Global Identification of O-GlcNAc-ModifiedProteins
Animesh Nandi, Robert Sprung, Deb K. Barma, Yingxin Zhao, Sung Chan Kim, John R. Falck, andYingming Zhao,,*
Departments of Biochemistry and Pharmacology, University of Texas Southwestern Medical Center,Dallas, Texas 75390-9038
The O-linked N-acetylglucosamine (O-GlcNAc) modifica-tion of serine/threonine residues is an abundant post-translational modification present in cytosolic and nuclearproteins. The functions and subproteome of O-GlcNAcmodification remain largely undefined. Here we report theapplication of the tagging-via-substrate (TAS) approach forglobal identification of O-GlcNAc-modified proteins. TheTAS method utilizes an O-GlcNAc azide analogue formetabolic labeling of O-GlcNAc-modified proteins, whichcan be chemoselectively conjugated for detection andenrichment of the proteins for proteomics studies. Ourstudy led to the identification of 199 putative O-GlcNAc-modified proteins from HeLa cells, among which 23 wereconfirmed using reciprocal immunoprecipitation. Func-tional classification shows that proteins with diversefunctions are modified by O-GlcNAc, implying that O-GlcNAc might be involved in the regulation of multiplecellular pathways.
The modification of nuclear and cytoplasmic proteins at serineand threonine residues with O-linked N-acetylglucosamine (O-GlcNAc) was first described over two decades ago.1 The modifica-tion was found in various classes of proteins including enzymes,transcription factors, cytoskeletal proteins, signaling proteins,receptors, nuclear pore complex proteins, and kinases.2 Similarto phosphorylation, the O-GlcNAc modification is dynamic with aturnover rate faster than that of the proteins it modifies.3 TheO-GlcNAc modification has been shown to affect protein-proteininteractions, protein-DNA interactions, protein stability andactivity, and cell signaling cascades.4 Disregulation of the O-GlcNAc modification has been implicated in the development ofdisease states including diabetes, cancer, and Alzheimers.2,4 Giventhe potentially broad regulatory influence of the O-GlcNAcmodification, a more comprehensive understanding of the targetsof O-GlcNAc transferase is needed to elucidate its functionalconsequences.
In this report, we describe the global detection and proteomicanalysis of O-GlcNAc-modified proteins in HeLa cells. An affinity-tagged version of the O-GlcNAc modification is metabolicallyincorporated onto proteins using an azide-tagged analogue ofN-acetylglucosamine. The azido-GlcNAc-modified proteins thuscontain an azide handle for chemoselective conjugation using abiotinylated phosphine reagent. The resulting conjugates wereaffinity-purified with streptavidin beads and subsequently digestedwith trypsin and analyzed by nano-HPLC-MS/MS. Using thisstrategy, we identified 199 azido-GlcNAc-modified proteins in HeLacells. We subsequently validated the presence of this modificationamong 10 previously reported and 13 newly identified O-GlcNAc-modified proteins using specific antibodies. Our results reveal thatproteins with a wide range of functions are modified by O-GlcNAc,implying its diverse cellular functions.
EXPERIMENTAL PROCEDURESMaterials. DMEM and glucose-free DMEM were purchased
from Life Technologies (Gaithesburg, MD). Bovine serum albu-min, trichloroacetic acid (TCA), sodium dodecyl sulfate (SDS),NP40, DMSO, and glucosamine were from Sigma (St. Louis, MO).Streptavidin agarose beads and D-biotin were from Pierce Bio-technology (Rockford, IL). Biotinylated phosphine capture reagent3 (Figure 1) and peracetylated N-(2-azidoacetyl)glucosamine 1(Figure 1) were synthesized in-house. Primary antibodies andProtein A/G agarose beads were from Santa Cruz BiotechnologyInc. (Santa Cruz, CA). Western Lighting plus chemiluminescencedetection kit was from Perkin-Elmer Life Sciences (Boston, MA).Bradford protein estimation reagent and Bio-gel P6 DG desaltingcolumn were from Bio-Rad (Hercules, CA). Streptavidin HRP wasfrom Amersham (Piscataway, NJ). Protease inhibitor cocktail wasfrom Calbiochem (San Diego, CA).
Methods. Cell Culture and Metabolic Labeling. Hela cells werecultured in DMEM (4.5 g of glucose/L) supplemented with 10%FBS and antibiotics at 37 C with 5% CO2. For labeling, culturemedium was replaced at 70% cell confluency with DMEM (1 g ofglucose/L) containing 250 M peracetylated GlcNAc or peracety-lated azido-GlcNAc in DMSO. The cells were labeled for 24 h.Three hours before harvest, glucosamine was added to thecultures to a final concentration of 4 mM as an inhibitor ofO-GlcNAcase during the harvest procedure.
Isolation of Nucleocytoplasmic Proteins. Fifty dishes (15 cm) oflabeled cells were harvested by scraping in chilled PBS containing4 mM glucosamine. The cells were collected by centrifugation at
* Corresponding author. E-mail: firstname.lastname@example.org. Fax: (214) 648-2797. Tel: (214) 648-7947.
Department of Biochemistry. Department of Pharmacology.
(1) Torres, C. R.; Hart, G. W. J. Biol. Chem. 1984, 259, 3308-3317.(2) Wells, L.; Vosseller, K.; Hart, G. W. Science 2001, 291, 2376-2378.(3) Comer, F. I.; Hart, G. W. J. Biol. Chem. 2000, 275, 29179-29182.(4) Vosseller, K.; Wells, L.; Hart, G. W. Biochimie 2001, 83, 575-581.
Anal. Chem. 2006, 78, 452-458
452 Analytical Chemistry, Vol. 78, No. 2, January 15, 2006 10.1021/ac051207j CCC: $33.50 2006 American Chemical SocietyPublished on Web 12/15/2005
1000g. Four milliliters of hypotonic lysis buffer (10 mM HEPES,10 mM KCl, 1.5 mM MgCl2, and protease inhibitor cocktail) wasadded to the pellet and the resultant mixture incubated on ice for30 min. The sample was Dounce homogenized using a B-ratedpestle, and NaCl was added drop by drop to a final concentrationof 400 mM to lyse nuclei as previously described.5 The samplewas ultracentrifuged at 100000g for 1 h at 4 C. The supernatant(nucleocytosolic fraction) was carefully removed, and the proteinwas estimated by the Bradford method.
Conjugation with Biotinylated Capture Reagent and AffinityPurification. The nucleocytosolic protein extract was precipitatedwith eight volumes of ice-cold acetone and one volume of TCAfor 2 h at -20 C. The protein pellet was obtained by centrifugationat 20000g, washed with ice-cold acetone, and resuspended in PBScontaining 2% SDS at a concentration of 4-8 g/L. For conjuga-tion, biotinylated phosphine capture reagent was added to a finalconcentration of 50 M. The samples were kept agitating for 10-12 h in the dark at room temperature. Unconjugated capturereagent was removed using a Bio-gel P6 DG desalting column.The samples were diluted with PBS to a final SDS concentrationof 0.2% and mixed with streptavidin beads (for protein affinitypurification and subsequent identification) or avidin monomerbeads (for verification of azido-GlcNAc modification by Westernblot analysis) for 1 h at room temperature. For protein identifica-tion, the streptavidin beads were washed with PBS containing 2%SDS three times followed by 8 M urea three times and finallywith 1 M KCl three times. The beads were then washed with 50mM NH4HCO3 (pH 8.0) and digested overnight at 37 C with 0.5g of trypsin. The peptides were collected, and the beads werewashed with buffer (ACN/HOAc/water, 40:1:59, v/v/v). Theeluates were pooled and dried in a Speed-Vac for protein
identification. For Western blotting analysis, avidin monomerbeads were collected by centrifugation and washed with PBScontaining 0.5% NP40 five times. The beads were then boiled in1 SDS sample buffer prior to SDS-PAGE.
Protein Identification by Nano-HPLC-MS/MS. Tryptic peptidesobtained above were cleaned with ZipTip C18 (Millipore, Bedford,MA) prior to nano-HPLC/tandem mass spectrometry analysis.Nano-HPLC/tandem mass spectrometry analysis was performedin an LCQ DECA XP ion trap mass spectrometer (ThermoFinni-gan, San Jose, CA) equipped with a nano-ESI source (Ther-moFinnigan). The electrospray source was coupled online withan Agilent 1100 series nano flow HPLC system (Agilent, Palo Alto,CA). A 2-L aliquot of the peptide solution in buffer A (2%acetonitrile/ 97.9% water/0.1% acetic acid (v/v/v)) was manuallyloaded into a capillary HPLC column (50 mm length 75 mi.d., 5-m particle size, 300- pore diameter) packed in-house withLuna C18 resin (Phenomenex, St. Torrance, CA). The peptideswere eluted from the column with a gradient of 5-80% buffer B(90% acetonitrile/9.9% water/0.1% acetic acid (v/v/v)) in buffer Aover 30 min. The eluted peptides were electrosprayed directly intothe LCQ mass spectrometer. The MS/MS spectra were acquiredin a data-dependent mode that determined the masses of theparent ions and fragments of the three strongest ions.
Protein Sequence Database Search. Tandem mass spectra weresearched against NCBI-nr database with MASCOT search engine(Matrix Science, London, U.K.). Enzyme was specified as trypsinwith one or two missing cleavages. Mass error for parent ion masswas set as (4 Da and for fragment ion as (0.5 Da. Spectra with+1, +2, and +3 charge states were considered. If more than onespectrum were assigned to one peptide, each spectrum was givena Mascot score and only the spectrum with the highest score wasused for fragmentation analysis. Peptides identified with a Mascotscore higher than 30 were considered as potential positive
(5) Comer, F. I.; Vosseller, K.; Wells, L.; Accavitti, M. A.; Hart, G. W. Anal.Biochem. 2001, 293, 169-177.
Figure 1. Schematic representation of the TAS technology. (A) Metabolic incorporation of O-GlcNAc into proteins. PeracetylatedN-(2-azidoacetyl)glucosamine 1 is converted in cells to UDP-azido-GlcNAc 2, which is used by O-GlcNAc transferase for O-GlcNAc modificationof target proteins. Protein is represented by the ribbon structure. (B) Conjugation reaction between azido-GlcNAc-modified protein and biotinylatedphosphine capture reagent 3 for subsequent detection and isolation.
Analytical Chemistry, Vol. 78, No. 2, January 15, 2006 453
identifications, and each of them was manually verified by themethod specified as following.
Manual Verification of Protein Identification. Strict manualanalysis was applied to validate protein identification results usinga procedure previously described.6 The following criteria wereused for manual verification. y, b, and a ions as well as their waterloss or amine loss peaks are considered. All the major isotope-resolved peaks should match fragment masses of the identifiedpeptide. The isotope-resolved peaks were emphasized because asingle peak could come from an electronic spark and are less likelyto be relevant to peptide fragments. The major isotope-resolvedpeaks are defined as (1) those isotope-resolved daughter ions withm/z higher than parent m/z and intensity higher than 5% of themaximum intensity or (2) those isotopically resolved peaks withintensities higher than 20% of the maximum intensity and m/zvalues between one-third of the parent m/z and the parent m/z.Typically >7 isotope-resolved peaks were matched to theoreticalmasses of the peptide fragments.
Identified proteins were functionally classified based on theirannotation in the NCBI protein database (www.ncbi.nlm.nih.gov)and NIAIDs DAVID.7,8
Verification of Azido-GlcNAc Modification. Cells were harvestedin PBS containing 0.5% NP40. The cell lysate was centrifuged at20000g for 1 h at 4 C, and the supernatant was carefully removed.For immunoprecipitation, 5 mg of whole cell lysate was mixedwith 4 g of specific antibody and 20 L of protein A/G agarosebeads for 2 h at room temperature. The beads were collected bycentrifugation and washed with PBS containing 0.5% NP40 fivetimes. The proteins were eluted by boiling for 5 min in 2% SDScontaining PBS. The supernatant was removed and then mixedwith biotinylated capture reagent 3 (final concentration 200 M).The conjugation reaction was carried out at room temperaturefor 12 h. The unconjugated capture reagent was removed using aBio-gel P6 DG desalting column. SDS sample buffer was addedto the eluted biotin-conjugated protein sample and analyzed byWestern blot using streptavidin-HRP. Alternatively, protein O-GlcNAc modification was confirmed by avidin/Western blottinganalysis. In this experiment, the azido-O-GlcNAc-modified proteinswere conjugated using the same method as described above andisolated by avidin-conjugated agarose beads. The isolated proteinswere eluted by boiling in SDS sample buffer and analyzed byWestern blotting analysis using an antibody of interest.
RESULTSSelective Metabolic Labeling of O-GlcNAc-Modified Pro-
teins. The effective proteomic analysis of proteins bearing specificposttranslational modifications requires selective enrichment ofthe proteins of interest from a complex protein mixture. Withrespect to the O-GlcNAc modification, the TAS strategy involvesmetabolic labeling of cells with an azide-derivatized analogue ofperacetylated N-acetylglucosamine 1 (Figure 1). This compoundis modified by the cells endogenous metabolic machinery intoUDP-azido-GlcNAc 2, which is appended onto proteins in place
of the O-GlcNAc modification. Incorporation of the azide-taggedanalogue of O-GlcNAc onto proteins would then allow the selectiveisolation, detection, and characterization of O-GlcNAc-modifiedproteins via the enhanced Staudinger ligation between the azideand a phosphine probe 3 engineered with an affinity tag, such asbiotin.9 In this study, we applied the previously characterized TAStechnology toward the proteomic analysis of O-GlcNAc-modifiedproteins in HeLa cells. This strategy takes advantage of thepromiscuity of metabolic enzymes in tolerating small perturbationsin the structure of the modification substrate.9-11
Detection of Azido-GlcNAc Modification in Nucleocyto-plasmic Proteins. Metabolic incorporation of an azide-taggedGlcNAc molecule onto O-GlcNAc-modified proteins would allowthe efficient detection, isolation, and characterization of O-GlcNAc-modified proteins. To demonstrate our ability to label and detectO-GlcNAc-modified proteins, HeLa cells were grown to 70%confluency and labeled for 24 h with either peracetylated GlcNAcor peracetylated azido-GlcNAc. Nucleocytoplasmic proteins wereextracted from the labeled cells and conjugated with biotinylatedphosphine capture reagent 3. As shown in Figure 2, O-GlcNAc-modified proteins can be detected in samples prepared from azido-GlcNAc-labeled HeLa cells, only after conjugation with 3 (lane 3,Figure 2). The Western blotting signal can be competitivelyinhibited by performing the capture reaction in the presence of0.1 mM concentration of an exogenous azide-containing substrate,azido-farnesyl diphosphate (FPP-N3) (lane 2). The signal can alsobe competitively inhibited by probing the nitrocellulose membranein the presence of 0.1 mM D-biotin (lanes 4-6). In addition, there
(6) Chen, Y.; Kwon, S. W.; Kim, S. C.; Zhao, Y. J. Proteome Res. 2005, 4, 998-1005.
(7) Dennis, G., Jr.; Sherman, B. T.; Hosack, D. A.; Yang, J.; Gao, W.; Lane, H.C.; Lempicki, R. A. Genome Biol. 2003, 4, P3.
(8) Hosack, D. A.; Dennis, G., Jr.; Sherman, B. T.;...