Quantitative time-resolved chemoproteomics revealsthat stable O-GlcNAc regulates box C/DsnoRNP biogenesisWei Qina,b, Pinou Lva,b, Xinqi Fana, Baiyi Quana, Yuntao Zhua, Ke Qina, Ying Chena, Chu Wanga,b,c,d,e,1,and Xing Chena,b,c,d,e,1

aCollege of Chemistry and Molecular Engineering, Peking University, Beijing 100871, China; bPeking–Tsinghua Center for Life Sciences, Peking University,Beijing 100871, China; cBeijing National Laboratory for Molecular Sciences, Peking University, Beijing 100871, China; dSynthetic and FunctionalBiomolecules Center, Peking University, Beijing 100871, China; and eKey Laboratory of Bioorganic Chemistry and Molecular Engineering of the Ministry ofEducation, Peking University, Beijing 100871, China

Edited by Carolyn R. Bertozzi, Stanford University, Stanford, CA, and approved June 29, 2017 (received for review February 16, 2017)

O-linked GlcNAcylation (O-GlcNAcylation), a ubiquitous posttransla-tional modification on intracellular proteins, is dynamically regu-lated in cells. To analyze the turnover dynamics of O-GlcNAcylatedproteins, we developed a quantitative time-resolvedO-linked GlcNAcproteomics (qTOP) strategy based on metabolic pulse-chase labelingwith an O-GlcNAc chemical reporter and stable isotope labeling withamino acids in cell culture (SILAC). Applying qTOP, we quantified theturnover rates of 533 O-GlcNAcylated proteins in NIH 3T3 cells anddiscovered that about 14% exhibited minimal removal of O-GlcNAcor degradation of protein backbones. The stability of those hyperstableO-GlcNAcylated proteins was more sensitive to O-GlcNAcylation in-hibition compared with the more dynamic populations. Among thehyperstable population were three core proteins of box C/D smallnucleolar ribonucleoprotein complexes (snoRNPs): fibrillarin (FBL),nucleolar protein 5A (NOP56), and nucleolar protein 5 (NOP58). Weshowed that O-GlcNAcylation stabilized these proteins and was es-sential for snoRNP assembly. Blocking O-GlcNAcylation on FBL al-tered the 2′-O-methylation of rRNAs and impaired cancer cellproliferation and tumor formation in vivo.

O-GlcNAcylation | metabolic labeling | proteomics | protein stability |snoRNP

In mammalian cells, a large number of nuclear, cytoplasmic,and mitochondrial proteins are posttranslationally or cotransla-

tionally modified with a GlcNAc monosaccharide β-linked to serineor threonine residues, which is termed as O-linked GlcNAcylation(O-GlcNAcylation) (1, 2). The O-GlcNAc modification has beenimplicated in a whole range of cellular processes, including tran-scriptional regulation (3), signal transduction (4), and stress re-sponse (5). At the protein level, O-GlcNAcylation was found toregulate protein stability (6–10) and activity (11–14). Remarkably, asole known enzyme O-linked GlcNAc transferase (OGT) catalyzesall cellular O-GlcNAc modifications (15, 16). Another enzyme,O-linked GlcNAcase (OGA), can remove O-GlcNAc from modi-fied proteins, making O-GlcNAcylation a reversible modification(17). To date, the number of candidateO-GlcNAcylated proteins inmammalian cells has accumulated to more than 1,000, mainly ow-ing to recent MS-based proteomic profiling in various cell lines (18–24). Such a broad scope of protein substrates raises several inter-esting questions about O-GlcNAcylation. Are all of the proteinO-GlcNAcylation events reversible in cells? Do O-GlcNAcylatedproteins turn over at different rates? Is the stability of O-GlcNAy-lated proteins differentially affected by the modification?To address these important questions, we sought to develop a

quantitative time-resolved method for globally profiling the turn-over dynamics ofO-GlcNAcylated proteins in living cells. The MS-based O-GlcNAcome profiling has been largely facilitated byemerging methods for selective enrichment of O-GlcNAcylatedproteins, such as affinity purification using O-GlcNAc–recognizingantibodies (19, 25) or lectins (20, 26), and chemoenzymatic (21, 27)

or metabolic labeling of O-GlcNAc with bioorthogonal chemicalreporters followed by click labeling with affinity tags (23, 28–31). Incontrast to most of the enrichment methods that are performed invitro, metabolic incorporation of O-GlcNAc reporters enableschemical labeling of O-GlcNAc in living cells. Moreover, metaboliclabeling is compatible with pulse-chase experiments (32–35), andtherefore, chemical reporters of O-GlcNAc may potentially beexploited to measure the O-GlcNAcylation dynamics and turnoverrates of O-GlcNAcylated proteins. However, proteomic studiesbased on O-GlcNAc chemical reporters have so far been per-formed using spectral counting, a semiquantitative method.Herein, we describe a quantitative time-resolvedO-linked GlcNAc

proteomics (qTOP) strategy for global analysis of the turnover dy-namics of O-GlcNAcylated proteins in living cells. The qTOP ap-proach combines the strengths of two protocols: (i) the metabolicpulse-chase labeling with an O-GlcNAc chemical reporter in livingcells, which enables time-resolved enrichment of O-GlcNAcylatedproteins, and (ii) the stable isotope labeling with amino acids in cellculture (SILAC), which permits quantitative analysis of the enrichedO-GlcNAc proteomes. Applying qTOP, we successfully quanti-fied the degradation rates of 533 O-GlcNAcylated proteins inNIH 3T3 cells, which were categorized into three subgroups:“hyperstable,” “dynamic,” and “hyperdynamic,” The hyperstablepopulation accounted for an unexpectedly high portion (∼14%) of

Significance

In mammalian cells, more than 1,000 intracellular proteins areposttranslationally modified with O-linked GlcNAc (O-GlcNAc),which regulates many important biological processes. TheO-GlcNAc modification has been found to dynamically cycle on andoff the modified proteins. How O-GlcNAc affects protein stabilityremains to be investigated at the proteome level. In this work, wedeveloped a quantitative time-resolved proteomic strategy to an-alyze the turnover dynamics of O-GlcNAcylated proteins. Wediscovered that not all protein O-GlcNAcylation events were re-versible and that a subset of O-GlcNAcylated proteins exhibitedminimal removal ofO-GlcNAc or degradation of protein backbones.Our work reveals stable O-GlcNAc as an important regulatorymechanism for stabilizing proteins, such as core proteins of box C/Dsmall nucleolar ribonucleoprotein complexes.

Author contributions: W.Q., C.W., and X.C. designed research; W.Q., P.L., X.F., B.Q., Y.Z.,K.Q., and Y.C. performed research; W.Q., C.W., and X.C. analyzed data; and W.Q., C.W.,and X.C. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence may be addressed. Email: chuwang@pku.edu.cn or xingchen@pku.edu.cn.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1702688114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1702688114 PNAS | Published online July 31, 2017 | E6749–E6758

CHEM

ISTR

YBIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

14, 2

021

the O-GlcNAcyated proteome that we quantified. More impor-tantly, those hyperstable O-GlcNAc modifications had more pro-found effects on stabilizing the modified proteins, among whichwere three core proteins of box C/D small nucleolar ribonucleo-protein complexes (snoRNPs)—fibrillarin (FBL), nucleolar pro-tein 5A (NOP56), and nucleolar protein 5 (NOP58).Box C/D snoRNPs consist of a box C/D small nucleolar RNA

(snoRNA) and a set of four core proteins—FBL, NOP56, NOP58,and nonhistone chromosome protein 2-like 1 (NHP2L1; SNU13,15.5K). They play an essential role in ribosome assembly bymethylating rRNAs at the 2′-O-ribose with FBL as the methyl-transferase (36). In this work, qTOP revealed that FBL, NOP56,and NOP58 were modified with stable O-GlcNAc. We furthercharacterized the functional consequences of O-GlcNAcylationfor box C/D snoRNPs and showed that it played an important rolein snoRNP assembly and biogenesis via stabilizing the modifiedcore proteins and regulating their subnuclear localization and in-teractions. Furthermore, we identified a major O-GlcNAcylationsite on FBL and found that O-GlcNAcylation of FBL regulatedrRNA methylation patterns and promoted tumorigenesis.

ResultsSILAC-Based Quantification Improves O-GlcNAcylated ProteomeProfiling. To establish the qTOP platform, we first implementedSILAC into the chemical reporter-based O-GlcNAc proteomicidentification. Several GlcNAc and GalNAc analogs containingan azide or alkyne have been developed as O-GlcNAc chemicalreporters, which showed varied efficiencies and specificitiesfor O-GlcNAc labeling (23, 24, 28–31) (Fig. S1). We chose per-acetylated 6-azido-6-deoxy-GlcNAc (Ac36AzGlcNAc) for thiswork, because it was shown to exclusively label O-GlcNAc withoutentering other glycosylation pathways and enabled identification of366 O-GlcNAcylated proteins in NIH 3T3 cells by spectralcounting (23). NIH 3T3 cells grown in culture medium containingisotopically labeled “heavy” lysine and arginine or standard “light”medium were treated with 200 μM Ac36AzGlcNAc for 36 h (Fig.S2A). In-gel fluorescence scanning of the light and heavy cell ly-sates reacted with alkyne-Cy5 via Cu(I)-catalyzed azide-alkynecycloaddition (i.e., click chemistry) exhibited identical labelingpatterns and intensities, indicating that metabolic incorporation of6-azido-6-deoxy-GlcNAc (6AzGlcNAc) does not differentiate be-tween light and heavy cells (Fig. S2B). Furthermore, 6AzGlcNAclabeling was OGT-dependent (Fig. S2C).Because SILAC-based quantification has been previously used

to improve the accuracy and sensitivity of palmitoylation pro-teomics using an alkyne-containing palmitic acid analog (33), wesought similar improvements for large-scale identification ofO-GlcNAcylated proteins. We first tested the accuracy of SILACquantification. The heavy and light isotope-labeled cell lysateswere combined at a 1:1 ratio followed by reaction with alkyne-biotin, streptavidin enrichment, and on-bead trypsin digestion.The resulting peptides were analyzed by liquid chromatography(LC)-MS/MS using the multidimensional protein identificationtechnology (37) protocol (Fig. S2A). Peptides were identified byProLuCID (38), and their light/heavy SILAC ratios were quan-tified using CIMAGE (39). The protein ratio was then quantifiedas the median ratio across all peptides assigned to a specificprotein. The distribution of protein ratios had a mean of 0.99,accurately matching the dilution factor value (Fig. S2D). Wethen applied the SILAC-based quantitative chemoproteomics toidentify O-GlcNAcylated proteins by performing the labeling ex-periments with Ac36AzGlcNAc as the chemical reporter (Fig. 1Aand Fig. S2E). In the standard “forward” SILAC experiment, thelight cells were treated with Ac36AzGlcNAc as a reporter, and theheavy cells were treated with peracetylated GlcNAc (Ac4GlcNAc)as a control. We also performed a replicate “reverse” SILACexperiment, where the labeling order was switched (i.e., heavy withthe reporter and light as the control) to increase the robustness and

accuracy for quantitation. The 6AzGlcNAc-incorporated proteinswere enriched and quantified against the control, which identified1,139 and 1,451 O-GlcNAcylated proteins in the forward and re-verse labeling experiments, respectively (Fig. S2F); 896 proteinswere reciprocally enriched and quantified in both experiments(Fig. 1B and Dataset S1). Reanalyzing the same datasets with thespectral counting method, we could identify 699 O-GlcNAcylatedproteins, of which 634 overlapped with the SILAC quantification(Fig. 1C). Notably, by implementing the SILAC quantificationmethod, the Ac36AzGlcNAc-based chemoproteomics identifiedan additional 262 O-GlcNAcylated proteins with low spectralcounts in NIH 3T3 cells, such as AHNAK, GAK, and UBR4(Fig. 1D). These results show that SILAC quantification cansignificantly improve the accuracy and sensitivity of the chemicalreporter-based O-GlcNAc proteomics.Excluding those with uncharacterized cellular localizations, all

of the O-GlcNAcylated proteins identified by Ac36AzGlcNAclabeling possess intracellular domains accessible by OGT, in agree-ment with the high specificity of Ac36AzGlcNAc for O-GlcNAclabeling (Fig. 1E). We further validated the identified O-GlcNA-cylated proteins using a chemoenzymatic method based on a mutantgalactosyltransferase (Y289L GalT), which recognizes terminalGlcNAc moieties and allows for labeling of endogenous andnatural O-GlcNAc with uridine diphosphate N-azidoacetylga-lactosamine (UDP-GalNAz) in cell lysates (40, 41). By pre-forming the forward and reverse Y289L GalT labeling followedby click reaction with alkyne-biotin, streptavidin enrichment,and LC-MS/MS analysis, a total of 1,272 O-GlcNAcylated pro-teins were identified (Fig. S2G); 754 proteins were identified in bothAc36AzGlcNAc and Y289L GalT labeling, which were classified ashigh-confidence O-GlcNAcylated proteins (Fig. 1F).

Quantitative Analysis of the Dynamics of O-GlcNAcylated Proteins byqTOP. Having established that Ac36AzGlcNAc in combinationwith SILAC could quantify more than 750 O-GlcNAcylatedproteins in NIH 3T3 cells that could also be verified by Y289LGalT labeling, we next integrated the pulse-chase protocol toconstruct the qTOP platform for profiling the turnover dynamicsof O-GlcNAc proteome (Fig. 2A and Fig. S3A). Light and heavycells were pulse labeled with Ac36AzGlcNAc for 36 h to ensuresufficient metabolic incorporation of the chemical reporter (Fig.S3B). In a forward qTOP experiment, the heavy cells were im-mediately harvested after pulse labeling to serve as the referenceat time of 0 h, and the light cells were chased with Ac4GlcNAcfor another 12 h (Fig. 2A). This chase time was optimized toensure that significant O-GlcNAc turnover events can be ob-served, and there are still enough chemical reporters remainingfor detection (Fig. S3C). Similarly, a reverse replicate of qTOPwas also performed, in which the pulse-labeled heavy cells werechased with Ac4GlcNAc (Fig. S3A). After mixing and lysing thecells, 6AzGlcNAc-labeled proteins were enriched, analyzed byLC-MS/MS, and quantified between the chased and referencesamples. The residual amount of the pulse-labeled fraction ofindividual proteins after 12 h was revealed by the qTOP ratio(chased vs. reference), with smaller ratios corresponding tofaster turnover rates of O-GlcNAcylated proteins. By isolating6AzGlcNAc-incorporated proteins, qTOP quantified exclusivelythe O-GlcNAcylated proteoform, avoiding interferences im-posed by the unmodified fraction. Moreover, the turnover ofO-GlcNAcylated proteins measured in qTOP could result fromprotein degradation and/or OGA-catalyzed removal of 6AzGlc-NAc from the pulse-labeled proteins (Fig. S3D).In three biological replicates, we acquired qTOP ratios for

533 of 754 high-confidence O-GlcNAcylated proteins (Fig. S3Eand Dataset S2), which were normally distributed at a mean of0.74 with an SD of 0.17 (Fig. 2B). The proteins were classifiedinto three groups by ranking their qTOP ratios: hyperstableO-GlcNAcylated proteins with ratios > 0.91 (i.e., mean + SD),

E6750 | www.pnas.org/cgi/doi/10.1073/pnas.1702688114 Qin et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

14, 2

021

dynamic O-GlcNAcylated proteins with ratios ≤ 0.91 and ≥ 0.57(i.e., mean – SD), and hyperdynamic O-GlcNAcylated proteinswith ratios < 0.57 (Fig. 2B and Dataset S2). Consistent with thegeneral view that O-GlcNAc modifications are often rapidly re-versible and dynamic (42–45), 86% of the O-GlcNAcylatedproteins quantified in the qTOP experiments showed a dynamicturnover in 12 h. Meanwhile, it is also intriguing to observe thatquite a significant portion (∼14%) of the O-GlcNAcylated pro-teins (i.e., 75 proteins in the hyperstable group) showed not onlyminimal degradation of the protein backbone but also, minimalremoval of O-GlcNAc within 12 h, reflecting their stable naturewith slow turnover dynamics.

Stable O-GlcNAc Modification Has a Profound Effect on StabilizingProteins. Given that a number of proteins have been found tobe stabilized by O-GlcNAc modification, such as SP1, p53,BMAL1/CLOCK, and nuclear pore complex protein 62 (NUP62)(2, 7–9), we asked whether the hyperstable O-GlcNAc revealedby qTOP contributes to the stability of those O-GlcNAcylatedproteins. To answer this question, we used an OGT inhibitor, per-acetylated 2-acetamido-2-deoxy-5-thio-D-glucopyranose (Ac45SGlc-NAc) (46), to globally lower the cellular level of O-GlcNAcylationand then measured changes of protein abundance by SILAC-basedquantitative proteomic analysis (Fig. 3A). Heavy NIH 3T3 cellswere treated with 50 μM Ac45SGlcNAc for 48 h, and light cells

were treated with the DMSO control. After mixing and lysing thecells, 2,500 proteins were quantified with high confidence (i.e., atleast twice in three biological replicates), and the heavy to lightratios indicated the changes of protein abundance on O-GlcNA-cylation inhibition (Fig. 3B, Dataset S3, and Fig. S4A).The list of 2,500 quantified proteins covered 62 of 75 hyperstable

O-GlcNAcylated proteins, 295 of 378 dynamic O-GlcNAcylatedproteins, and 43 of 80 hyperdynamicO-GlcNAcylated proteins (Fig.3C). On OGT inhibition, >80% of the hyperstableO-GlcNAcylatedproteins exhibited SILAC ratios smaller than 1.0, indicating thattheir abundance was decreased when O-GlcNAcylation was inhibi-ted, whereas the percentages were much lower (57 and 60%) in thedynamic and hyperdynamic populations, respectively (Fig. 3D).Furthermore, the extent of decrease was significantly greater for thehyperstable O-GlcNAcylated proteins than the dynamic and hyper-dynamic populations (the mean ratio of 0.89 for the hyperstablegroup vs. 0.98 for the dynamic group and 0.99 for the highly dynamicgroup). Collectively, these results suggest that stable O-GlcNAc hasmore profound effects on stabilizing proteins.

Three Box C/D snoRNP Core Proteins Are Modified with Stable O-GlcNAc.The 75 hyperstable O-GlcNAcylated proteins are mainly associatedwith three biological processes, including transport, metabolicprocess, and RNA regulation (Fig. 2C). Among the list of transport-associated proteins, nucleoporin NUP214, a nuclear pore complex

Fig. 1. Quantitative identification of O-GlcNAcylated proteins by SILAC-based chemoproteomics. (A) Schematic of the workflow of SILAC-based quantitativeO-GlcNAcylation proteomics using metabolic labeling of chemical reporters. In the forward SILAC, light and heavy cells were treated with 200 μMAc36AzGlcNAc and Ac4GlcNAc for 36 h, respectively. The lysates were mixed at a 1:1 ratio, reacted with alkyne-biotin for enrichment, digested by trypsin, andanalyzed by LC/LC-MS/MS. The ratio of each identified peptide was quantified by comparing areas under the light and heavy chromatographic peaks, and theratio for each protein was calculated as the median of all of the relevant peptide ratios. In the reverse SILAC, light cells were treated with Ac4GlcNAc, andheavy cells were labeled with Ac36AzGlcNAc the scheme is shown in Fig. S2E. (B) Overlap of the identified O-GlcNAcylated proteins (enrichment ratios ≥ 2)between the forward and reverse SILAC experiments; 896 proteins were enriched in both experiments. (C) Overlap of the O-GlcNAcylated proteins assignedby SILAC and spectral counting. (D) MS1 chromatographic peaks of the representative peptides from AHNAK, cyclin G-associated kinase (GAK), andE3 ubiquitin-protein ligase UBR4, which are newly identified as O-GlcNAcylated proteins by SILAC-based chemoproteomics. The light and heavy chro-matographic traces are colored in red and blue, respectively. Green solid lines delineate the chromatographic peak boundary for quantification. Black as-terisks indicate the MS/MS event that supports the identification of the corresponding peptide. (E) Cellular localization analysis of 896 6AzGlcNAc-labeledproteins. Intracellular proteins include those with defined locations in the cytoplasm, nucleus, or mitochondria. Extracellular/luminal proteins include thosewith defined locations exclusively outside the plasma membrane or within the lumen. Dual proteins are those with both intracellular and extracellular/luminaldomains. Unassigned proteins are those with no defined information on their subcellular locations. (F) Overlap of the O-GlcNAcylated proteins identified byAc36AzGlcNAc and Y289L GalT labeling. The proteins identified by the two labeling methods were classified as high-confidence O-GlcNAcylated proteins.

Qin et al. PNAS | Published online July 31, 2017 | E6751

CHEM

ISTR

YBIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

14, 2

021

component protein, exhibited no detectable protein degradationor removal of O-GlcNAc in the qTOP experiments (Fig. S3F).Consistent with our hypothesis that stable O-GlcNAc stabilizesproteins, NUP214 was indeed reported to be stabilized by O-GlcNAcylation previously (10). When we looked into the stableO-GlcNAcylated proteins that are associated with RNA regulation,we noticed that three of four core protein components of box C/DsnoRNPs were among the list, which are FBL, NOP56, andNOP58 (Fig. 2D). Given the functional importance of these pro-teins in regulating snoRNP assembly and biogenesis, we continuedto investigate the impact of O-GlcNAcylation on their proteinstability as well as functional consequences.FBL, NOP56, and NOP58 all showed selective enrichment in

SILAC-based identification of O-GlcNAcylated proteins in Ac36AzGlc-NAc-labeled cells (Fig. 4A). To further validate that they are bona

fide O-GlcNAcylated proteins, each of three proteins wasrecombinantly expressed with a C-terminal Flag-His tandem tag inHeLa cells treated with Ac4GalNAz and immunoprecipitated tomeasure its labeling by the chemical reporter. We chose Ac4Gal-NAz here as the chemical reporter, because it enters the GalNAcsalvage pathway to form UDP-GalNAz, which is interconvertedwith UDP-N-azidoacetylglucosamine (UDP-GlcNAz), and providesa much better O-GlcNAc labeling efficiency (24, 29). Despite thatAc4GalNAz also labels cell surface glycans and therefore, compli-cates the global proteomic profiling, it would not interference withbiochemical studies on individual immunoprecipitated O-GlcNA-cylated proteins. The cell lysates were conjugated with alkyne-biotinand immunoprecipitated with an anti-His antibody. Western blot-ting with an antibiotin antibody showed that FBL, NOP56, andNOP58 were all modified with GlcNAz (Fig. 4B). Furthermore,

Fig. 2. Turnover dynamics ofO-GlcNAcylated proteins revealed by qTOP. (A) Schematic of the qTOP workflow. The light and heavy cells were pulse labeled with200 μMAc36AzGlcNAc for 36 h. In the forward qTOP experiment, the heavy cells are harvested immediately, and the light cells are chased with 200 μMAc4GlcNAcfor another 12 h. The cells were then mixed, lysed, and conjugated with alkyne-biotin for enrichment and quantitative proteomic analysis. The workflow of thereverse qTOP experiment, in which the heavy cells are chased instead, is shown in Fig. S3A. The SILAC ratio for each given protein (chased vs. reference) is definedas its qTOP ratio. (B) Histogram of the qTOP ratios for 533 high-confidence O-GlcNAcylated proteins quantified in three replicates of qTOP experiments (mean =0.74 and SD = 0.17). Using means ± SD as cutoffs, these proteins were classified as hyperstable, dynamic, and hyperdynamic. (C) Gene ontology analysis of75 hyperstable O-GlcNAcylated proteins regarding the biological processes in which they are involved. (D) qTOP ratios of FBL, NOP56, and NOP58. For eachprotein, the sequence of one representative peptide and its associated MS1 chromatographic peaks are shown in Upper, and its qTOP ratio quantified from threereplicates of qTOP experiments (mean ± SD) is shown in Lower. N.S., not significant (one sample t test against 1.0, the value indicating no degradation).

E6752 | www.pnas.org/cgi/doi/10.1073/pnas.1702688114 Qin et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

14, 2

021

FBL, NOP56, and NOP58 were all identified in the Y289L GalT-based proteomics, confirming their modifications with natureGlcNAc on the endogenous proteins (Fig. S5A).We adapted a previously developed method based on a re-

solvable mass tag to quantify the O-GlcNAcylation stoichiometryon the endogenous FBL, NOP56, and NOP58 (47). Cell lysateswere treated with Y289L GalT and UDP-GalNAz to tagO-GlcNAc with azides, which were subsequently click labeled withan alkyne-functionalized PEG mass tag with the molecular weightof about 2,000 or 5,000 (alkyne-PEG2k or alkyne-PEG5k) and re-solved by SDS/PAGE. Western blotting with antibodies againstFBL, NOP56, or NOP58 exhibited a shift of molecular mass at-tributed to O-GlcNAc moieties chemoenzymatically labeled withthe PEG mass tag (Fig. 4C). The O-GlcNAcylation stoichiome-tries were estimated by quantifying the relative band intensities ofmodified and unmodified proteins. Using the alkyne-PEG2K, thestoichiometry of FBL O-GlcNAcylation was estimated to be 40%.The alkyne-PEG5K resulted in a lower labeling of FBL O-GlcNAc,which was probably caused by higher steric hindrance of the larger

PEG chain (Fig. S5B). Because of the higher molecular weights,NOP56 and NOP58 are better separated in SDS/PAGE with thealkyne-PEG5K tag, giving the estimated O-GlcNAcylation stoichi-ometries as 49 and 45%, respectively (Fig. 4C). Because it is pos-sible that the PEGylation reactions did not proceed to completion,the O-GlcNAcylation stoichiometries could be underestimated tosome extent.Our O-GlcNAc proteomic profiling identified FBL, NOP56, and

NOP58, leaving NHP2L1 as the only box C/D snoRNP proteincomponent that is not O-GlcNAcylated. To confirm this result, wefurther examined the O-GlcNAcylation state of NHP2L1 by usingthe Ac4GalNAz labeling and the PEG mass tag-based assays,both of which exhibited no detectable O-GlcNAc modification onNHP2L1 (Fig. S5C).We next attempted to identify the exact sites of O-GlcNAcy-

lation on these proteins. Flag/His-tagged proteins were purifiedfrom HeLa cells and subjected to analysis by tandem MS withelectron transfer dissociation (ETD) fragmentation to preservethe labile O-GlcNAc glycosidic linkage. The tandem ETD spectrum

Fig. 3. Quantification of the changes of protein abundance on OGT inhibition by SILAC-based quantitative proteomics. (A) Workflow of quantitativeproteomics for cells treated with an OGT inhibitor. Equal numbers of heavy and light cells are treated with 50 μM Ac45SGlcNAc and DMSO, respectively, for48 h. After mixing and lysing the cells, the whole proteomes are digested by trypsin. The peptides were fractionated by high pH reversed-phase liquidchromatography (RPLC) and analyzed by SILAC-based quantitative proteomics. (B) Overlap of the proteins quantified in three replicate experiments. A totalof 2,500 proteins were quantified in at least two of three replicates and selected for analysis of protein abundance change on OGT inhibition. (C) Overlap of2,500 proteins quantified on OGT inhibition with the hyperstable, dynamic, and hyperdynamic O-GlcNAcylated proteins from the qTOP experiments.(D) Abundance changes of O-GlcNAcylated proteins on OGT inhibition. For each of the hyperstable, dynamic, and hyperdynamic populations of O-GlcNAcylatedproteins quantified by qTOP, a boxplot was drawn showing the distribution of protein abundance change on OGT inhibition. The SILAC ratios (OGT inhibitionvs. control) of each protein are shown as gray diamonds, with those of FBL, NOP56, and NOP58 highlighted in blue. Box limits represent 25th percentiles,medians, and 75th percentiles. Whiskers extend to nonoutliers. ***P < 0.001 (one-sided Wilcoxon rank sum test).

Qin et al. PNAS | Published online July 31, 2017 | E6753

CHEM

ISTR

YBIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

14, 2

021

unambiguously identified serine-142 (S142) of FBL as a site ofO-GlcNAcylation (Fig. 4D and Fig. S5D). To further verify theO-GlcNAcylation on S142 of FBL, an S142A mutant of FBL,FBLS142A-Flag/His, was overexpressed in HeLa cells treated withAc4GalNAz. The lysates were reacted with alkyne-biotin andimmunoprecipitated with the anti-His antibody, and Western blot-ting with the antibiotin antibody showed significantly reducedGlcNAz labeling on FBLS142A (Fig. 4E). Moreover, Westernblotting with an O-GlcNAc–recognizing antibody RL2 detecteda strongO-GlcNAylation signal on purified Flag/His-FBL in contrastto Flag/His-FBLS142A, confirming S142 as the major O-GlcNAcmodification site of FBL (Fig. 4F). Our attempts to identifyO-GlcNAc modification sites on NOP56 and NOP58 using ETDMS were not successful.

O-GlcNAcylation Stabilizes Box C/D snoRNP Core Proteins. Next,we assessed whether O-GlcNAcylation regulates the expressionlevels of FBL, NOP56, and NOP58. The overall cellular level ofO-GlcNAc was inhibited by the OGT inhibitor Ac45SGlcNAc. Con-sistent with our proteomic quantification (Fig. 3D and Fig. S4B),the abundance of all of these three proteins, as detected byWesternblot analysis, was significantly diminished on deprivation ofO-GlcNAc in a time-dependent manner (Fig. 5A and Fig. S6 A–C).Quantitative RT-PCR (qRT-PCR) analysis showed that the mRNAlevels of FBL, NOP56, and NOP58 were not lowered on OGTinhibition (Fig. S6 E–G), excluding the possibility of transcriptional

regulation. By knocking down OGT with siRNA to reduceO-GlcNAcylation (Fig. S7A), we also observed that the protein levelsof FBL, NOP56, and NOP58 were greatly reduced with no ap-parent decrease in their mRNA levels (Fig. S7 B–D and F).To test whether the reduced protein level was owing to de-

creased protein stability on O-GlcNAc deprivation, the cellswere treated with cycloheximide (CHX) to stop protein synthe-sis. The immunoblotting of FBL, NOP56, and NOP58 showedthat protein degradation was significantly accelerated by inhib-iting OGT either genetically (Fig. S7G) or pharmacologically(Fig. 5B and Fig. S8 A–C).Interestingly, the protein level of NHP2L1 was also diminished

by OGT inhibition (Fig. 5A and Fig. S6D) and knockdown (Fig.S7F), whereas its mRNA level was not lowered (Figs. S6H and S7E).Correlatively, the degradation of NHP2L1 was accelerated by OGTknockdown (Fig. S7G) and inhibition (Fig. 5B and Fig. S8D). Be-cause NHP2L1 was not O-GlcNAc modified, we suspected that thewhole box C/D snoRNP complex might be destabilized when one ofthe essential components was disrupted. We therefore knockeddown FBL by siRNA, which as expected, resulted in down-regulationof NHP2L1 expression as well as the expression of NOP56 andNOP58 (Fig. S9). Similar results were observed for siRNA knock-down of NOP56 or NOP58. Taken together, these results show thatO-GlcNAc regulates the stability and integrity of box C/D snoRNPby stabilizing FBL, NOP56, and NOP58.

Fig. 4. Core proteins of box C/D snoRNP are O-GlcNAcylated. (A) MS1 chromatographic peaks of representative peptides of FBL, NOP56, and NOP58 fromthe SILAC-based identification of O-GlcNAcylated proteins showing the selective enrichment of these proteins through 6AzGlcNAc labeling. (B) Detection ofO-GlcNAcylation on immunoprecipitated FBL, NOP56, and NOP58. HeLa cells transiently expressing Flag/His-tagged FBL, NOP56, or NOP58 were treated with100 μM Ac4GalNAz or Ac4GlcNAc as a negative control for 36 h. The cells were lysed and reacted with 100 μM alkyne-biotin followed by immunoprecipitation (IP)using an anti-His antibody. With similar amounts of the immunoprecipitated proteins, the antibiotin blotting showed the specific incorporation of the GlcNAzchemical reporter in FBL, NOP56, and NOP58. (C) Detection of O-GlcNAcylation on endogenous FBL, NOP56, and NOP58 using chemoenzymatic labeling with aresolvable mass tag. The solid and dashed arrows indicate untagged proteins and proteins modified with the PEG mass tag, respectively. (D) Identification ofSer142 on FBL as theO-GlcNAcylation site by ETD. Prominent c/z fragment ions used to identify theO-GlcNAcylated peptides from FBL were shown in red, and theraw ETDMS/MS spectrum is shown in Fig. S5D. (E) Verification ofO-GlcNAcylation of Ser142 on immunoprecipitated FBL by chemical reporter. Antibiotin Westernblot on FBLWT-Flag/His and FBLS142A-Flag/His immunoprecipitated from cells treated with Ac4GalNAz or Ac4GlcNAc. Anti-His blot indicated the loading ofthe immunoprecipitated proteins. (F) Verification of O-GlcNAcylation of Ser142 on immunoprecipitated FBL by O-GlcNAc antibody. Western blot detection ofO-GlcNAc by using the RL2 antibody on isolated FBLWT-Flag/His and FBLS142A-Flag/His. Anti-His blot indicated the loading of the immunoprecipitated proteins.

E6754 | www.pnas.org/cgi/doi/10.1073/pnas.1702688114 Qin et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

14, 2

021

Fig. 5. O-GlcNAcylation stabilizes box C/D core proteins and regulates the assembly of box C/D snoRNPs and rRNA methylation. (A) Time-dependent loss ofbox C/D core proteins on OGT inhibition. HeLa cells were treated with 50 μM Ac45SGlcNAc for up to 72 h, and the protein level was monitored by immu-noblotting. RL2 was used to monitor the cellular level of O-GlcNAc. Actin was used as the loading control. The relative protein level was shown under eachband. Statistical analysis is shown in Fig. S6 A–D. (B) Inhibition of O-GlcNAcylation expedites degradation of box C/D core proteins. HeLa cells were treatedwith DMSO or 50 μM Ac45SGlcNAc for 48 h followed by treatment with 50 μM CHX for up to 8 h and monitoring of the protein level. Actin was used as theloading control. The relative protein level is shown under the bands. Statistical analysis is shown in Fig. S8. (C) O-GlcNAcylation of Ser142 is critical formaintaining the stability of FBL. HeLa cells expressing FBLWT-Flag/His and FBLS142A-Flag/His were treated with 50 μM CHX for 12 h. The protein level wasmonitored by immunoblotting with an anti-His antibody. Actin was used as the loading control. The relative protein level is shown under each band. Sta-tistical analysis is shown in Fig. S10D. (D) Fluorescence imaging of HeLa cells expressing GFP-FBLWT or GFP-FBLS142A, which were costained with anti-NOP58 followed by an Alexa Fluor 555-conjugated secondary antibody (nucleolus; red signal) and Hoechst 33342 (nuclei; blue signal). (Scale bar: 1 μm.)(E) Boxplots of the nucleolar proportions of GFP-FBLWT and GFP-FBLS142A. The nucleolar proportion was estimated as the ratio of nucleolar fluorescenceintensity to nuclear fluorescence intensity of individual cells (n = 50). Box limits represent 25th percentiles, medians, and 75th percentiles. Whiskers extend tothe maximum and minimum. **P < 0.01 (Student’s t test). (F) O-GlcNAcylation of Ser142 on FBL regulates the assembly of box C/D snoRNPs. Coimmuno-precipitation (Co-IP) of NOP56, NOP58, and NHP2L1 with FBLWT-Flag/His and FBLS142A-Flag/His. Anti-His blotting was used to adjust the loading of precipitatedFBL. Values under each band indicate relative protein levels coimmunoprecipitated with FBL. In the bar graph, values represent means ± SD from threereplicate experiments. ***P < 0.001 (Student’s t test). (G) OGT inhibition reduces the overall methylation level on rRNAs. The relative amounts of Am and Gmwere quantified by LC-MS/MS. ***P < 0.001 (Student’s t test). (H) OGT inhibition reduces the relative methylation level at five sites of rRNA measured by qRT-PCR. Values represent means ± SD from three replicate experiments. *P < 0.05 (Student’s t test); **P < 0.01 (Student’s t test); ***P < 0.001 (Student’s t test).

Qin et al. PNAS | Published online July 31, 2017 | E6755

CHEM

ISTR

YBIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

14, 2

021

As exemplified by NHP2L1, it is possible that global deprivationof O-GlcNAc may indirectly affect the stability of certain proteins.To rule out such an indirect effect for FBL, we used its S142Amutant to specifically assay the functional role of O-GlcNAylationon its stability without interference by O-GlcNAc on other proteins.Computational modeling by Rosetta (48) indicated that the S142Amutation did not cause significant perturbation to the proteinstructural stability (Fig. S10A). Furthermore, FBLWT and FBLS142A

were expressed in Escherichia coli, which does not have O-GlcNAcmodification. The thermal shift assay on cell lysates showed thatFBLS142A exhibited unaltered thermal stability compared withFBLWT, suggesting that S142 mutation itself does not cause proteininstability (Fig. S10B). We then transfected HeLa cells with equalamounts of plasmids of FBLS142A-Flag/His and FBLWT-Flag/His,respectively. Comparable mRNA levels were confirmed by qRT-PCR (Fig. S10C). As expected, Western blot analysis showed thatthe protein expression level of FBLS142A-Flag/His was significantlylower and that the degradation rate was much faster than that of theWT FBL-Flag/His (Fig. 5C and Fig. S10D). Furthermore, confocalfluorescence microscopy confirmed that the S142A mutationresulted in a decreased protein level of GFP-FBL in HeLa cells(Fig. S10 E and F). Collectively, these results show that the stabilityof FBL was directly regulated by O-GlcNAcylation at S142.

O-GlcNAc Is Essential for Subcellular Location of FBL and Its Interactionswith Other Core Proteins. The box C/D snoRNP proteins are syn-thesized in the cytosol, imported into the nucleoplasm and Cajalbodies for assembly with box C/D RNAs, and then, transported tothe nucleolus as the mature snoRNPs (49). To investigate the ef-fects of O-GlcNAcylation on the subcellular location of FBL, HeLacells transiently expressing GFP-FBL or GFF-FBLS142A were visu-alized by immunofluorescence microscopy. Hoechst staining andimmunofluorescence staining of endogenous NOP58 were used todelineate the boundaries of nucleus and nucleolus, respectively (Fig.5D). The distribution of FBL or FBLS142A in each subcellular regionwas quantified by GFP fluorescence intensity, which showed thatboth the overall expression level and the nucleolar/nuclear per-centage of GFP-FBLS142A were significantly lower than those ofGFP-FBL (Fig. 5E). These results indicate that O-GlcNAc is im-portant in regulating the subnuclear localization of FBL.To assay how O-GlcNAcylation of FBL affects its interactions

with other box C/D snoRNP core proteins, FBLS142A-Flag/Hisand FBLWT-Flag/His were expressed in HeLa cells and immu-noprecipitated using the anti-Flag antibody. The FBL-interacting

proteins were coimmunoprecipitated, and their abundance wascompared by Western blot analysis. Starting with the sameamount of FBLS142A-Flag/His and FBLWT-Flag/His, much lessNOP56, NOP58, and NHP2L1 were coimmunoprecipitated withthe mutant FBL than with the WT, indicating that loss ofO-GlcNAcylation at S142 of FBL disrupted its interactions withcore proteins of box C/D snoRNPs and the assembly of themature complex (Fig. 5F).

O-GlcNAc–Mediated Alteration of rRNA Methylation. Given thatO-GlcNAcylation regulates the stability of C/D snoRNP coreproteins, we assessed whether changes in the O-GlcNAc levelmay alter rRNA methylation, one of the key functions of the boxC/D snoRNPs. The 5S, 18S, and 28S rRNAs from Ac45SGlc-NAc- and DMSO-treated cells were isolated and digested intonucleosides. The relative amounts of methylated adenosine (Am)and methylated guanine (Gm) were quantified by LC-MS/MS (50).Our results showed that O-GlcNAcylation inhibition reducedmethylation globally in rRNAs (Fig. 5G). In addition, we ana-lyzed the methylation level of five known methylation sites lo-cated in the 5S, 18S, and 28S rRNAs using a site-specificsemiquantitative qRT-PCR–based method (51, 52). Consistentwith the global reduction of rRNA methylation, all of five siteswere significantly less methylated in the Ac45SGlcNAc-treatedcells, where the overall O-GlcNAc level was decreased and theexpression of FBL was repressed (Fig. 5H). These results in-dicate that aberrant O-GlcNAcylation alters the rRNA methyl-ation pattern, presumably through destabilizing core proteinsand regulating the assembly and biogenesis of box C/D snoRNPs.

O-GlcNAcylation of FBL at S142 Facilitates Tumorigenesis. FBLoverexpression has been implicated in tumorigenesis, probablythrough modulation of ribosome activity by altering rRNA methyl-ation (52–54). To investigate whether O-GlcNAcylation of FBLcontributes to tumorigenesis, we stably expressed FBL or FBLS142A

in MCF7 cancer cells, and in agreement with the previous results(52), overexpression of FBL in MCF7 cells significantly increasedcolony formation and cell proliferation (Fig. 6A and B). Mutation oftheO-GlcNAcylation site at S142 completely abolished these effects,suggesting that O-GlcNAc modification is essential for the FBL-promoted colony formation and cell proliferation. In addition,overexpression of FBL but not FBLS142A protected MCF7 cellsagainst the drug doxorubicin (Fig. 6C). The impact of FBLO-GlcNAcylation on tumor growth was further investigated in vivoby using a xenograft mouse model. MCF7 cells stably expressingFBL or FBLS142A were injected into nude mice. Decreased tumormass was observed in mice injected with FBLS142A-expressing cellscompared with mice injected with FBLWT-expressing cells (Fig. 6D).Taken together, these in vitro and in vivo results show that thehyperstable O-GlcNAcylation of FBL at Ser142, through maintain-ing the protein stability, promotes cancer cell proliferation andtumor formation.

DiscussionIt is intriguing that over 1,000 proteins are O-GlcNAcylated by asingle glycosyltransferase OGT. In this regard, oligosaccharyl-transferase, an endoplasmic reticulum (ER) membrane proteincomplex that catalyzes the first step of protein N-linked glycosyl-ation, is one of the few enzymes that have comparably broadprotein substrates (55). Although N-glycosylation is generallyregarded as an irreversible modification for membrane-associatedand -secreted proteins, O-GlcNAcylation is thought to be dy-namically regulated by OGA in concerted with OGT. The level ofO-GlcNAcylation fluctuates in response to the nutrient conditionand cellular stress (5, 56–58). However, whether OGA can removeO-GlcNAc from all modified proteins and whether the turnoverdynamics is differentially regulated among O-GlcNAcylated pro-teins remain largely uninvestigated.

Fig. 6. O-GlcNAcylation of FBL contributes to cell proliferation and tumorgrowth. (A–C) Colony formation (A), cell proliferation (B), and cell growthwith doxorubicin (C) of MCF7 cells expressing GFP, FBLWT, or FBLS142A. Valuesrepresent means ± SD from three replicate experiments. N.S., not significant(one-way ANOVA). ***P < 0.001. (D) Tumor formation in nude mice injectedwith MCF7 cells expressing FBLWT or FBLS142A. Shown in Upper are repre-sentative photographs of dissected tumors after 1 wk of growth in mice.Shown in Lower are masses of the dissected tumors. Values representmeans ± SD from three replicate experiments. ***P < 0.001 (Student’s t test).

E6756 | www.pnas.org/cgi/doi/10.1073/pnas.1702688114 Qin et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

14, 2

021

In this study, we developed the qTOP platform that enablesquantitative analysis of the turnover rates of O-GlcNAcylatedproteins. Our analysis revealed a population of hyperstableO-GlcNAcylated proteins, with O-GlcNAcylation that is mini-mally perturbed within a 12-h timeframe. Moreover, we foundthat the abundance of those proteins carrying hyperstableO-GlcNAc is more susceptible to OGT inhibition than those in thedynamic and hyperdynamic populations. We, therefore, propose thatthe stable O-GlcNAc with slow turnover kinetics has importantfunctional implications in regulating the stability of modified proteins.Among the hyperstable O-GlcNAcylated proteins identified in

this study are FBL, NOP56, and NOP58, three of the box C/DsnoRNP core proteins. The assembly pathway of box C/D snoRNPshas been extensively studied, and various assembly factors, such asthe HSP90/R2TP chaperone/cochaperone system, have been char-acterized (49). However, posttranslational modifications (PTMs) ofthe core proteins and their functional roles in snoRNP biogenesisare rarely studied. Recent proteomic profiling identified NOP58as a substrate of small ubiquitin-related modifier (SUMO), whichpromotes its high-affinity binding with snoRNAs (59, 60).Moreover, SUMO modification of NOP58 seems to be regulatedby flanking serine phosphorylation (60). Our study revealsO-GlcNAcylation as another PTM that regulates snoRNP as-sembly and biogenesis through regulating the stability, subcellularlocation, and interactions of the core proteins. The extensive in-terplays between those PTMs are very possible and remain tobe investigated.Alteration of O-GlcNAcylation impacts the functions of

box C/D snoRNPs. For example, we show that inhibition ofO-GlcNAcylation impairs the rRNA methylation level and pat-tern. Altered rRNA methylation was reported to affect ribosomeproduction and translation (52, 61). Correlatively, ribosomeproteins were also found to be modified with O-GlcNAc (62, 63).Together, O-GlcNAc has proven an important PTM for regu-lating ribosome biogenesis and functions.Elevated O-GlcNAcylation has been found in various cancer

types and emerged as a general feature of cancer (64). Many ofthe O-GlcNAcylated proteins in cancer cells are associated withtumorigenesis. One mechanism through which O-GlcNAc con-tributes to tumorigenesis is to regulate activities of those pro-teins. For example, O-GlcNAcylation of phosphofructokinase1 was found to block its kinase activity and increase glucose fluxthrough the pentose phosphate pathway (PPP), which promotedcancer cell proliferation and tumor formation (12). Conversely,O-GlcNAc modification could activate the rate-limiting enzymeof the PPP, glucose-6-phosphate dehydrogenase, thus increasingglucose flux through the PPP (14). Alternatively, O-GlcNAcyla-tion may stabilize proteins promoting tumorigenesis. The ex-amples include c-MYC (65), FoxM1 (66), and Snail1 (67). In thiswork, we show that O-GlcNAcylation on box C/D snoRNP coreproteins, particularly FBL, enhances the stability of box C/DsnoRNPs, which impacts rRNA methylation and ribosome bio-genesis. The dysregulation of ribosome biogenesis in cancer cellsplays key roles in oncogenesis (68). Therefore, our work providesanother link between O-GlcNAc and tumorigenesis.Metabolic labeling with chemical reporters, such as 6AzGlc-

NAc, enables the pulse-chase experiments for dynamic profilingof O-GlcNAc, which is difficult to achieve by using other labelingmethods. Notably, 6AzGlcNAc incorporation results in an un-natural alteration on O-GlcNAc. Although we show that theresulting O-6AzGlcNAc can be removed by OGA, it might bepossible that it is removed at a different rate. Nevertheless, the

O-GlcNAc dynamics and the effects on protein stability quanti-fied by qTOP are validated by SILAC-based proteomic quantifi-cation of the protein abundance on OGT inhibition. In addition,functional studies on individual proteins were cross-validated byusing methods detecting natural O-GlcNAc, such as chemo-enzymatic labeling with Y289L GalT and an anti–O-GlcNAc an-tibody, in addition to the chemical reporter-based labeling.Other than the cytoplasmic and nuclear O-GlcNAc modifica-

tion, an EGF-specific O-GlcNAcylation was recently discoveredto occur in the ER for several secreted proteins containing EGFdomains (69, 70). The modification is catalyzed by an ER-localized EGF-specific O-GlcNAc transferase (EOGT). In thiswork, none of these EOGT-modified proteins were quantified inqTOP. Interestingly, a recent study detected and identified aGlcNAc modification on cysteine of a few proteins, suggestingthe possible occurrence of S-GlcNAcylation (71). Because it wasshown that S-GlcNAc was not cleaved by OGA, the fact that theAc36AzGlcNAc labeling is sensitive to OGA suggests that thisparticular chemical reporter mainly labels O-GlcNAc. Further-more, we performed quantitative proteomics with the Y289LGalT labeling to ensure that those proteins analyzed with qTOPare indeed O-GlcNAcylated.In conclusion, we present a chemoproteomic strategy for se-

lective enrichment and quantitative analysis of O-GlcNAcylatedproteins as well as their turnover profiles in living cells. The largenumber of O-GlcNAcylated proteins categorized with distinctturnover profiles provides a rich and valuable resource for betterunderstanding the O-GlcNAc biology. However, it should benoted that the qTOP methodology currently does not differen-tiate the removal of O-GlcNAc modification from the degrada-tion of modified proteins, and we, therefore, chose to focus onthe hyperstable population where both the protein and themodification appear to be highly stable with slow turnover ki-netics. For the dynamic and hyperdynamic populations, it ispossible that O-GlcNAc turns over more rapidly than protein orvice versa. To further distinguish between those situations, weenvision expansion of qTOP with multiplexed quantitative pro-teomics. Equally important is to develop effective methods forsite-specific profiling of O-GlcNAcylation that can be integratedinto qTOP, and this integration would ultimately enable detailedanalysis for proteins bearing multiple O-GlcNAc modificationswith different turnover kinetics.

Materials and MethodsDetails are in SI Materials and Methods, which includes detailed methods forquantitative chemoproteomic profiling, fluorescence imaging, rRNA meth-ylation quantification, quantitative real-time PCR, in-gel fluorescence, cellproliferation assay, and in vivo tumor xenografts. All animal experimentswere performed in accordance with guidelines approved by the InstitutionalAnimal Care and Use Committee of Peking University accredited byAssociation for Assessment and Accreditation of Laboratory Animal Care(AAALAC) International.

ACKNOWLEDGMENTS. We thank Dr. Guifang Jia and Ye Wang for help withLC-MS/MS quantification of rRNA methylation, Dr. Yuan Liu for help withRosetta simulation, the MS facility of the National Center for ProteinSciences at Peking University for assistance with ETD MS, and the computingplatform of the Center for Life Science for supporting the LC-MS/MS dataanalysis. This work is supported by National Natural Science Foundation ofChina Grants 21472008 (to C.W.), 81490740 (to C.W.), 21425204 (to X.C.),21521003 (to X.C.), and 21672013 (to X.C.); National Key Research andDevelopment Projects Grant 2016YFA0501500 (to C.W. and X.C.); and a“1000 Talents Plan” Young Investigator Award (to C.W.).

1. Hart GW, Slawson C, Ramirez-Correa G, Lagerlof O (2011) Cross talk betweenO-GlcNAcylation

and phosphorylation: Roles in signaling, transcription, and chronic disease.Annu Rev Biochem

80:825–858.2. Zhu Y, et al. (2015) O-GlcNAc occurs cotranslationally to stabilize nascent polypeptide

chains. Nat Chem Biol 11:319–325.

3. Yang X, Zhang F, Kudlow JE (2002) Recruitment of O-GlcNAc transferase to promoters

by corepressor mSin3A: Coupling protein O-GlcNAcylation to transcriptional re-

pression. Cell 110:69–80.4. Wells L, Vosseller K, Hart GW (2001) Glycosylation of nucleocytoplasmic proteins:

Signal transduction and O-GlcNAc. Science 291:2376–2378.

Qin et al. PNAS | Published online July 31, 2017 | E6757

CHEM

ISTR

YBIOCH

EMISTR

YPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

14, 2

021

5. Wang ZV, et al. (2014) Spliced X-box binding protein 1 couples the unfolded proteinresponse to hexosamine biosynthetic pathway. Cell 156:1179–1192.

6. Ruan H-B, Nie Y, Yang X (2013) Regulation of protein degradation by O-GlcNAcylation:Crosstalk with ubiquitination. Mol Cell Proteomics 12:3489–3497.

7. Zhang F, et al. (2003) O-GlcNAc modification is an endogenous inhibitor of the pro-teasome. Cell 115:715–725.

8. Yang WH, et al. (2006) Modification of p53 with O-linked N-acetylglucosamine reg-ulates p53 activity and stability. Nat Cell Biol 8:1074–1083.

9. Li M-D, et al. (2013) O-GlcNAc signaling entrains the circadian clock by inhibitingBMAL1/CLOCK ubiquitination. Cell Metab 17:303–310.

10. Zhu Y, et al. (2016) Post-translational O-GlcNAcylation is essential for nuclear poreintegrity and maintenance of the pore selectivity filter. J Mol Cell Biol 8:2–16.

11. Erickson JR, et al. (2013) Diabetic hyperglycaemia activates CaMKII and arrhythmiasby O-linked glycosylation. Nature 502:372–376.

12. Yi W, et al. (2012) Phosphofructokinase 1 glycosylation regulates cell growth andmetabolism. Science 337:975–980.

13. Kaasik K, et al. (2013) Glucose sensor O-GlcNAcylation coordinates with phosphory-lation to regulate circadian clock. Cell Metab 17:291–302.

14. Rao X, et al. (2015) O-GlcNAcylation of G6PD promotes the pentose phosphatepathway and tumor growth. Nat Commun 6:8468.

15. Haltiwanger RS, Holt GD, Hart GW (1990) Enzymatic addition of O-GlcNAc to nuclearand cytoplasmic proteins. Identification of a uridine diphospho-N-acetylglucosamine:peptide beta-N-acetylglucosaminyltransferase. J Biol Chem 265:2563–2568.

16. Lubas WA, Frank DW, Krause M, Hanover JA (1997) O-linked GlcNAc transferase is aconserved nucleocytoplasmic protein containing tetratricopeptide repeats. J BiolChem 272:9316–9324.

17. Gao Y, Wells L, Comer FI, Parker GJ, Hart GW (2001) Dynamic O-glycosylation ofnuclear and cytosolic proteins: Cloning and characterization of a neutral, cytosolicbeta-N-acetylglucosaminidase from human brain. J Biol Chem 276:9838–9845.

18. Ma J, Hart GW (2014) O-GlcNAc profiling: From proteins to proteomes. Clin Proteomics11:8.

19. Wang Z, Pandey A, Hart GW (2007) Dynamic interplay between O-linkedN-acetylglucosaminylation and glycogen synthase kinase-3-dependent phosphorylation.Mol Cell Proteomics 6:1365–1379.

20. Trinidad JC, et al. (2012) Global identification and characterization of bothO-GlcNAcylation and phosphorylation at the murine synapse. Mol Cell Proteomics 11:215–229.

21. Alfaro JF, et al. (2012) Tandem mass spectrometry identifies many mouse brainO-GlcNAcylated proteins including EGF domain-specific O-GlcNAc transferase tar-gets. Proc Natl Acad Sci USA 109:7280–7285.

22. Hahne H, et al. (2013) Proteome wide purification and identification of O-GlcNAc-modified proteins using click chemistry and mass spectrometry. J Proteome Res 12:927–936.

23. Chuh KN, Zaro BW, Piller F, Piller V, Pratt MR (2014) Changes in metabolic chemicalreporter structure yield a selective probe of O-GlcNAc modification. J Am Chem Soc136:12283–12295.

24. Woo CM, Iavarone AT, Spiciarich DR, Palaniappan KK, Bertozzi CR (2015) Isotope-targeted glycoproteomics (IsoTaG): A mass-independent platform for intact N- andO-glycopeptide discovery and analysis. Nat Methods 12:561–567.

25. Zhao P, et al. (2011) Combining high-energy C-trap dissociation and electron transferdissociation for protein O-GlcNAc modification site assignment. J Proteome Res 10:4088–4104.

26. Vosseller K, et al. (2006) O-linked N-acetylglucosamine proteomics of postsynapticdensity preparations using lectin weak affinity chromatography and mass spec-trometry. Mol Cell Proteomics 5:923–934.

27. Khidekel N, Ficarro SB, Peters EC, Hsieh-Wilson LC (2004) Exploring the O-GlcNAcproteome: Direct identification of O-GlcNAc-modified proteins from the brain. ProcNatl Acad Sci USA 101:13132–13137.

28. Vocadlo DJ, Hang HC, Kim E-J, Hanover JA, Bertozzi CR (2003) A chemical approachfor identifying O-GlcNAc-modified proteins in cells. Proc Natl Acad Sci USA 100:9116–9121.

29. Boyce M, et al. (2011) Metabolic cross-talk allows labeling of O-linked beta-N-acetylglucosamine-modified proteins via the N-acetylgalactosamine salvage path-way. Proc Natl Acad Sci USA 108:3141–3146.

30. Zaro BW, Yang Y-Y, Hang HC, Pratt MR (2011) Chemical reporters for fluorescentdetection and identification of O-GlcNAc-modified proteins reveal glycosylation ofthe ubiquitin ligase NEDD4-1. Proc Natl Acad Sci USA 108:8146–8151.

31. Li J, et al. (2016) An OGA-Resistant Probe Allows Specific Visualization and AccurateIdentification of O-GlcNAc-Modified Proteins in Cells. ACS Chem Biol 11:3002–3006.

32. Zhang MM, Tsou LK, Charron G, Raghavan AS, Hang HC (2010) Tandem fluorescenceimaging of dynamic S-acylation and protein turnover. Proc Natl Acad Sci USA 107:8627–8632.

33. Martin BR, Wang C, Adibekian A, Tully SE, Cravatt BF (2011) Global profiling of dy-namic protein palmitoylation. Nat Methods 9:84–89.

34. Xie R, et al. (2016) In vivo metabolic labeling of sialoglycans in the mouse brain byusing a liposome-assisted bioorthogonal reporter strategy. Proc Natl Acad Sci USA113:5173–5178.

35. McShane E, et al. (2016) Kinetic analysis of protein stability reveals age-dependentdegradation. Cell 167:803–815.e21.

36. Watkins NJ, Bohnsack MT (2012) The box C/D and H/ACA snoRNPs: Key players in themodification, processing and the dynamic folding of ribosomal RNA.Wiley InterdiscipRev RNA 3:397–414.

37. Washburn MP, Wolters D, Yates JR, 3rd (2001) Large-scale analysis of the yeast pro-teome by multidimensional protein identification technology. Nat Biotechnol 19:242–247.

38. Xu T, et al. (2015) ProLuCID: An improved SEQUEST-like algorithm with enhancedsensitivity and specificity. J Proteomics 129:16–24.

39. Weerapana E, et al. (2010) Quantitative reactivity profiling predicts functional cys-teines in proteomes. Nature 468:790–795.

40. Ramakrishnan B, Qasba PK (2002) Structure-based design of beta 1,4-galactosyl-transferase I (beta 4Gal-T1) with equally efficient N-acetylgalactosaminyltransferaseactivity: Point mutation broadens beta 4Gal-T1 donor specificity. J Biol Chem 277:20833–20839.

41. Clark PM, et al. (2008) Direct in-gel fluorescence detection and cellular imaging ofO-GlcNAc-modified proteins. J Am Chem Soc 130:11576–11577.

42. Roquemore EP, Chevrier MR, Cotter RJ, Hart GW (1996) Dynamic O-GlcNAcylation ofthe small heat shock protein alpha B-crystallin. Biochemistry 35:3578–3586.

43. Khidekel N, et al. (2007) Probing the dynamics of O-GlcNAc glycosylation in the brainusing quantitative proteomics. Nat Chem Biol 3:339–348.

44. Bond MR, Hanover JA (2015) A little sugar goes a long way: The cell biology of O-GlcNAc.J Cell Biol 208:869–880.

45. Wang X, et al. (2016) A novel quantitative mass spectrometry platform for de-termining protein O-GlcNAcylation dynamics. Mol Cell Proteomics 15:2462–2475.

46. Gloster TM, et al. (2011) Hijacking a biosynthetic pathway yields a glycosyltransferaseinhibitor within cells. Nat Chem Biol 7:174–181.

47. Rexach JE, et al. (2010) Quantification of O-glycosylation stoichiometry and dynamicsusing resolvable mass tags. Nat Chem Biol 6:645–651.

48. Kellogg EH, Leaver-Fay A, Baker D (2011) Role of conformational sampling in com-puting mutation-induced changes in protein structure and stability. Proteins 79:830–838.

49. Massenet S, Bertrand E, Verheggen C (2016) Assembly and trafficking of box C/D andH/ACA snoRNPs. RNA Biol 2016:1–13.

50. Jia G, et al. (2011) N6-methyladenosine in nuclear RNA is a major substrate of theobesity-associated FTO. Nat Chem Biol 7:885–887.

51. Maden BE (1988) Locations of methyl groups in 28 S rRNA of Xenopus laevis and man.Clustering in the conserved core of molecule. J Mol Biol 201:289–314.

52. Marcel V, et al. (2013) p53 Acts as a safeguard of translational control by regulatingfibrillarin and rRNA methylation in cancer. Cancer Cell 24:318–330.

53. Koh CM, et al. (2011) Alterations in nucleolar structure and gene expression programsin prostatic neoplasia are driven by the MYC oncogene. Am J Pathol 178:1824–1834.

54. Su H, et al. (2014) Elevated snoRNA biogenesis is essential in breast cancer. Oncogene33:1348–1358.

55. Zielinska DF, Gnad F, Wi�sniewski JR, Mann M (2010) Precision mapping of an in vivoN-glycoproteome reveals rigid topological and sequence constraints. Cell 141:897–907.

56. Bond MR, Hanover JA (2013) O-GlcNAc cycling: A link between metabolism andchronic disease. Annu Rev Nutr 33:205–229.

57. Pekkurnaz G, Trinidad JC, Wang X, Kong D, Schwarz TL (2014) Glucose regulatesmitochondrial motility via Milton modification by O-GlcNAc transferase. Cell 158:54–68.

58. Denzel MS, et al. (2014) Hexosamine pathway metabolites enhance protein qualitycontrol and prolong life. Cell 156:1167–1178.

59. Westman BJ, et al. (2010) A proteomic screen for nucleolar SUMO targets showsSUMOylation modulates the function of Nop5/Nop58. Mol Cell 39:618–631.

60. Matic I, et al. (2010) Site-specific identification of SUMO-2 targets in cells reveals aninverted SUMOylation motif and a hydrophobic cluster SUMOylation motif. Mol Cell39:641–652.

61. Belin S, et al. (2009) Dysregulation of ribosome biogenesis and translational capacityis associated with tumor progression of human breast cancer cells. PLoS One 4:e7147.

62. Ohn T, Kedersha N, Hickman T, Tisdale S, Anderson P (2008) A functional RNAi screenlinks O-GlcNAc modification of ribosomal proteins to stress granule and processingbody assembly. Nat Cell Biol 10:1224–1231.

63. Zeidan Q, Wang Z, De Maio A, Hart GW (2010) O-GlcNAc cycling enzymes associatewith the translational machinery and modify core ribosomal proteins. Mol Biol Cell21:1922–1936.

64. Ma Z, Vosseller K (2014) Cancer metabolism and elevated O-GlcNAc in oncogenicsignaling. J Biol Chem 289:34457–34465.

65. Itkonen HM, et al. (2013) O-GlcNAc transferase integrates metabolic pathways toregulate the stability of c-MYC in human prostate cancer cells. Cancer Res 73:5277–5287.

66. Caldwell SA, et al. (2010) Nutrient sensor O-GlcNAc transferase regulates breastcancer tumorigenesis through targeting of the oncogenic transcription factor FoxM1.Oncogene 29:2831–2842.

67. Park SY, et al. (2010) Snail1 is stabilized by O-GlcNAc modification in hyperglycaemiccondition. EMBO J 29:3787–3796.

68. Truitt ML, Ruggero D (2016) New frontiers in translational control of the cancer ge-nome. Nat Rev Cancer 16:288–304.

69. Sakaidani Y, et al. (2011) O-linked-N-acetylglucosamine on extracellular protein do-mains mediates epithelial cell-matrix interactions. Nat Commun 2:583.

70. Varshney S, Stanley P (2017) EOGT and O-GlcNAc on secreted and membrane pro-teins. Biochem Soc Trans 45:401–408.

71. Maynard JC, Burlingame AL,Medzihradszky KF (2016) Cysteine S-linked N-acetylglucosamine(S-GlcNAcylation), a newpost-translationalmodification inmammals.Mol Cell Proteomics 15:3405–3411.

E6758 | www.pnas.org/cgi/doi/10.1073/pnas.1702688114 Qin et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

14, 2

021