O-Linked N-Acetylglucosamine Modification of Insulin Receptor Substrate-1 Occurs in Close Proximity to Multiple SH2 Domain Binding Motifs*

Insulin receptor substrate-1 (IRS-1) is a highly phosphorylated adaptor protein critical to insulin and IGF-1 receptor signaling. Ser/Thr kinases impact the metabolic and mitogenic effects elicited by insulin and IGF-1 through feedback and feed forward regulation at the level of IRS-1. Ser/Thr residues of IRS-1 are also O-GlcNAc-modified, which may influence the phosphorylation status of the protein. To facilitate the understanding of the functional effects of O-GlcNAc modification on IRS-1-mediated signaling, we identified the sites of O-GlcNAc modification of rat and human IRS-1. Tandem mass spectrometric analysis of IRS-1, exogenously expressed in HEK293 cells, revealed that the C terminus, which is rich in docking sites for SH2 domain-containing proteins, was O-GlcNAc-modified at multiple residues. Rat IRS-1 was O-GlcNAc-modified at Ser914, Ser1009, Ser1036, and Ser1041. Human IRS-1 was O-GlcNAc-modified at Ser984 or Ser985, at Ser1011, and possibly at multiple sites within residues 1025–1045. O-GlcNAc modification at a conserved residue in rat (Ser1009) and human (Ser1011) IRS-1 is adjacent to a putative binding motif for the N-terminal SH2 domains of p85α and p85β regulatory subunits of phosphatidylinositol 3-kinase and the tyrosine phosphatase SHP2 (PTPN11). Immunoblot analysis using an antibody generated against human IRS-1 Ser1011 GlcNAc further confirmed the site of attachment and the identity of the +203.2-Da mass shift as β-N-acetylglucosamine. The accumulation of IRS-1 Ser1011 GlcNAc in HEPG2 liver cells and MC3T3-E1 preosteoblasts upon inhibition of O-GlcNAcase indicates that O-GlcNAcylation of endogenously expressed IRS-1 is a dynamic process that occurs at normal glucose concentrations (5 mm). O-GlcNAc modification did not occur at any known or newly identified Ser/Thr phosphorylation sites and in most cases occurred simultaneously with phosphorylation of nearby residues. These findings suggest that O-GlcNAc modification represents an additional layer of posttranslational regulation that may impact the specificity of effects elicited by insulin and IGF-1.

Insulin receptor substrate-1 (IRS-1) 1 is a highly phosphorylated adaptor protein critical to insulin and IGF-1 receptor signaling. Many of the metabolic and mitogenic effects elicited by insulin and IGF-1 are mediated and modulated by posttranslational modifications of IRS-1, and tight regulation at the posttranslational level is crucial for maintaining insulin sensitivity and controlling growth factor-induced proliferation. Following hormonal stimulation, IRS-1 is phosphorylated by the receptor tyrosine kinases creating SH2 domain docking sites for downstream binding partners including the p85 regulatory subunits of phosphatidylinositol 3-kinase, Grb2, and the tyrosine phosphatase SHP2 (PTPN11) (1). Binding of p85 phosphatidylinositol 3-kinase and Grb2 activate the PI3K/Akt and Ras-MAPK pathways, respectively, whereas binding of SHP2 results in tyrosine dephosphorylation and signal attenuation (2). Positive and negative feedback regulation by Ser/ Thr kinases, such as Akt (3), c-Jun N-terminal kinase (JNK) (4), S6K (5), and ERK (6), impact the interactions of IRS-1 with SH2 domain proteins and the receptor thereby affecting the duration and outcome of the signal. IRS-1 has been described as a central node for the integration of information regarding the nutrient and stress status of the cell (7). This information is encoded by site-specific phosphorylation by a number of kinases that regulate the specificity of effects that are elicited following receptor stimulation. Many sites of Ser/Thr phosphorylation have been identified on IRS-1, and cross-talk among Tyr and Ser/Thr phosphorylations at specific residues is evidence of dynamic and complex posttranslational regulation (8,9). Inappropriate phosphorylation of IRS-1 resulting in the disruption of interactions of IRS-1 with binding partners is implicated in the development of insulin resistance (10) and altered IGF-1 signaling in breast cancer tissue (11,12).
In addition to phosphorylation, Ser/Thr residues in IRS-1 are also dynamically modified by GlcNAc in a nutrient-responsive manner. As opposed to a negatively charged phosphate group, O-GlcNAcylation imparts a bulky, hydrophilic, electrostatically neutral moiety to Ser/Thr residues. The enzymes responsible for the incorporation and removal of the monosaccharide from proteins, O-GlcNAc-transferase and O-Glc-NAcase, respectively, are localized in the cytoplasm and the nucleus of all eukaryotic cells (13,14). Recent studies suggest that the activity of O-GlcNAc-transferase is regulated by insulin (15) and that localization of O-GlcNAc-transferase to the membrane is driven by direct association with phosphatidylinositide 3-phosphate (16). The abundance of O-GlcNAc modification on many proteins in the insulin signaling pathway increases with sustained high glucose and chronic insulin stimulation, and elevated O-GlcNAc modification of IRS-1 correlates with the development of insulin resistance in multiple cell types including 3T3-L1 adipocytes (17,18), MIN6 pancreatic beta cells (19), Fao rat hepatoma cells (16), human aortic endothelial cells (20), and skeletal muscle (21). The impact of O-GlcNAcylation on insulin signaling and diabetic complications was reviewed recently (22,23). The direct effect of O-GlcNAc modification on signaling via IRS-1 is not known because conditions that mimic those in the uncontrolled diabetic patient may also result in phosphorylation of IRS-1 at inhibitory sites (16,24) and O-GlcNAc modification of other proteins in the insulin signaling pathway, such as the insulin receptor, Akt (18), FoxO (25), AMP-activated protein kinase (26), and ␤-catenin (17).
To elucidate site-specific effects of O-GlcNAc modification on IRS-1-mediated signal transduction, we identified the sites of O-GlcNAc modification of rat and human IRS-1 by tandem mass spectrometry. To facilitate detection of the O-GlcNAcmodified peptides and assign the sites of modification, CID coupled with neutral loss-triggered MS3 and electron transfer dissociation (ETD) (27) tandem spectrometric approaches were used. Fragmentation of O-GlcNAc-modified peptides by ETD did not destroy the labile O-linkage (28) permitting direct detection of these peptides by the database searching algorithm ProteinProspector2 (29). O-GlcNAc modification occurred in close proximity to multiple SH2 domain binding motifs and within a region of IRS-1 shown previously to interact with the insulin and IGF-1 receptors (30).

EXPERIMENTAL PROCEDURES
Generation of Wild Type and Mutant Rat and Human IRS-1 cDNAs-His-and S-tagged rat IRS-1, amino acid residues 6 -1235, in pTriEX vector (31) served as a template for site-directed substitution of O-GlcNAc-modified serine residues with alanine. Rat IRS-1 S1036A, S1041A, S1036A/S1041A, and S1009A were generated using the oligonucleotide primers 5Ј-gccccccctccatccgccacagctagcgcttctgcttctgttacacct, 5Ј-gccccccctccatcctccacagctagcgctgctgcttctgttacacct, 5Ј-gccccccctccatccgccacagctagcgctgctgcttctgttacacct, and 5Ј-ccagtggccccagtcgcatatgctgacatgcgg, respectively, using the QuikChange kit (Stratagene) according to the manufacturer's recommendations. The cDNA encoding C-terminally hemagglutinin-tagged human IRS-1 was kindly provided by Adrian Lee (Baylor University) (27). Initial mass spectrometric analysis of the expressed protein did not reveal the sites of O-GlcNAc modification; therefore, the human IRS-1 cDNA was cloned into the NotI and PmlI sites of pTriEX4 vector (Novagen) to generate an N-terminally His-and S-tagged protein. A stop codon was inserted at the 3Ј end of the coding sequence to remove the C-terminal hemagglutinin tag from the expressed protein.
Mutagenesis reactions and manipulations of the cDNA were verified by sequencing.
Cell Culture and Expression of IRS-1-HEK293 cells were maintained in Iscove's modified Eagle's medium (BioSource) containing 10% heat inactivated FCS with 1% antibiotic/antimycotic (Invitrogen). HEK293 cells were transiently transfected with rat or human IRS-1 cDNA using Polyfect transfection reagent according to the manufacturer's recommendations (Qiagen). To facilitate mass spectrometric detection of O-GlcNAc-modified peptides, cells grown in 5 mM glucose were incubated with an O-GlcNAcase inhibitor (50 M O-(2acetamido-2-deoxy-D-glucopyranosylidene) amino-N-phenylcarbamate (PUGNAc); Toronto Chemicals) for 18 h prior to lysis. For coimmunoprecipitation experiments, cells were serum-starved for 16 h in the presence of absence of PUGNAc and then stimulated with 100 nM insulin or 10 ng/ml IGF-1 (Sigma) for varying times. Cells were lysed in 300 mM NaCl, 50 mM sodium phosphate, pH 8, 10 mM imidazole, 10 mM ␤-mercaptoethanol, 1% Nonidet P-40, 50 M PUGNAc, EDTA-free protease inhibitor mixture (Roche Applied Science), 10 mM sodium fluoride, 500 M sodium vanadate, 1 mM sodium pyrophosphate. His-tagged IRS-1 was enriched by incubation with nickel-nitrilotriacetic acid affinity resin (Novagen) for 1 h at room temperature followed by extensive washing with lysis buffer containing 20 mM imidazole. The protein was eluted in lysis buffer containing 250 mM imidazole and gel-purified by electrophoresis on 4 -10% XT Criterion gels (Bio-Rad). For mass spectrometric analysis, the protein was stained with E-zinc (Pierce), and the IRS-1 band was excised from the gel. Gel pieces were washed twice for 10 min with 100 mM ammonium bicarbonate, dehydrated with acetonitrile, and dried under vacuum. Cysteines were reduced with dithiothreitol and alkylated with iodoacetamide for 30 min at room temperature in the dark. Gel pieces were dehydrated with acetonitrile and dried completely prior to proteolytic digestion with trypsin (Promega) at 37°C or with Glu-C (Roche Applied Science) at 25°C for 18 h. Peptides were extracted with one wash of 25 mM ammonium bicarbonate for 20 min and three washes of 5% formic acid, 50% acetonitrile for 20 min each, dried under vacuum, reconstituted in 0.1% TFA, and cleaned up on C 18 microspin columns (Nest Group). Prior to LC/MS analysis, the samples were reconstituted with 10 l of 2% acetonitrile, 0.2% formic acid.
Immunoprecipitation and Western Blotting-Wild type and mutant IRS-1 protein, isolated by nickel affinity chromatography or immunoprecipitation, was separated by SDS-PAGE and transferred to nitrocellulose. Membranes were blocked with nonfat dry milk in Trisbuffered saline for detection by ECL or Odyssey blocking buffer (Licor Biosciences, Inc.) prior to fluorescence detection. Antibodies used for immunoprecipitation and immunoblotting of endogenously expressed IRS-1 were purchased from Bethyl Laboratories (A301-158) and Santa Cruz Biotechnology (A-19), respectively. A pan-specific mono-clonal anti-O-GlcNAc (CTD 110.6) IgM generated against YSPTS(O-GlcNAc)PSK (32) was purchased for Western blot (Pierce) or immunoprecipitation (Covance). The presence of recombinant wild type or mutant IRS-1 was detected by probing for the S-tag with S-protein conjugated to HRP (Novagen) and visualized using West Pico enhanced chemiluminescent reagent (Pierce). Fluorescently labeled secondary antibodies, IRDye 800CW and 680, were purchased from Licor Biosciences, Inc.
Reversed Phase HPLC and Mass Spectrometry-Digested peptides were separated by C 18 reversed phase nano-LC using a 75m ϫ 15-cm capillary column packed in house (YMC ODS-AQ 120A S5, Waters) with a gradient of 2-60% B in 110 min (B ϭ 0.02% heptafluorobutyric acid or 0.2% formic acid in acetonitrile) at 180 nl/min using an Ultimate 3000 nanoflow system with Chromeleon 6.8 software (Dionex, Sunnyvale, CA). Eluted peptides were analyzed by CID or ETD MS/MS with an LTQ XL ion trap mass spectrometer equipped with ETD capability (ThermoFisher). Multiple mass spectrometric approaches were utilized to detect O-GlcNAc-modified peptides and confirm the assignments of the sites of O-GlcNAc modification. Data-dependent neutral loss-triggered MS3 acquisition facilitated the identification of candidate O-GlcNAc-modified peptides. The instrument was programmed to acquire MS/MS/MS (MS3) data on precursor ions that exhibited a neutral loss corresponding to N-acetylglucosamine (203.2, 101.6, 67.7, or 50 Da for the 1ϩ, 2ϩ, 3ϩ, and 4ϩ charge states, respectively) upon collision-induced dissociation. A normalized collision energy of 35% was used for fragmentation by CID. The threshold for the selection of an ion for fragmentation was set to 500. In subsequent analyses, the instrument was programmed to acquire tandem mass spectra generated by alternating between fragmentation by CID and ETD or by ETD only on the five most abundant ions in the full mass spectrum survey scan (m/z range, 400 -2000). The parameters used for ETD were as follows: 100-ms activation time; emission current, 130 A; automatic gain control, 30,000; temperature, 170°C; supplemental activation was turned off; isolation width, 4 Da. With the exception of LC-MS/MS analyses performed with a parent mass inclusion list, dynamic exclusion was enabled to exclude ions from MS/MS selection for 3 min after being selected three times in a 30-s window.
Tandem mass spectra were searched against a database of human or rat IRS-1 using Bioworks 3.3.1 for the following variable modifications: phosphorylation at serine, threonine, or tyrosine; O-GlcNAc modification of serine and threonine (ϩ203.2 Da); and oxidation of methionine. Carbamidomethylcysteine was included as a static modification. Because of the labile nature of O-GlcNAc glycosylation, CID spectra also were searched for variable N-or C-terminal O-GlcNAc modification (ϩ203.2 Da) to mimic the partial to complete loss of the GlcNAc from the precursor and fragment ions upon collision-induced dissociation. Unlike the neutral loss of phosphoric acid (98 Da) from phosphoserine resulting in a serine that has lost the side chain hydroxyl group, neutral loss of the monosaccharide from O-GlcNAc serine leaves the serine intact. Thus, MS3 spectra acquired following neutral loss of N-acetylglucosamine were also searched using Bioworks. ETD MS/MS spectra were searched by Protein Prospector2 (29) for phosphorylation or O-GlcNAc modification of serine and threonine and for unanticipated posttranslational modifications. The exact sites of modification were confirmed by manual interpretation of CID and ETD spectra.
Peptide Synthesis and Antibody Production-A peptide corresponding to human IRS-1 residues 1003-1019 with an O-linked-␤-Nacetylglucosamine serine at 1011, Ac-CDTSPAAPVS(O-GlcNAc)YA-DMRTGI-amide, was synthesized and purified by Sussex Research (Ontario, Canada). An N-terminal Cys residue was included for conjugation of the peptide to keyhole limpet hemocyanin carrier protein, and a polyclonal antibody was generated in rabbit (Lampire Biological Laboratories, Pipersville, PA). To characterize the specificity of the antibody for the site of O-GlcNAc modification, wild type IRS-1 or a mutant of rat IRS-1 in which the site of O-GlcNAc modification was substituted with alanine (S1009A) was transiently expressed in HEK293 cells in the presence of the O-GlcNAcase inhibitor PUGNAc. Nickel-purified proteins were separated by gel electrophoresis and transferred to nitrocellulose, and the membranes were probed with a 1:3000 dilution of rabbit sera followed by a fluorescently labeled secondary IgG. The protein was visualized using an Odyssey instrument (Licor Biosciences, Inc.). Removal of the site of O-GlcNAc modification at Ser 1009 completely abolished antibody binding (supplemental Fig. 1). To confirm that the antibody recognized the monosaccharide and not the naked peptide backbone, a competition experiment was performed by preincubating the sera with either the unmodified or O-GlcNAc-modified peptide residues 1003-1019. Preincubation of the sera with the O-GlcNAc-modified peptide abolished antibody binding to rat IRS-1, whereas preincubation with the naked peptide did not alter antibody binding to IRS-1 (supplemental Fig. 1).
To check the specificity of the antibody for N-acetylglucosamine, a competition experiment was performed by preincubating the sera with 1 M N-acetylglucosamine (Sigma) prior to probing IRS-1 by Western blot. Preincubation of the antibody with the monosaccharide greatly reduced the staining of the IRS-1 protein. To further confirm that the antibody recognized the monosaccharide and not the naked peptide backbone, human IRS-1 isolated from PUGNAc-treated HEK293 cells was incubated in the presence or absence of ␤-Nacetyl-hexosaminidase (New England Biolabs). The reaction was performed at 37°C for 18 h in reaction buffer supplied by the manufacturer. The protein was separated by gel electrophoresis and probed by Western blot. Treatment of IRS-1 with hexosaminidase abolished antibody recognition of the protein (supplemental Fig. 1).

Ionization Efficiency and Estimated Stoichiometry of O-GlcNAcmodified Peptides-
The peptide Ac-Cys-1003-1019-amide was synthesized with and without the O-GlcNAc modification at Ser 1011 . The concentrations of the purified peptides were determined by amino acid analysis (Anaspec, San Jose, CA). To compare the ionization efficiencies and the possible loss of the labile monosaccharide in the ionization source of the mass spectrometer, equal molar ratios of O-GlcNAc-modified and unmodified peptides were combined and analyzed using the LC-MS/MS parameters described above. The peak areas of the unsmoothed extracted ion chromatograms corresponding to the doubly charged ions of the unmodified and O-GlcNAc-modified synthetic peptides were determined using the Qual browser feature in Xcalibur 2.0.5. Loading equimolar amounts of the synthetic peptides resulted in equal areas under the base peak ion chromatograms suggesting that the O-GlcNAc modification was not lost upon ionization in the source and that the O-GlcNAc-modified peptide was ionized as efficiently as the unmodified peptide (supplemental Fig. 3).
Estimates of the relative abundance of differentially modified peptides of rat IRS-1 isolated from cells treated with the O-GlcNAcase inhibitor PUGNAc were determined from the chromatographic peak areas of the extracted ion chromatograms corresponding to the modified and unmodified peptides. The stoichiometry of modification was expressed as a percentage of the chromatographic peak area of the modified peptide to a sum of peak areas of ions corresponding to the unmodified and modified peptides.
Phosphopeptide Interaction with SH2 Domain Protein Array-To determine candidate SH2 domain-containing proteins that may interact with a putative SH2 domain binding motif at Y 1012 ADM, an SH2 domain protein array was probed with a synthetic peptide corresponding to human IRS-1 phosphorylated at Tyr 1012 . The peptide Ac-Cys-1003-1019-amide phospho-Tyr 1012 , synthesized and purified by Anaspec, was biotinylated using maleimide-polyethylene glycol-biotin according to the manufacturer's instructions (Pierce). The biotinylated peptide was purified from the reaction mixture by reversed phase C 18 chromatography (C 18 Platinum EPS, 4.6 mm ϫ 25 cm, Alltech). The collected fractions were analyzed by MALDI-TOF-MS (Autoflex III Smartbeam, Bruker Daltonics, Billerica, MA), and the fraction containing the pure biotinylated peptide was dried under vacuum. The SH2 domain protein array included 38 different recombinant SH2 domains from 33 proteins immobilized on a membrane. Each SH2 domain was spotted in duplicate at 100 ng. Prior to incubation with the membrane, the biotinylated peptide was reconstituted in 15 l of water and 15 l of streptavidin-HRP conjugate and incubated for 30 min followed by the addition of 5 l of blocking buffer according to the manufacturer's instructions (Panomics, Fremont, CA). The SH2 domain protein array membrane was blocked and subsequently incubated with the HRP-labeled peptide overnight at 4°C. The membrane was washed thoroughly, and peptide binding to the protein array was visualized by enhanced chemiluminescence.

RESULTS
To identify the sites of O-GlcNAc modification of human and rat IRS-1, the proteins were transiently expressed in HEK293 cells grown in the presence of the O-GlcNAcase inhibitor PUGNAc. His-tagged IRS-1 was enriched by nickel affinity chromatography and gel-purified. The IRS-1 protein was excised from the gel and digested with trypsin and/or Glu-C. Extracted peptides were separated and analyzed by LC-MS/MS using collision-induced dissociation with neutral loss-triggered MS3 and electron transfer dissociation. Trypsin digestion yielded 84 and 75% sequence coverage of the 131-kDa rat and human IRS-1 proteins, respectively.
O-GlcNAc Modification of Rat IRS-1 at Ser 1036 and Ser 1041 and Human IRS-1 Residues 1025-1045-O-GlcNAc modification of Ser 1036 was identified previously by a chemical derivatization approach utilizing ␤-elimination and Michael addition with dithiothreitol (31). However, the presence of O-GlcNAc modification elsewhere in the protein, which has 181 Ser and 63 Thr residues, was evident following immunoblot analysis of a mutant of IRS-1 in which the site of O-GlcNAc modification had been substituted with alanine (S1036A). Removal of this site of O-GlcNAc modification reduced but did not completely abolish immunoreactivity of the pan-specific anti-O-GlcNAc antibody CTD 110.6 against IRS-1 as determined by Western blot (31). To facilitate the detection of other sites of O-GlcNAc modification, rat IRS-1 was analyzed by ETD MS/MS. This methodology is amenable to the detection of labile amino acid modifications and leaves O-GlcNAc Ser intact during fragmentation of the peptide backbone for MS/MS analysis (28). The fragmentation pattern of the O-GlcNAcylated tryptic peptide 1027-1074 of rat IRS-1 by ETD is consistent with the co-elution of a mixture of mono-GlcNAc-modified peptides in which only the site of modification at Ser 1036 could be resolved (Fig. 1A). LC-MS/MS analysis of the S1036A mutant of IRS-1 revealed a second site of O-GlcNAc modification within residues 1032-1041 (Fig. 1B). The observed molecular mass of the O-GlcNAc-modified peptide 1021-1051 S1036A following Glu-C digestion was 3083.  Table I). These modifications flank the SH2 domain binding motif at Y 895 VN, which when phosphorylated by the insulin receptor creates a docking site for Grb2 and signal transduction to the Ras-MAPK pathway (33). The CID tandem mass spectrum of the doubly charged precursor at m/z 1419.2 confirmed the previously reported phosphorylation at Ser 891 (34) and revealed O-GlcNAc modification at Ser 914 (Fig. 3A). Following collision-induced dissociation, the major fragment ions in the spectrum at m/z 1317.7 and 1268.8 correspond to neutral loss of the GlcNAc and the subsequent loss of phosphoric acid, respectively. Partial retention of GlcNAc on the y series of ions generated a complex tandem mass spectrum that gave poor peptide probability scores by automated database searching algorithms, such as SEQUEST. This behavior contributes to the difficulty in identifying O-GlcNAc-modified peptides by conventional MS/MS sequencing approaches. This peptide was detected as a result of triggering the acquisition of an MS3 spectrum following neutral loss of the monosaccharide. The chromatographic separation and an estimate of the relative stoichiometry of the differentially modified forms of this peptide of IRS-1 isolated from cells treated with the O-GlcNAcase inhibitor are shown in Fig. 3B. O-GlcNAc modification of the homologous human IRS-1 peptide, 892-922, was not observed. However, this peptide was phosphorylated at Ser 892 (34), and a novel site of phosphorylation was identified at Ser 918 in human IRS-1 (supplemental Fig. 4).
O-GlcNAc Modification of Human IRS-1 Residues 981-998 -Human IRS-1 was O-GlcNAc-modified at Ser 984 or Ser 985 (Table I and Fig. 4). This is in close proximity to the SH2 domain binding motif, which when phosphorylated at Y 989 MTM by the insulin receptor creates a docking site for the p85 subunit of PI3K (35). This region of IRS-1, residues 950 -986, has also been shown to interact with the insulin and  22 ion that retained the monosaccharide, the fragmentation pattern is consistent with the assignment of the O-GlcNAc modification at residue 1041, which was further confirmed by site-directed mutagenesis. b and y ions are labeled according to the unmodified peptide. Fragment ions that retained the GlcNAc moiety are indicated with an asterisk. Within the peptide sequence shown, the amino acid that was substituted by site-directed mutagenesis, S1036A, is underlined, and the newly identified site of O-GlcNAc modification is indicated with a "g." IGF-1 receptors (30). The presence of the b 8 GlcNAc ion in the CID tandem mass spectrum of the doubly charged precursor ion at m/z 1139.9 confirmed that the modification is within the N terminus of the peptide presumably at either Ser 984 or Ser 985 (Fig. 4). The unmodified peptide yielded a similar fragmentation pattern with the most abundant ions resulting from dissociation at the Asp-Tyr bond generating the b 8 and y 10 ions (supplemental Fig. 5).  (Table I and Fig. 5). The CID tandem mass spectrum of the O-GlcNAcylated peptide 997-1014 (precursor ion at m/z 1095.3) is consistent with O-GlcNAc modification at Ser 1009 (Fig. 5A). The assignment of the site of O-GlcNAc modification was further confirmed following fragmentation of the peptide by electron transfer dissociation (Fig. 5B). The CID tandem mass spectrum of the doubly charged precursor ion at m/z 1135.0 is consistent with O-GlcNAc modification at Ser 1009 and phosphorylation at Ser 1003 (Fig. 5C). The chromatographic separation and rela- Human IRS-1 was phosphorylated and O-GlcNAc-modified at the homologous residues Ser 1005 and Ser 1011 , respectively. Interestingly, phosphorylation at Ser 1005 decreases in euglycemic hyperinsulinemic clamp studies (36) indicating that modifications within this region of IRS-1 are involved in acute responses to insulin in vivo. The CID tandem mass spectrum of human IRS-1 residues 999 -1016 at m/z 1081.7 is consistent with the assignment of O-Glc-NAc modification at Ser 1011 (Fig. 5D).

O-GlcNAc Modification of Rat IRS-1 at Ser 1009 and Human
The presence of this site-specific glycosylation in endogenously expressed IRS-1 was confirmed by Western blotting and tandem mass spectrometry. IRS-1 was immunoprecipitated from HEPG2 human hepatoma cells or murine MC3T3-E1 osteoblast precursor cells grown in 5 mM glucose in the absence or presence of PUGNAc. Tandem mass spectrometric analysis of IRS-1 isolated from MC3T3-E1 cells confirmed O-GlcNAc modification of mouse IRS-1 at Ser 1005 (supplemental Fig. 6) Using an antibody generated against a synthetic peptide corresponding to human IRS-1 residues 1003-1019 Ser 1011 GlcNAc (supplemental Fig. 1), the immu-  (37), although this YXXM motif, which is preceded by hydrophobic amino acids and does not possess the Tyr ϩ 1 methionine residue, has not been observed by in vitro phosphorylation by the insulin receptor nor has phosphorylation at this residue been detected in recent studies characterizing the temporal dynamics of insulin-stimulated tyrosine phosphorylation (9,35,38). This motif is one of the nine YXXM motifs in IRS-1, six of which are Phosphorylated forms of these peptides are also shown. p, phosphorylation, g, O-GlcNAc modification. References previously reporting sites of posttranslational modification are indicated in parentheses. The following chemical modifications occurred during sample preparation (data not shown). Peptides with N-terminal glutamine were partially cyclized to pyroglutamic acid. Peptides were observed with and without oxidation of methionine. Residues 892-922 were observed with ϩ126 Da at Tyr 895 most likely due to iodination occurring during alkylation of Cys with iodoacetamide. BEMAD, ␤-elimination and Michael addition with dithiothreitol. YMXM. To ascertain candidate binding partners for this particular motif, we probed an SH2 domain protein array with an HRP-labeled synthetic peptide corresponding to human IRS-1 residues 1003-1019, phospho-Tyr 1012 (Fig. 7). The peptide interacted with the N-terminal SH2 domains of the p85␣ and p85␤ regulatory subunits of PI3K and, consistent with phosphotyrosine peptide pulldown experiments in C2C12 cells (39), with the N-terminal SH2 domain of the tyrosine phosphatase SHP2 (PTPN11).
Because O-GlcNAc modification and tyrosine phosphorylation of IRS-1 occur at substoichiometric levels and the protein has been shown to be localized to different subcellular pools depending on the cell type, we questioned whether these modifications were occurring on the same pool of IRS-1. To probe the O-GlcNAc-modified fraction of IRS-1 for insulin-stimulated tyrosine phosphorylation, rat IRS-1 was isolated from HEK293 cells that had been incubated overnight with the O-GlcNAcase inhibitor in the absence or presence of 100 nM insulin. Total Fragmentation of peptides by CID and ETD revealed four sites of O-GlcNAc modification at residues Ser 914 , Ser 1009 , Ser 1036 , and Ser 1041 in rat IRS-1. Because not all of these residues are conserved in human IRS-1, the human homolog was also characterized. Analysis of human IRS-1 revealed novel sites of phosphorylation at Ser 918 , Ser 1041 , and Ser 1043 and O-GlcNAc modification at Ser 984 or Ser 985 , at Ser 1011 , and possibly at multiple sites between residues 1025-1045 (Table I). A sequence alignment of rat and human IRS-1 (residues 892-1059) with the observed sites of posttranslational modification and the proximity of the O-GlcNAc-modified residues to SH2 domain binding motifs, interaction surfaces for the insulin and IGF-1 receptors, and two human polymorphisms (40,41) is shown in Fig. 8. Interestingly, the sites of O-GlcNAc modification observed occur within a region of the protein that is unique to IRS-1 and is not conserved in IRS-2.
O-GlcNAc modification of human IRS-1 at 984/985 occurs in close proximity to the DY 989 MTM docking site for the p85 subunit of PI3K. The effect of O-GlcNAc modification at the Tyr Ϫ 5 position on receptor-mediated phosphorylation at Tyr 989 and on SH2 domain interaction will be examined in future studies. This site of O-GlcNAc modification also occurs within residues 950 -986 to which the insulin and IGF-1 receptors have recently been shown to bind (30). A single amino acid polymorphism, G972R, within this region has been shown to impact the kinetics of the receptor-substrate interaction and autophosphorylation of the insulin receptor and attenuate signaling to both the PI3K/Akt and Grb2/ERK axes (30,36). Thus, it is conceivable that O-GlcNAc modification at this site could impinge on insulin signaling by multiple mechanisms.
O-GlcNAc modification of human IRS-1 at Ser 1011 is adjacent to a YXXM motif that resides within a sequence conserved in mammalian IRS-1 homologs, PVSYADM. The presence of O-GlcNAc modification at this site in endogenously expressed human and mouse IRS-1 was confirmed by Western blot analysis using a site-specific antibody. Based on the interaction of the corresponding phospho-Tyr peptide with the SH2 domains of p85 PI3K and SHP2, phosphorylation of Tyr 1012 may contribute to PI3K/Akt signaling, or conversely, it may facilitate Tyr dephosphorylation thereby attenuating signaling through IRS-1. Based on previous studies illustrating the critical role of other SH2 binding motifs in mediating the interactions of IRS-1 with PI3K (42) and SHP2 (2), the extent to which phosphorylation at this Tyr impacts insulin signaling is not clear. The effect of O-GlcNAc modification at Ser 1011 , which may alter phosphorylation at Tyr 1012 or the binding of SH2 domains to this motif, will be addressed in future studies.
As a highly posttranslationally modified adaptor protein that interacts with many binding partners, dynamic crosstalk among modifications of IRS-1 may contribute to specificity of effects elicited following receptor stimulation. Reversible posttranslational modifications acting either independently or in combination may induce or antagonize protein interactions by multiple mechanisms (for a review, see Ref. 43). O-GlcNAc modification mutually excludes phosphorylation of identical or adjacent residues in several proteins; therefore, we analyzed the phosphorylation state of the O-GlcNAc-modified peptides observed. In our mass spectrometric characterization of recombinant IRS-1 isolated from cells treated with PUGNAc or stimulated with insulin for varying times, we did not observe evidence for reciprocal or mutually exclusive O-GlcNAc modification and Ser/Thr phosphorylation within the proteolytic peptides ex-amined. In fact, most of the O-GlcNAc-modified peptides observed were also phosphorylated at nearby serines or threonines within consensus sequences for proline-directed kinases. Because of the difficulty in resolving the site(s) of O-GlcNAc modification of human IRS-1 to specific residue(s) within peptide 1035-1045, SSSSAASASPT, which is phosphorylated at Ser 1041 and Ser 1043 , we cannot rule out the occurrence of reciprocity of modifications within this peptide. The bottom up approach of proteolytically digesting the protein prior to MS analysis limits the observable connectivity of modifications to short peptide sequences. We probed for potential reciprocity of O-GlcNAc modification and insulin-stimulated tyrosine phosphorylation of the intact protein by immunoblot analysis and found that the O-GlcNAc-modified fraction of IRS-1 remained an available substrate for insulin-stimulated tyrosine phosphorylation. These observations suggest that rather than mutually exclusive site-specific competition with phosphorylation the O-GlcNAc modification may provide an additional layer to the complex posttranslational regulation of IRS-1.
The investigation of the sites and functional effects of O-GlcNAc modification on many proteins has been neglected because of the difficulty in detecting the modification by conventional tandem mass spectrometric approaches. However, electron transfer dissociation and robust search algorithms providing automated identification of O-GlcNAc-modified peptides will greatly facilitate the advancement of this field. Furthermore, middle down approaches utilizing ETD will permit analysis of the temporal dynamics of O-GlcNAc modification and connectivity of this modification to other modifications within larger peptide sequences.
The function of O-GlcNAc modification of IRS-1 is unclear. It may affect protein interactions either by blocking hydrophobic interactions, by recruiting a different subset of binding partners, or by altering subcellular localization. Furthermore, the effects may vary by the cellular milieu; e.g. they may be cell-specific. Given the critical role of the posttranslational modifications of IRS-1 in mediating and modulating insulin and IGF-1 receptor signaling, in vivo studies concerning the effects of the detected sites of O-GlcNAc modification on insulin and IGF-1 receptor signaling seem warranted. FIG. 8. Posttranslational modifications, polymorphisms, and protein interaction motifs in C terminus of IRS-1. The sequence alignment of human and rat IRS-1 is shown with the sites of O-GlcNAc modification indicated by hexagons. Ambiguous sites occurring within the overscored sequences are indicated with a question mark. Sites of phosphorylation are indicated with diamonds. Underlined residues are known sites of human polymorphism, G972R and S1043Y. The C-terminal region of human IRS-1 shown to interact with the insulin and IGF-1 receptors is indicated. Known sites of insulin receptor-mediated Tyr phosphorylation (asterisks) that induce interactions with the p85 subunit of PI3K, Grb2, or SHP2 are shown. The N-terminal SH2 domains of the p85␣ and p85␤ PI3K and SHP2 shown to interact with a phospho-Y 1012 ADM-containing peptide in this study are indicated ( †).