identical conditions. Field background fluorescence values (where there was no tissue) were subtracted from intensity values for each area analyzed.
Mean intensity values for an antibody staining were then corrected for nonspecific background staining by subtracting the mean fluorescence intensities of corresponding pre-immune serum. Thus, IRS-1 and IRS-1 PS789 immunoreactivities were expressed as corrected mean pixel intensities for each animal.

Enhanced Serine Kinase Activity in Liver Extracts of Zucker
Fatty Rats-JCR:LA-cp rats and Zucker fatty rats are two genetically unrelated, widely used animal models of insulin resistance and obesity (41, 49 -52). We had previously identified enhanced serine kinase activity in liver extracts from JCR: LA-cp obese rats (40). We now report the same finding in Zucker fatty (fa/fa) rats. Kinase activity in liver extracts from Zucker rats was compared with JCR:LA-cp rats using our in vitro kinase assay based on 50% (NH 4 ) 2 SO 4 precipitates of liver extracts as the source of kinase and a GST fusion protein containing the 526-to 859-amino acid region of IRS-1 (GST-IRS-1-(526 -859)) as substrate (40). Serine kinase activity was 3-and 2.5-fold higher in the Zucker fatty rats and JCR:LA-cp obese rats, respectively, compared with their lean controls (Fig. 1A).
To see if the same site within IRS-1 was serine-phosphorylated in Zucker and JCR:LA-cp rats, phosphorylated GST-IRS1-(526 -859) was digested with trypsin and the tryptic phosphopeptides were analyzed by Tricine SDS-PAGE and HPLC analyses. An identical tryptic phosphopeptide (P3) was identified by Tricine SDS-PAGE analysis in samples phosphorylated by liver extracts from both JCR:LA-cp and Zucker rats, with greatly increased density in the obese rats over their lean controls (Fig. 1B, lanes a-d). When GST-IRS1-(526 -859) was phosphorylated by Erk-2, P1 and P2 were the predominant tryptic phosphopeptides, and P3 was not detected (Fig. 1B, lane e). HPLC analysis showed a single tryptic phosphopeptide (eluted at 12.8 min) phosphorylated by the JCR:LA-cp rats (Fig. 1C, top panel) that was ϳ2-fold increased in the obese rats over the lean controls. This phosphopeptide was P3, because it co-migrated with P3 on Tricine SDS-PAGE (data not shown). An identical HPLC elution profile was obtained from tryptic peptide phosphorylated by extracts from the Zucker rats (Fig.  1C, middle panel). In contrast, the phosphorylation pattern of GST-IRS1-(526 -859) generated by recombinant Erk-2 was totally different with multiple tryptic phosphopeptides, none of which eluted at 12.8 min (Fig. 1C, bottom panel). These data indicate that the enhanced serine kinase activity in both insulin-resistant models phosphorylates IRS-1 at seemingly identical serine residues, which are distinctly different from the ones phosphorylated by Erk-2 kinase, confirming that the responsible kinase in these insulin-resistant animals is not an MAPK (40).
To see if other regions of IRS-1 are serine-phosphorylated by the enhanced serine kinase activity, we performed a similar experiment on full-length recombinant IRS-1 protein that was prepared from CHO/IR/IRS-1. As expected, increased phosphorylation of IRS-1 was detected by in vitro kinase assay using extracts from both the obese JCR:LA-cp and Zucker rats ( Fig.  2A, lanes b and d versus lanes a and c). Tryptic phosphopeptide mapping analyzed by Tricine SDS-PAGE showed that P3 was the major tryptic phosphopeptide that increased in both the obese JCR:LA-cp and the Zucker fatty rats (Fig. 2B, lanes b and  d versus lanes a and c), suggesting the phosphorylation site in the full-length IRS-1 and GST-IRS1-(526 -859) was identical. A few minor additional phosphopeptides were found when using extracts from the Zucker rats; however, there was no significant difference between the lean and fatty rats (Fig. 2B,  lanes c and d). Recombinant Erk-2 phosphorylated the fulllength IRS-1 mainly on P1 and P2 tryptic peptides with an additional few weak phosphopeptides; however, P3 tryptic peptide was not detected (Fig. 2B, lane e). These data confirm that P3 is the only tryptic phosphopeptide in IRS-1 induced by the enhanced serine kinase activity of the insulin-resistant rats.
Determination of the Specific Serine Phosphorylation Site for Enhanced Serine Kinase Activity-To determine the specific serine phosphorylation site(s), the tryptic phosphopeptide P3 was purified by HPLC (Fig. 3A, left panel) followed by manual radiosequencing analysis based on Edman degradation (10,45). The radioactivity was released at the fourth cycle of Edman degradation, indicating the phosphorylated serine was at the fourth position (Fig. 3A, right panel). Trypsin cleaves peptide bonds at the C-terminal of lysine or arginine except between proline and lysine or arginine (53). Serine 789 was the only candidate in GST-IRS1-(526 -859) to be a fourth residue from a tryptic cleavage site (Fig. 3A, right panel). Having noticed that the Ϫ3 position of Ser 789 is a leucine residue that can be cleaved at the C terminus by chymotrypsin (53), phosphorylated GST-IRS1-(526 -859) was digested with chymotrypsin followed by HPLC separation. A single phosphopeptide was isolated (Fig. 3B, left panel), and, as expected, radioactivity was detected at the third position by manual radiosequencing analysis (Fig. 3B, right panel). Tryptic and chymotryptic phosphopeptides phosphorylated by extracts from JCR:LA-cp rats were analyzed in an identical fashion, and similar results were obtained (data not shown).
To further confirm that the phosphorylation site was Ser 789 , two additional GST fusion proteins were generated: one containing amino acids 765-816 of IRS-1 (GST-IRS1-(765-816)) and another containing IRS-1-(526 -859) with Ser 789 mutated to glycine (GST-IRS1-(526 -859)/S 789 G) (Fig. 4A). These GST fusion proteins were exposed to 50% (NH 4 ) 2 SO 4 precipitates of liver extracts or recombinant Erk-2 in the in vitro kinase assay. As expected, GST-IRS1-(765-816) was as good a substrate as GST-IRS1-(526 -859) for the liver extracts (Fig. 4B, panel III  versus panel I, lanes a-d) despite lacking all of the potential MAPK phosphorylation sites (Fig. 4A). Also, Erk-2 failed to phosphorylate this substrate (Fig. 4B, panel III, lane e). Furthermore, there was no detectable phosphorylation of GST-IRS1-(526 -859)/S 789 G by the liver extracts (Fig. 4B, panel II  versus panel I, lanes a-d), although the mutation at Ser 789 had no effect on phosphorylation by Erk-2 (Fig. 4B, panels I and II, lane e). We also examined tryptic phosphopeptides of these GST fusion proteins on Tricine SDS-PAGE. Converting Ser 789 to glycine completely eliminated P3 phosphorylation by the liver extracts (Fig. 4C, lanes a-d); however, it had no effect on phosphorylation of P1 and P2 by recombinant Erk-2 (Fig. 4C, lane e). Deleting both MAPK sites in GST-IRS1-(765-816) had no effect on the phosphorylation of P3 by the liver extracts (Fig.  4C, lanes f-i); in contrast, it completely abolished the phosphorylation of P1 and P2 phosphopeptides by recombinant Erk-2 (Fig. 4C, lane j). Thus, the Ser 789 residue in IRS-1 is the phosphorylation site for the enhanced serine kinase activity identified in insulin-resistant animals, and the serine kinase is not an MAPK.
Enhanced Serine Kinase Activity Is Not Contributed by an AMP-activated Protein Kinase-The serine 789 phosphorylation site in IRS-1 is highly conserved among human, rat, and mouse (Fig. 5). It is surrounded by leucines at the P Ϫ 5 and P ϩ 4 positions, which resemble a phosphorylation motif for AMP-activated protein kinase (43,54). AMP-activated protein kinase (AMPK) is a heterotrimer of three subunits, i.e. ␣, ␤, and ␥. The 63-kDa ␣-subunit contains the kinase domain, and two isoforms of the ␣-subunits (␣1 and ␣2) have been described (55). AMP binds to the ␣-subunit of AMPK and increases its kinase activity by 5-to 8-fold (56). Recently, AMPK has been reported to phosphorylate Ser 789 of IRS-1 (39). To see if the enhanced serine kinase activity detected in the insulin-resistant rats is derived from AMPK, we performed the in vitro kinase assay in the presence or absence of AMP using GST-IRS1-(765-816) as a substrate. As previously noted, kinase activity in liver extracts was higher in obese JCR:LA-cp rats than in lean control rats in the absence of AMP. In the presence of AMP, the kinase activity did not increase in either the lean or the obese tissue extract (Fig. 6A). Failure of AMP to activate serine kinase activity was also observed in the extracts prepared from Zucker rats as well as from insulin-treated CHO cells (data not shown).
To see if depletion of AMPK by specific antibodies will alter the enhanced serine kinase activity in liver extracts, AMPK was immunoprecipitated from liver extracts by anti-␣1, anti-␣2 antibodies or non-immune serum, and kinase activities in both supernatant and the immune complex were measured. In the supernatants after immunodepletion of AMPK, there was slightly decreased kinase activity in lean animals (Fig. 6B,  upper panel, lane a versus lanes b and c); however, the en-FIG. 3. Determination of the phosphorylation site for the enhanced serine kinase activity. GST-IRS1-(526 -859) was phosphorylated by the 50% (NH 4 ) 2 SO 4 precipitates of liver extracts from Zucker fatty rats and separated by 10% SDS-PAGE. Phosphorylated GST-IRS1-(526 -859) was extracted from gels and digested with trypsin (A) or chymotrypsin (B). Tryptic and chymotryptic peptides were separated by HPLC, and radioactivity was monitored by an on-line radioactive detector. Phosphorylated tryptic peptide (A) and chymotryptic peptide (B) were collected from the HPLC and subjected to manual radiosequencing analysis. Radioactivity loaded on the disc was considered as 100%. The open bars represent the percentage of radioactivity left on the disc, and filled bars represent the percentage of radioactivity released from the disc. The candidate phosphorylation sites based on manual radiosequencing analysis are listed.
hanced serine kinase activity in obese animal was not affected (Fig. 6B, upper panel, lanes e and f versus lane d). Consistent with Fig. 6A, AMP failed to activate kinase activities in the supernatants (Fig. 6B, upper panel, lanes g-l versus a-f). In contrast to the supernatant, AMP-activated kinase activities were readily detected in both anti-␣1 and anti-␣2 immune complexes whether using GST-IRS1-(765-816) as a substrate (Fig. 6B, lower panel) or SAM (specific AMPK substrate) (Fig.  6C). However, there was no difference between lean and obese animals (Fig. 6, B, lower panel, and C). The fact that AMPK activities can only be detected in immune complexes but not in tissue extracts (Fig. 6A) indicates that AMPK activity was suppressed in liver extracts. Together, these data suggest that the enhanced serine kinase activity in insulin-resistant animals is not AMP-activated kinase, indicating a novel serine kinase.
In Vivo Detection of Serine 789 Phosphorylation of IRS-1 in Liver Tissues-The fact that enhanced serine kinase phosphorylates IRS-1 in vitro does not necessarily mean it occurs in vivo. To confirm that enhanced serine kinase phosphorylates IRS-1 in vivo, we generated an antibody against a synthetic peptide of IRS-1 that contained Ser(P) 789 (␣IRS1 PS789 ). The specificity of the antibody was tested by immunoblotting against GST-IRS1-(765-816), which had been exposed to the liver extracts in the presence (phosphorylated) or absence of ATP (unphosphorylated) in the in vitro kinase assay. Although there was 0.5 g of GST-IRS1-(765-816) protein in each lane, GST-IRS1-(765-816) was detected by ␣IRS1 PS789 only if ATP was present in the in vitro kinase assay (Fig. 7A, lanes c, d, g,  and h versus lanes a, b, e, and f), indicating the specificity of the antibody against phosphorylated GST-IRS1-(765-816). Because IRS-1 has been shown to be heavily phosphorylated at multiple serine sites in vivo in the basal state (42), we tested whether ␣IRS1 PS789 recognizes serine phosphorylation sites other than Ser 789 . This was done by immunoblotting full-length IRS-1 or Ser 789 to alanine mutant (IRS-1 S789A ) isolated from CHO/IR cells overexpressing wild type (CHO/IR/IRS-1) or IRS-1 S789A (CHO/IR/IRS-1 S789A ), respectively. After exposed to the liver extracts by in vitro kinase assay, wild type, but not IRS-1 S789A , was recognized by ␣IRS1 PS789 (Fig. 7B, lanes a-d versus  f-g). Although both substrates were phosphorylated by Erk-2 equally well (Figs. 2A, lane e, and 4B, panels I and II, lane e), they were not recognized by ␣IRS1 PS789 (Fig. 7B, lanes e and h). Thus, ␣IRS1 PS789 is specific for Ser(P) 789 of IRS-1 and does not recognize phosphorylated IRS-1 at serines other than Ser 789 .
We used this highly specific antibody to investigate liver sections from the lean and obese Zucker rats. Total IRS-1-and Ser 789 -phosphorylated IRS-1 were analyzed in situ by immunofluorescence staining of liver sections from 10-week-old Zucker fatty rats and lean controls (n ϭ 3 each group). Utilizing large batch staining and sampling analysis with identical confocal imaging parameters for each sample, we observed a decreased total IRS-1 immunofluorescence signal but an increased Ser 789 phosphorylation signal in the obese rat liver when compared with their lean controls (Fig. 8A, panels a-d).
Semiquantitative analyses of IRS-1 and IRS-1 PS789 immunofluorescence signals consistently revealed a 76% decreased IRS-1 immunoreactivity and 247% increased IRS-1 PS789 immunoreactivity in the obese group versus lean controls (Fig. 8B). Similar results were obtained when JCR:LA-cp rats were examined (data not shown).
IRS-1 total protein or Ser 789 -phosphorylated IRS-1 were also immunoprecipitated from liver lysates of Zucker fatty rats and lean controls (n ϭ 3 each group) by ␣IRS-1 or ␣IRS1 PS789 , followed by immunoblotting analysis with the same antibodies. In Zucker fatty rats, there was a lower level of total IRS-1 (58% of lean control) (Fig. 9, A, lanes b, d, and f, versus lanes a, c, and  e, and C). In contrast, Ser 789 phosphorylation of IRS-1 as detected by ␣IRS1 PS789 was significantly increased in the liver of Zucker fatty rats (191%) opposed to the lean controls (Fig. 9, B,  lanes b, d, and f, versus lanes a, c, and e, and C). Together with the immunofluorescent staining, these data demonstrate, for the first time, that Ser 789 phosphorylation of IRS-1 is significantly increased in the liver of insulin-resistant rat models.

DISCUSSION
Serine phosphorylation of IRS-1 has been implicated as a mechanism of attenuated insulin signaling, the so-called "insulin resistance" (20 -29, 37). Identification of serine phosphorylation sites on IRS-1 and the kinase that phosphorylates IRS-1 are important steps toward the understanding of the molecular mechanism of insulin resistance. We previously identified enhanced serine kinase activity in liver extracts of JCR:LA-cp obese rats using an in vitro kinase assay combined with phosphotryptic peptide mapping analysis (40). The current study identified that the phosphorylation site for this kinase is Ser 789 of IRS-1. Furthermore, we now demonstrate in vivo for the first time that Ser 789 phosphorylation of IRS-1 is increased in liver of the insulin-resistant rat model using both immunofluorescence staining or immunoblotting analysis. The fact that identical results were obtained in two genetically unrelated insulin-resistant rat models, JCR:LA-cp and Zucker rats (41,49), suggests that our findings may generally apply to the insulin-resistant state. Conclusive evidence of increased Ser 789 phosphorylation of IRS-1 in vivo together with the in vitro detection of enhanced serine kinase activity in both insulin-resistant animal models leads us to speculate that this as yet unidentified serine kinase, which physiologically or pathophysiologically modulates insulin sensitivity, is a mechanism of impaired insulin effectiveness in liver.
IRS proteins are important signaling molecules that become tyrosyl-phosphorylated during insulin stimulation and activate the insulin signaling network through interaction with downstream signaling molecules, including PI3K, Grb-2/Sos, and SHP-2 (5,57). Recent studies in cultured cells have shown that IRS-1 becomes serine-phosphorylated after prolonged exposure to many factors, including insulin, TNF-␣, glucose, or free fatty acids, and consequently fails to become tyrosyl-phosphorylated, resulting in attenuation of the insulin response (20 -29, 37). Based on these findings, an attractive hypothesis has emerged that serine phosphorylation of IRS proteins is a cause of insulin resistance. Searching for IRS-1 serine kinases led to the identification of many serine kinases, including casein kinase II, PI3K, Akt and mTor (33,37,38,58). Glycogen-synthase kinase 3 and MAPK have also been reported to phosphorylate IRS-1 in vitro and in vivo and to impair insulin action in cultured cells (31,32). Recently, TNF-␣-induced serine phosphorylation of IRS-1 has been mapped to Ser 307 in IRS-1, which is phosphorylated in vitro by c-Jun N-terminal kinase (36). Mutation on this site completely abolished the ability of TNF-␣ to induce insulin resistance in the mutant cells (36,59). However, the relevance of this phosphorylation on IRS-1 to the insulin resistance in animal models has not been shown.
We thus took an alternate approach to investigate serine phosphorylation of IRS-1 using tissue extracts from insulinresistant rodents. Most importantly, we used two genetically distinct models, Zucker fatty (fa/fa) rats and JCR:LA-cp rats (41, 49 -52, 60). We established an in vitro kinase assay using GST-IRS-1 fragments as substrates and tissue extracts as the kinase source (40) and identified an enhanced serine kinase activity, in liver extracts from both obese animals, that phosphorylates IRS-1 at a region between amino acid residues 526 and 859. The current study has determined the Ser 789 residue in IRS-1 as the phosphorylation site for this serine kinase based on several criteria. First, radioamino acid sequencing analysis data predicted that Ser 789 was the phosphorylation site. Second, GST-IRS1-(765-816) eliminated all the MAPK phosphorylation sites without altering phosphorylation by the enhanced serine kinase activity. Third, mutation of Ser 789 in IRS-1 completely abolished phosphorylation by the enhanced serine kinase activity. Furthermore, Ser 789 appears to be the only major phosphorylation site in IRS-1 for the enhanced serine kinase activity, because full-length IRS-1 produced one predominant tryptic phosphopeptide (P3) that was identical to that derived from phosphorylated GST-IRS1-(526 -859) or GST-IRS1-(765-816) on Tricine SDS-PAGE and HPLC profiles.
The previously cited studies have shown that various reagents (TNF-␣, insulin, and glucose) promote serine phosphorylation of IRS-1, but there has been no in vivo evidence prior to this study supporting the validity of these findings to the insulin resistance in animal models. Mapping the serine phosphorylation site in IRS-1 to Ser 789 allowed us to generate a phosphoserine-specific antibody (␣IRS1 PS789 ) for in vivo testing. The specificity of ␣IRS1 PS789 was confirmed by several criteria. First, ␣IRS1 PS789 did not recognize GST-IRS1-(765-816) protein even in a high concentration (0.5 g/lane) by immunoblotting unless it was phosphorylated by liver extracts in vitro in the presence of ATP. Second, ␣IRS1 PS789 recognized wild type IRS-1 by immunoblotting when phosphorylated in vitro by the liver extracts, but mutation on Ser 789 of IRS-1 completely eliminated this recognition. Because both wild type IRS-1 and IRS-1 S789A in CHO/IR/IRS-1 or CHO/IR/IRS-1 S789A cells are heavily phosphorylated on serines in the basal state (manifested by the identical high molecular mass of 185 kDa on SDS-PAGE gel) (40,42), our data show that ␣IRS1 PS789 is highly specific for Ser(P) 789 and does not recognize other phosphoserines in IRS-1.
␣IRS1 PS789 was subsequently used to assess the Ser 789 phosphorylation of IRS-1 in vivo using liver slices from normally sensitive and insulin-resistant Zucker rats. Using semiquanti- tative immunofluorescence imaging, we clearly demonstrated that Ser 789 phosphorylation of IRS-1 markedly increased in livers of the two studied insulin-resistant animal models. Increased Ser 789 phosphorylation of IRS-1 in insulin-resistant animals was further confirmed by immunoprecipitation and immunoblotting analysis. This finding is dramatic, because the level of IRS-1 protein was decreased in the insulin-resistant obese rats as reported from other laboratories (15,18). Increased Ser 789 phosphorylation of IRS-1 in vivo in insulinresistant states leads us to speculate that the enhanced serine kinase activity could be the potential molecular linker for insulin resistance in livers of these animals.
The identity of the responsible serine kinase is currently unknown. Serine 789 exists in a highly conserved motif among human, rat, and mouse IRS-1 (Fig. 5) and is not part of a known motif for MAPK, JNK, GST-3, casein kinase II, PI3K, Akt, or mTor. Interestingly, this motif is similar to one of the phosphorylation motifs for AMP-activated kinase: YXXXXSXXXY, where Y is a hydrophobic residue (Met, Val, Leu, Ile, or Phe) (43,54). The fact that AMPK is able to phosphorylate IRS-1 at Ser 789 (39) raises the possibility that the serine kinase might be an AMPK. However, our results disagree with this possibility. First, the enhanced serine kinase activity in the insulin-resistant animal is not activated by AMP. Second, AMP-activated kinase activity is suppressed in the liver extracts where the enhanced serine kinase activity is readily detected. Third, AMPK is not enhanced in insulin-resistant animals, consistent with the literature (55). Finally, the enhanced serine kinase cannot be depleted from tissue extracts by anti-AMPK immunoprecipitation. Collectively, these data suggest that, although it can phosphorylate IRS-1 on Ser 789 in vitro, AMPK activity is not the operative serine kinase in the studied insulin-resistant animals. The definitive identification will depend on the purification and characterization of this serine kinase.
MAPK has been shown to phosphorylate IRS-1 in vitro and in vivo at Ser 612 , leading to the inhibition of subsequent tyrosine phosphorylation by the insulin receptor (30,31,61). Moreover, the phosphorylation of a synthetic IRS-1 peptide containing Ser 612 by liver extracts was significantly higher in ob/ob mice than in lean controls (30). We did not detect enhanced MAPK activity in liver extracts from either the JCR:LA-cp obese rats or Zucker fatty rats, even when myelin basic protein was used as a substrate (data not shown). We do not believe that this discrepancy was due to the preparation procedure of liver extracts (50% (NH 4 ) 2 SO 4 precipitation), because this same procedure was able to enrich MAPK activity in rat liver (40). Thus, our results do not support MAPK being the operative kinase in the livers of the studied insulin-resistant rat models.
The biochemical mechanism by which serine phosphorylation attenuates insulin signaling is an intriguing issue that currently lacks definitive explanation. Increased serine phosphorylation of IRS-1 or IRS-2 has been shown to disrupt its binding to the juxtamembrane region of the insulin receptor and impair its ability to undergo insulin-induced tyrosine phosphorylation (38,62). It also disrupts the intracellular distribution of IRS proteins, resulting in a dissociation of IRS proteins from the low density microsomes and subsequent degradation (63,64). Therefore, serine phosphorylation of IRS-1 may disrupt the normal interaction of IRS-1 with other proteins or lipids that are required for insulin signaling.
Recently, serine/threonine phosphorylation has been found to involve protein-protein interactions that play an important role in signal transduction (65,66). WD40 domains, leucinerich repeats in some F-box proteins, and WW domains are able to directly bind to phosphoserine residues in proteins that are targeted for ubiquitin-proteasome-mediated degradation (65,67,68). A decreased level of IRS-1 protein has been reported in numerous insulin-resistant animal models and non-insulindependent diabetes mellitus patients (15)(16)(17)(18)(19)69). We have recently reported that proteasome-mediated degradation of IRS-1 may be one of the possible mechanisms for controlling cellular levels of IRS-1 (70,71). It is possible that phosphorylation of Ser 789 links IRS-1 to the ubiquitin-proteasome degradation pathway. This possibility is currently under investigation.
In summery, our results demonstrated in vitro and in vivo for the first time that a novel serine kinase activity assessed by Ser 789 phosphorylation of IRS-1 closely correlates with the insulin-resistant state in two genetically unrelated insulinresistant animal models. We speculate that this potential serine kinase may physiologically or pathophysiologically modulate levels of phosphorylation of IRS-1 at Ser 789 and thus the insulin sensitivity.