O-Linked-N-acetylglucosamine on extracellular protein domains mediates epithelial cell-matrix interactions

Abstract

The O-linked-N-acetylglucosamine (O-GlcNAc) modification of cytoplasmic and nuclear proteins regulates basic cellular functions and is involved in the aetiology of diabetes and neurodegeneration. This intracellular O-GlcNAcylation is catalyzed by a single O-GlcNAc transferase, OGT. Here we report a novel OGT, EOGT, responsible for extracellular O-GlcNAcylation. Although both OGT and EOGT are regulated by hexosamine flux, EOGT localizes to the lumen of the endoplasmic reticulum and transfers GlcNAc to epidermal growth factor-like domains in an OGT-independent manner. Loss of Eogt gives phenotypes similar to those caused by defects in the apical extracellular matrix. Dumpy (Dp), a membrane-anchored extracellular protein, is O-GlcNAcylated, and EOGT is required for Dp-dependent epithelial cell-matrix interactions. Thus, O-GlcNAcylation of secreted and membrane glycoproteins is a novel mediator of cell-cell or cell-matrix interactions at the cell surface.

ARTICLE

Received 6 Jun 2011 | Accepted 11 Nov 2011 | Published 13 Dec 2011 DOI: 10.1038/ncomms1591

O-Linked-N-acetylglucosamine on extracellular protein domains mediates epithelial cell matrix interactions

Yuta Sakaidani1,2,*, Tomoko Nomura1,*, Aiko Matsuura2,*, Makiko Ito2, Emiko Suzuki3, Kosuke Murakami2, Daita Nadano2, Tsukasa Matsuda2, Koichi Furukawa1 & Tetsuya Okajima1,2

The O-linked-N-acetylglucosamine (O-GlcNAc) modication of cytoplasmic and nuclear proteins regulates basic cellular functions and is involved in the aetiology of diabetes and neurodegeneration. This intracellular O-GlcNAcylation is catalyzed by a single O-GlcNAc transferase, OGT. Here we report a novel OGT, EOGT, responsible for extracellular O-GlcNAcylation. Although both OGT and EOGT are regulated by hexosamine ux, EOGT localizes to the lumen of the endoplasmic reticulum and transfers GlcNAc to epidermal growth factor-like domains in an OGT-independent manner. Loss of Eogt givesphenotypessimilarto those caused by defects in the apical extracellular matrix. Dumpy (Dp), a membrane-anchored extracellular protein, is O-GlcNAcylated, and EOGT is required for Dp-dependent epithelial cell matrix interactions. Thus, O-GlcNAcylation of secreted and membrane glycoproteins is a novel mediator of cellcell or cellmatrix interactions at the cell surface.

1 Department of Biochemistry II, Nagoya University Graduate School of Medicine, 65 Tsurumai, Showa-ku, Nagoya 466-0065, Japan . 2Department of Applied Molecular Biosciences, Nagoya University Graduate School of Bioagricultural Sciences, Furo-cho, Chikusa-ku, Nagoya 464-8601, Japan . 3Structural Biology Center, National Institute of Genetics, Department of Genetics, The Graduate University for Advanced Studies, Mishima 411-8540, Japan. *These authors contributed equally to this work. Correspondence and requests for materials should be addressed to T.O. (email: [email protected]).

1

NATURE COMMUNICATIONS | 2:583 | DOI: 10.1038/ncomms1591 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1591

O-Linked- N-acetylglucosamine(O-GlcNAc)modication is a unique post-translational modication that cooperates with phosphorylation to regulate a variety of cellular proc

esses, including intracellular signalling, cytokinesis, transcription and protein stability, and serves to modulate nutrient sensing 13. O-GlcNAcylation is also involved in epigenetic control of cell dierentiation and is implicated in the aetiology of several human diseases, including type 2 diabetes and neurodegeneration 46. Although many reports have described biological roles of O-GlcNA-cylation, they were limited to the intracellular protein functions 7, as the modication had been believed to be restricted to the cytoplasm and nucleus 7 . However, a few reports have identied O-GlcNAc in the extracellular environment 8,9. Nevertheless, this could be attributed to a non-canonical secretion pathway for proteins that bypass the endoplasmic reticulum (ER) and Golgi compartments 7. To date, a sole O-GlcNAc transferase, OGT 10 , has been shown to catalyze intracellular O-GlcNAcylation 11,12.

Here we show that O-GlcNAc is expressed in the apical extra-cellular matrix (aECM) and that it is required for epithelial cell adhesion / interaction with the ECM in Drosophila. Th is cellmatrix

interaction is mediated by a novel ER-resident O-GlcNAc transferase, EOGT. Th is enzyme is responsible for O-GlcNAcylation of proteins in the secretory pathway independently of OGT activity. Dumpy (Dp), a membrane-anchored extracellular protein, is O-GlcNAcylated, and EOGT mediates cell matrix interactions in a Dp-dependent manner. These ndings reveal new biological roles for O-GlcNAcylation in secreted and membrane glycoproteins and the extracellular environment.

ResultsEOGT is required for extracellular O-GlcNAcylation. O-GlcNAc transferase activity responsible for modifying the epidermal growth factor (EGF) repeats in the extracellular domain of Notch was previously detected in the membrane fraction of Drosophila S2 cells 9. We performed an RNA interference (RNAi)-based search for genes that specically aect O-GlcNAcylation by following changes in glycosylation by matrix-assisted laser desorption / ionization time of ight-mass spectrometry. As other Notch O-glycosyltransferases are ER-resident proteins 1315 , we initiated screening by targeting putative glycosyltransferases possessing an ER retention signal,

Eogt RNAi

IB: V5

CG9109

Control

CG31915 RNAi

Cont

Ofut1

Eogt

CG31915

8,949

8,946

Eogt RNAi

8,879

8,746 8,950

9,084

Control

(kDa)

(kDa)

171 117

171 117

Relative intensity

(kDa)

15

10

3.5

15 103.5

204

205

IB: O-GlcNAc

IB: FLAG (N-EGF:FLAG)

8,584

IP: FLAG

IB: O-GlcNAc (CTD110.6)

Eogt dsRNA

8,791

8,584

9,082

8,787 8,580

9,078

268

EGF20:V5His

268

IP: Notch

+ IB: O-GlcNAc (CTD110.6)

IB: Notch

m/z m/z m/z

Dros

1 MPILPILIGILHLS------------LAEDAKHLDGFSLPSLPSEHLIRYLN----TFPKMLMLLVFGVLLHEVPLSGQDKAHSEADDAPGKALYDYSSLRLPAEHIPFFLHNNRHVASV

LKQQLPTNLTGKGTISSACWGHERDCTPAGRFQTPQCPGEHTGWARSKEAQVRTFYNQADCREDSHCPYKKHLENLNYCWGYEKSCAPEFRFGSPVCSYVDLGWTDTLESAQDMFWRQAD

FGYIQEQLSQLTPQCVPTYLGDSSLECTHYLRFCRGRNLLFDFRGLEQREERIRYHMDVLFGYARERLGEIRTICQPERASDSSLVCSRYLQYCRATGLYLDLRNIKRN--HDRFKEDFL

GPGQLLGHCKLNRTRLSGEMEHIGSALQSWGPELRNFDVLPHPVLESGLCDVVVNTPTFIQGGEIGGYCKLDSHALVSEGQRK-SPLQSWFAELQGYTQLNFRPIEDAKCDIVVEKPTYF

MKIDATYNMYHHFCDFFNLYASLFVNQSHPAAFNTDVQILIWET--YPYDSPFRDTFKAFMKLDAGINMYHHFCDFLNLYLTQHVNN----SFSTDVYIVMWDTSTYGYGDLFSDTWKAF

SQRPVWTLSDVEGKRVCFKNVVLPLLPRMIFGLFYNTPIIQGCSNSGLFRAFSEFILHRL

TDYDVIHLKTYDSKKVCFKEAVFSLLPRMRYGLFYNTPLISGCQNTGLFRAFSQHVLHRL

QIPYKPPQQ-KIRITYLSRRTKYRQVLNEDELLAPLEANDKYDVQRVSYER--LPFTNQLNITQEGPKDGKVRVTILARSTEYRKILNQDELVNALKTVSTFEVRVVDYKYRELGFLDQL

AITRNTDILIGMHGAGLTHLLFLPNWACIFELYNCEDPNCYKDLARLRGVRYRTWEQRDLRITHNTDIFIGMHGAGLTHLLFLPDWAAVFELYNCEDERCYLDLARLRGIHYITWRKPSK

VYPQDEGHHPEGGAHAKFTNYSFDVKEFVHLVDGAAEEILSHKEFPRRASENPSKTQRNEVFPQDKGHHPTLGEHPKFTNYSFDVEEFMYLVLQAAEHVLQHPQWPFKKKHDEL------

L-

Figure 1 | Identication of an EGF domain-specic O-GlcNAc transferase, EOGT. ( a) Matrix-assisted laser desorption/ionizationtime of ight mass spectrum of Notch EGF20 prepared from culture media of wild-type or dsRNA-treated S2 cells. S2 cells untreated or treated with dsRNA corresponding to CG31915 or Eogt were transiently transfected with a construct encoding EGF20:V5His. After afnity-purication using Ni-magnetic beads, proteins were analyzed by MALDI-TOF-MS using the 4700 Proteomic Analyzer in the linear positive mode as described previously 9. The mass/charge (m/z) values are shown in bold numbers. Peak-to-peak mass increment corresponding to O-GlcNAc modication is indicated by double-headed red arrows. The theoretical average mass for a singly charged and unglycosylated EGF20 peptide with three disulphide bonds is 8,438. ( b) Notch EGF20 fragments puried from culture media of S2 cells treated without or with Eogt dsRNA were subjected to immunoblotting (IB) with CTD110.6 (O-GlcNAc) and V5 antibodies.( c) Alignment of amino-acid sequences of Drosophila and mouse EOGT. Black-shaded residues indicate identity; grey-shaded residues indicate similarity.( d) Immunoblot analysis for O-GlcNAcylation of N-EGF:FLAG isolated from the medium of S2 cells. Cells were co-transfected with the indicated constructs. IP, immunoprecipitation. ( e) Effect of Eogt RNAi on O-GlcNAcylation of endogenous Notch. Proteins immunoprecipitated from Kc cell lysates with anti-Notch antibody were detected by anti-O-GlcNAc antibody.

2

NATURE COMMUNICATIONS | 2:583 | DOI: 10.1038/ncomms1591 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1591

ARTICLE

which are listed in the CAZy database ( http://www.cazy.org ). S2 cells were treated with double-stranded RNA (dsRNA) corresponding to the selected gene and simultaneously transfected to express Notch EGF domain 20 (EGF20). We identied one candidate gene, EOGT (EGF-domain O-GlcNAc transferase; Supplementary Fig. S1 ), whose downregulation decreased the signals corresponding to O-GlcNAcylated EGF20 peptides ( Fig. 1a,b and Supplementary Fig. S2 ). EOGT is conserved from Drosophila to mammals ( Fig. 1c ), but shows no apparent homology to previously isolated GlcNActransferases, such as OGT or GlcNAcT enzymes. Co-expression of FLAG (DYKDDDDK)-tagged Notch EGF repeats (N-EGF:FLAG) with EOGT or other selected genes revealed that EOGT specically elevates the O-GlcNAc level on EGF domains ( Fig. 1d ). Furthermore, O-GlcNAcylation of endogenous Notch was nearly eliminated by Eogt RNAi ( Fig. 1e ). Th us, EOGT retains an activity required for extracellular O-GlcNAcylation.

EOGT encodes an EGF domain O-GlcNAc transferase. To investigate whether EOGT itself has O-GlcNAc transferase activity, we isolated V5-hexahistidine-tagged EOGT (EOGT:V5His) from the membrane fraction of cultured S2 cells. The membrane fraction proteins or puried EOGT:V5His was incubated with yeast-expressed unglycosylated EGF20 and uridine diphosphate (UDP)-[ 3 H]GlcNAc in buer. Transfer of [ 3 H]GlcNAc to EGF20 was detected only in the presence of EOGT:V5His ( Fig. 2a,b ). Divalent cations, especially Mn 2+ , stimulated EOGT activity, although they were not absolutely required ( Fig. 2c,d ). Lack of a requirement for Mn 2+ has been reported in a subset of glycosyltransferases, including GlcNAcT-V 16 and plant 2-xylosyltransferase 17 . EOGT shows partial homology to the 2-xylosyltransferase (19.4% identity to Arabidopsis thaliana 2-xylosyltransferase), suggesting that EOGT and the plant enzyme share common structural features. Analysis of donor substrate specicity revealed that EOGT uses UDP-GlcNAc,

but not UDP-GalNAc, a sugar donor for mucin-type O-glycosylation ( Fig. 2e ). As acceptor substrates, folded EGF domains, but not structurally altered variants of EGF20, served as good substrates. EGF20 GlcFuc (lacking O-glucosylation and O-fucosylation sites) was also GlcNAcylated by EOGT, albeit less effi ciently than wild-type EGF20, whereas EGF20 T34A (a mutation of Th r34, the O-glycosylation site in EGF20, to Ala) was not detectably O-GlcNAcylated ( Fig. 2e and Supplementary Fig. S3 ). Th us, EOGT is an EGF domain O-GlcNAc transferase that specically modies Th r34 of EGF20.

EOGT is involved in OGT-independent O-GlcNAcylation. The hexosamine pathway of glucose metabolism leads to UDP-GlcNAc synthesis 1 . As UDP-GlcNAc serves as a sugar donor for both OGT and EOGT, we investigated whether extracellular O-GlcNA-cylation is linked to OGT-dependent glycosylation. Glutamine: fructose-6-phosphate aminotransferase is a rate-limiting enzyme for glucose entry into the hexosamine pathway. Exogenous GlcNAc can bypass this step, directly entering the hexosamine pathway and being converted to UDP-GlcNAc. Supplementing the medium with GlcNAc markedly increased O-GlcNAc levels on endogenous Notch and bulk intracellular proteins, suggesting that both intracellular and extracellular O-GlcNAcylations are controlled by a common hexosamine pathway ( Fig. 3a ). However, Eogt RNAi nearly eliminated O-GlcNAcylation of secreted N-EGF:FLAG, whereas O-GlcNA cylation of intracellular proteins was unaected. Conversely, Ogt RNAi or inhibition of O-GlcNAcase by PUGNAc aected intracellular, but not extracellular, O-GlcNAc levels ( Fig. 3b ). Th us, OGT and EOGT act selectively and independently to control intracellular and extracellular O-GlcNAcylation, respectively.

EOGT localizes to the lumen of the ER. We next examined whether the distinct roles of OGT and EOGT are derived from their dierent subcellular localization. As expected from its signal peptide and car-

12,000

EDTA (5 mM)

15,000

(kDa)

47.5

[3 H]GlcNAc (d.p.m.)

[3 H]GlcNAc (d.p.m.)

175

600

Activity

(pmole h1 mg1)

Activity

(pmole h1 mg1)

8,000

10,000

83

400

4,000

5,000

200

0 + +

S2/Mock S2/EOGT

Membrane fraction

0 Mock EOGT Mock EOGT

Membrane fraction

Affinitypurifed enzyme

25

15

0 0 20 25

10

5

MnCl2(mM)

400

Activity

(pmole h1 mg1)

K

600

Donor

300

100

0

Activity

(pmole h1 mg1)

EGF20:V5His

Glc Fuc

1000

200

400

800

600

200

400

UDP-Glc UDP-GalNAc

UDP-GlcNAc

200

None

EDTA

Mn

UDP-Xyl

Mg

Ca

0

UDP-Gal

0 Native

Unfolded

T34A

EGF20C7A, C18A

EGF20 C-terminus

Acceptor

EGF20:V5His

Figure 2 | Detection and characterization of O-GlcNAc transferase activity of EOGT. ( a,b) In vitro glycosylation assays. O-GlcNAc transferase activity was measured using membrane fraction proteins or afnity-puried proteins, recombinant EGF20 prepared from yeast and UDP-[ 3H]GlcNAc. (a) S2 cells were transfected with or without a construct expressing EOGT:V5His, and membrane fraction proteins were used for enzyme source. Reactions were performed in either the presence or absence of 5 mM EDTA. ( b) S2 cells were transfected with or without a construct expressing EOGT:V5His, and membrane fraction proteins were used for enzyme source. Where indicated, membrane fraction proteins were further puried with Ni-magnetic beads. CBB staining of the puried enzyme is also shown (right). ( c) Metal requirements of EOGT:V5His. ( d) Effect of increasing concentration of MnCl 2 on

O-GlcNAc transferase activity. ( e) Donor (left) and acceptor (right) specicity of EOGT:V5His. ( ce) Vertical bars represent a range of values obtained from duplicate samples. At least three experiments were performed, which each gave the similar results.

3

NATURE COMMUNICATIONS | 2:583 | DOI: 10.1038/ncomms1591 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1591

Glucose

GlcNAc

Cont

Eogt dsRNA

OGT dsRNA PUGNAc

+ + +

+ +

EOGT

ILSHKEFPRRASENPSKTQRNEL

EOGT RNEL

ILSHKEFPRRASENPSKTQ

(kDa)

EOGT

KDEL signal

460

268 460

268

460

268

117

71

268

238

117

IB: O-GlcNAc (CTD110.6)

IB: Notch

IB: O-GlcNAc (RL2)

IB: Notch

EOGT:V5His

Lysate IP: Notch

V5His

238

117

Medium IP:FLAG

Lysate

IB: O-GlcNAc (CTD110.6)

IB: FLAG (N-EGF:FLAG)

IB: -Tubulin

FLAG:EOGT

FLAG:EOGT RNEL

FLAG

Signal peptide

460 268 238

117

71

55

Lysate

EOGT GFP-KDEL Merged

IB:O-GlcNAc (RL2)

Glucose

Glycolysis

GFAT

ER GlcNAc

GlcNAc

PUGNAc

FLAG:EOGT

FLAG:EOGT

RNEL

IB: -Tubulin

EOGT

EOGT:V5His

Mock

EOGT

EOGT:V5His

EOGT

OGT

Mock

Mock

(kDa)

(kDa)

58

80

58

Medium IP: FLAG

Lysate

IB: FLAG

IB: FLAG

IB: EOGT

IB: V5

Fructose-6-phosphate

UDP-GlcNAc

58

80

58

IB: -Tubulin

Medium

Lysate

O-GlcNAcase

Figure 3 | EOGT is involved in OGT-independent O-GlcNAcylation. ( a) Immunoblot analysis for O-GlcNAcylation of whole-cell lysates (RL-2) or anti-Notch immunoprecipitants (CTD110.6) of Kc cells; the cells were cultured with the indicated supplements. ( b) Immunoblot analysis for O-GlcNAcylation of N-EGF:FLAG isolated from the medium and whole-cell lysates of S2 cells; the cells were treated with indicated dsRNAs or PUGNAc. ( c) Schematicsof EOGT constructs and C-terminal sequences, with proposed KDEL signal boxed. ( d) Confocal images of S2 cells transfected with GFP:KDEL and EOGT. Scale bar, 20 m. (e) S2 cells were transfected with control vector (Mock), constructs expressing FLAG:EOGT or FLAG:EOGT RNEL. Ability of the cellsto retain each construct was analyzed by immunoblotting of cell lysates and FLAG-tag puried secreted proteins. Immunoblot with anti-FLAG antibody revealed the mislocalization of FLAG:EOGT RNEL. (f) S2 cells were transfected with control vector (Mock), or constructs expressing EOGT or EOGT:V5His. Ability of the cells to retain each construct was analyzed by immunoblotting (IB) of cell lysates and secreted proteins using EOGT and V5 antibodies. Note that addition of V5His tag to EOGT precludes detection by EOGT antibody. ( g) Schematic of the EOGT-dependent O-GlcNAcylation pathway. GFAT, glutamine:fructose-6-phosphate aminotransferase;IP, immunoprecipitation.

boxyl-terminal KDEL-like sequence RNEL ( Fig. 3c ), EOGT localized in organelles that co-stained with the ER marker GFP-KDEL ( Fig. 3d ). Expression of a construct lacking the RNEL sequence (EOGT RNEL ) or C-terminally tagged with V5His (EOGT:V5His) resulted in a marked shi in localization from cell lysate to culture medium ( Fig. 3e,f ), suggesting that the KDEL-like sequence serves as an ER retrieval signal. Th us, EOGT is a soluble ER protein that catalyzes OGT-independent O-GlcNAcylation of secreted and membrane glycoproteins ( Fig. 3g ).

Mutation of Eogt causes defects in the aECM. To explore the biological events associated with extracellular O-GlcNAcylation, we generated Drosophila mutants of Eogt (Fig.4aandSupplementary Fig. S4 ). In situ hybridization study revealed that the highest expression of Eogt , likely the maternal contribution, was observed in preblastoderm-stage embryos ( Fig. 4b ). At later stages of embryogenesis, the expression is ubiquitous but the expression level decreases as the stage proceeds ( Fig. 4c e ). Strikingly, although mutants lacking both maternal and zygotic (M / Z) expression of Eogt failed to hatch, but no apparent defect was observed in Notch-dependent neurogenesis ( Fig. 4f ), and the pattern of somatic musculature was unchanged ( Fig. 4g ). However, aer muscle contraction during late stage 17, all segments were compacted and the muscle bands were correspondingly shorter ( Fig. 4h,i ), suggesting that Eogt is essential during embryogenesis, but in a Notch-independent manner.

Th e morphological defects in Eogt M/Z embryos are similar to mutants for Piopio (Pio), a zona pellucida (ZP) protein required for the integrity of the aECM (known as cuticle) 18.Embryoslacking ZP proteins such as Papillote, Dp and Pio show a defect in the formation of the innermost layer of the aECM and its attachment to the epidermis 19. Eogt 7.4 homozygotes and Eogt 7.4/LL00272 transheterozygotes died mostly during second-instar or second / third-instar interface, but some survived until early third-instar. Those Eogt mutant larvae displayed cuticle defect ( Fig. 4j,k ) and irregular tracheal morphology, which appears to correspond to the defects observed in dp mutants ( Fig. 4l ) 19,20. Ultrastructural analysis of larval epidermis revealed disruption of the deposition zone of the endocuticle, leading to separation of the epidermis from the chitin layers (Fig.4m).Th ese results suggest that Eogt contributes to the aECM or its link to the epidermis.

Th e integrity of aECM is also required for the at morphology of adult wings, and deciency in the aforementioned ZP proteins or Eogt results in wing blistering ( Supplementary Fig. S5 ) 19,21. Although deciencies in integrin-dependent epithelial cell contact with basal ECM also causes wing blistering, PS integrin activity and localization were unaected by EOGT removal ( Supplementary Fig. S6a,b ). Eogt does not interact genetically with mys ( PS integrin; Supplementary Fig. S6c ), and muscle detachment, a characteristic loss-of-function phenotype for integrin, was not observed in Eogt embryos (Fig.4gi).Th ese results suggest that EOGT mediates adhesion of the wing epithelium in an integrin function-independent manner.

4

NATURE COMMUNICATIONS | 2:583 | DOI: 10.1038/ncomms1591 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1591

ARTICLE

LL00272

BG00673

WT Eogt

WT

WT

Eogt

WT

ELAV

Eogt

Eogt7.4

WT

Eogt

Eogt7.4

7.4

34.2

*

btl-Gal4 UAS-dpRNAi

Tropomyosin

WT

Eogt

Eogt

7.4/+

7.4/LL00272

Epicuticle Chitin layer

Deposition zone

Epidermal microvilli

F-actin

WT

F-actin

Eogt7.4

Figure 4 | Generation and characterization of

Drosophila Eogt mutants. ( a) Schematic of the Eogt locus. Exons (bars) and the open reading frame (black bars) are depicted. Both Eogt 7.4 and Eogt 34.2 lack sequences corresponding to the translational start codon (ATG 197), signal peptide and a portion of the catalytic domain ( ~ Phe 395 in Eogt 7.4 and ~Thr222 in Eogt 34.2 ) that are highly conserved from Drosophila and mammals. Thus, these two lines are expected to be null alleles. ( be) In situ hybridization analysis of Eogt expression during embryogenesis. ( b) The highest expression of Eogt, likely the maternal contribution, was observed in preblastoderm-stage embryos. ( c), Stage 11; ( d), stage 13; ( e), stage 16 embryos. Scale bars, 50 m. (f) Wild-type (WT) and

Eogt 7.4 embryos lacking maternal and zygotic expression ( Eogt M/Z embryos) were stained for the neuronal marker ELAV. ( g) Stage 16 WT and Eogt M/Z embryos stained for tropomyosin. ( h) Stage 17 WT and Eogt M/Z embryos were stained with phalloidin. ( i) An overlay of the differential interference contrast (DIC) and uorescence image shown in h. (fi) Scale bars, 100 m. (j) Live images of WT and Eogt 7.4 homozygous second-instar larvae.

Enlargement of the boxed area is shown below. In Eogt 7.4 larvae, cuticles were partially detached from the epidermis (arrowheads), although the larval body was not compacted and the body size was comparable with that of the WT larvae. The trachea remains attached to posterior spiracles ( asterisk). Scale bars, 400 m. (k) Cuticle detachment of Eogt 7.4/LL00272 third-instar larvae. Enlargement of the boxed area is shown below. Scale bar, 500 m. (l)

Tracheal morphology of WT, Eogt 7.4 mutant and btl-Gal4 UAS-dp RNAi second-instar larvae. Dorsal view of larvae with similar body length and diameter is shown. Scale bar, 200 m. In the Eogt 7.4 mutant and btl-Gal4 UAS-dp RNAi larvae, the tracheae are distorted relative to the WT larvae (arrowheads). ( m)

Ultrastructural images of WT and Eogt 7.4 homozygous second-instar larvae. Scale bar, 2 m.

EOGT O-GlcNAcylates Dp.AmongtheaforementionedZPproteins, only Dp, a giant 2.5 MDa membrane-anchored extracellular protein, contains EGF-like domains and a very large number of them (308 EGF-like repeats; Fig. 5a,b ) 22. Immunoblotting of cuticle extracts revealed that O-GlcNAcylation occurs almost exclusively on a single high-molecular-weight protein, which was identied as Dp by immunoprecipitation analysis ( Fig. 5c,d ). In contrast, glycosylation of the other O-GlcNAcylated proteins detected in total larval lysates was not impaired in Eogt mutants ( Fig. 5e and Supplementary Fig. S2). Th is supports our proposal that EOGT does not aect OGT-dependent intracellular O-GlcNAcylation. Partial fragments of EGF DPY EGF repeats of Dp expressed in S2 cells demonstrated that EGF94 103 and EGF108 117, but not EGF84 93, are O-GlcNAcylated. Moreover, depletion of EOGT by RNAi yielded a marked decrease in the O-GlcNAc level ( Fig. 5f,g and Supplementary Fig. S2 ). Th e remaining O-GlcNAc signal in EGF94 103 is because of the residual enzyme activity in EOGT-depleted cells as no EOGT homologues are present in Drosophila. Alignment of these sequences with Notch EGF20 suggested that Th r / Ser residues located at the h position aer the h conserved Cys residue are O-GlcNAc modication sites ( Fig. 5b ). Consistently, Ala substitutions of the putative glycosylation sites of EGF94-103 and EGF108-117 ( GlcNAc) abrogated O-GlcNAcylation ( Fig. 5f and Supplementary Fig. S2 ). Although the precise O-GlcNAcylation sites on Dp have not yet been mapped experimentally, these data clearly showed that C 5XXXXT/SG is the modication site for O-GlcNAcylation of EGF domains, and that up to 86 EGF domains could be O-GlcNAcylated ( Supplementary Fig. S7 ). Interestingly, these modication sites are mostly located within the second EGF-like domain of the EGFDPYEGF repeats.

Eogt is required for Dp-dependent cell matrix interactions. Dp is thought to function in maintaining mechanical tension between the apical surface of epithelial cells and the overlying cuticles 22. As

described above, loss of Dp causes defects in the association of the apical surface of the epidermis with the body wall cuticle. However, in other tissues, Dp is not necessary for adhesion of epithelium to the cuticle. In adult wings, inactivation of dp does not aect adhesion between the apical cell surface and the aECM, but instead leads to loss of contact between the basal surfaces of dorsal and ventral epithelium 19 . Furthermore, dp mutation causes overgrowth of the trachea and invagination of the thoracic cuticle at the muscle-attachment sites on the notum (known as vortex and comma phenotypes) possibly because of reduced mechanical tension 22 . However, the molecular mechanisms by which Dp mediates epithelial cell matrix interactions have not previously been investigated.

To assess the functional relevance of biochemical data, the genetic interaction between Eogt and dp was analyzed in the wing. Notably, a single copy of the dp allele markedly increased the penetrance of wing blisters caused by Eogt RNAi in the posterior wing compartment ( Fig. 6a ). Similarly, Eogt and dp interacted to produce vortex and comma phenotypes on the notum ( Fig. 6b ) 23.Collectively, these data suggest that Eogt is involved in Dp-dependent cellmatrix interactions.

To address the eect of O-GlcNAcylation on Dp localization and expression, we analyzed larval tracheae while taking advantage of the relative convenience of isolation of mutant tissues for immunohistochemical and biochemical analyses. In wild-type larval tracheae, the apical localization of Dp overlaps or is closely apposed to the cuticles. In contrast, the cuticle co-localization is markedly compromised in the Eogt mutants ( Fig. 6c ). Immunoblot analysis of total lysates prepared from larval tracheae showed that O-GlcNAcylation of Dp is specically aected in Eogt mutants, while the amount of Dp is not detectably altered ( Fig. 6d ). Accordingly, O-GlcNAc staining of wild-type tracheae detects intracellular O-GlcNAc as well as the apical expression of O-GlcNAc, and the latter is selectively compromised in Eogt mutant cells (Fig. 6e). Furthermore, RNAi-induced downregulation of dp resulted in a

NATURE COMMUNICATIONS | 2:583 | DOI: 10.1038/ncomms1591 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

5

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1591

Signal peptide Proline-rich region

P-F repeats (Thr/Ser-rich region)

ZP domain

Transmembrane domain

EGF and DPY modules

EGF and DPY modules

EGF domain

EGF and DPY modules

64

1 65 259 260 308

O-GlcNAc

Notch EGF20 CSSNPCQHGGTCYDKLNAFSCQCMPGYTGQKC

Dumpy EGF EGF

DPY

EGF84 CEPSPCGPNSRCKATPDGYAACSCLPNFKGAPPV--CQPECVVSSECAPNQACLNQRCTDPCPGICGGGARCEVLNHNPICSCEANFEGDPFVAC EGF85EGF86 CVPSPCGPNSICQIKQNRPV-CSCVANYIGSPPY--CRPECTLSSECPSDKACINEKCQNPCANVCGHNARCTVIAHSAHCSCDEDYEGDAFIGC EGF87EGF88 CYPNPCAENAVCTPYNNAAR-CTCIEPYNGDPYSTGCRPECIYSSECPSSLACIKQHCRDPCTAACGANAECTVVNHLPSCSCTRGFEGNPFDGC EGF89EGF90 CEPNPCGPNSICRSVEGHPT-CSCQVGYFGAPPQ--CRPECVVSSECAQHLSCINQKCMDPCVGTCGFNAKCQVNNHNPICSCPANYEGNPFEQC EGF91EGF92 CLPSPCGSNSICRNVNNRAE-CSCAPGMFGAPPN--CRPECVINQDCPSNRACIRQRCEDPCIGICGFNAVCSTQNHQPKCSCIESFEGDPYTAC EGF93

EGF94 CYPSPCGANAICRVRNGAGS-CSCIQNYFGDPYIN-CRPECVQNSDCPNNRACINMKCRDPCANACGFNAICRVAHHQPVCSCEPHLTGNPLRAC EGF95EGF96 CRPSPCGLFSTCHVVGERPV-CACLPDYMGAPPN--CKPECMTSAECPSDRACINQRCKDPCPGTCGYNARCRCTNHSPICSCYDGYTGDPFHQC EGF97EGF98 CVPSPCGPNSQCQVSSSGAV-CSCVTNYIGRPPG--CRPECSINSECPARMACINARCADPCIGSCGNNALCHVSLHAPVCMCEPGYSGDPFSGC EGF99EGF100 CRPSPCGLNALCEERNQAAA-CKCLPEYFGDPYVE-CRPECVINSDCPRSRACVNQKCVDPCPGMCGHNALCAVFNHAPNCECLPGYTGNPIVGC EGF101EGF102 CQPSPCGLYSNCRPVNGHAV-CSCVPSYIGSPPN--CRPECMSSSECAQDKSCLNERCKDPCPGTCGNNALCRVVNHNPICSCSPGFSGDPFVRC EGF103

EGF108 CRPSPCGPYSQCREVNGHAV-CSCVTNYIGTPPA--CRPECSVSSECAQDRACVNQRCADPCPGTCGNEAICKVTNHNPICSCPAGYSGDPFVRC EGF109EGF110 CVPSPCGRNSQCRVVGETGV-CSCLPNFVGRAPN--CRPECTINTECPANLACINERCQDPCPGSCGFNAFCSVVNHSPICTCDSGYTGDPFAGC EGF111EGF112 CQPSPCGPNAECRERNGAGS-CTCLPEYFGDPYSG-CRPECVVNSDCSRDKSCVNQKCVDPCPGVCGLNAQCRVSNHLPSCSCLAGYTGNPSSAC EGF113EGF114 CRPSPCGPYSQCREVDGHAV-CSCLQGFIGSAPN--CRPECIISSDCAQNLNCQNQKCVDPCPGTCGIEARCQVINHYPACSCAPGFTGDPFNRC EGF115EGF116 CIPSPCGPNSKCLDVRGSPA-CSCLPDYLGRPPN--CRPECLSSADCPANLACVNQRCSNPCIGACGLHSVCTVIKHRPACECVPGYTGDPFSGC EGF117

WT

Eogt

IB: O-GlcNAc

WT

Eogt

(kDa)

Eogt overexpress

- [phone]

- + - +

EGF

EGF8493

EGF94103

]

]

IP: DpIB: O-GlcNAc

IP: Dp IB: Dp

Mock

1,048

1,048

(kDa)

58

46

30

58

46

30

58

46

30

500

251

121

55

IP: V5IB: O-GlcNAc

IP: V5 IB: V5

51

(kDa)

164

71

WT

Eogt

Cuticle extracts

Total lysate

Lectin

Lectin + chitin

Anti-V5

(kDa)

51

IB: O-GlcNAc

500 279

164

121

71

Eogt dsRNA

- 108117

94103

Wild type

GlcNAc

Wild type

GlcNAc

+ - + 108117

94103

]

1,048

1,048

IB: Dp

IB: Dp

IB: Tubulin

62

49.5

32.5

IP: V5IB: O-GlcNAc

IP: V5 IB: V5

55

62

49.5

32.5

Figure 5 | EOGT O-GlcNAcylates Dumpy. ( a) Schematic representation of the structure of Dp. ( b) Partial alignment of EGF DPY EGF repeats of Dp. O-GlcNAcylation sites are shown in red. ( c) Immunoblot analysis of total lysates and cuticle extracts prepared from wild-type (WT) and Eogt 7.4

second-instar larvae. ( d) Immunoblot analysis of anti-Dp immunoprecipitants isolated from cuticle extracts of WT and Eogt 7.4 second-instar larvae.( e) Immunoblot analysis of total lysates prepared from WT and Eogt 7.4 second-instar larvae. O-GlcNAcylation of a high-molecular-weight protein is specically affected in Eogt mutants (arrow). ( f) Immunoblot analysis of Dp fragments and the mutants lacking O-GlcNAcylation sites ( GlcNAc) isolated from the medium of S2 cells. Where indicated, Eogt is exogenously expressed or silenced by dsRNA treatment. ( g) O-GlcNAcylated Dp fragments bindto a plant chitin-binding lectin, wheat germ agglutinin (WGA). Isolated Dp fragments were subjected to lectin blotting with / without chitin hydrolysates. O-GlcNAc on Dp has the ability to associate with WGA, and binding is inhibited by chitin. IB, immunoblotting; IP, immunoprecipitation.

marked decrease in the apical O-GlcNAc staining ( Fig. 6e ). This suggests that Dp is the major O-GlcNAcylated protein in the tracheal cuticles. Taken together, these results suggest that O-GlcNAcylation is required for the correct localization of Dp in the cuticles, but not for apical secretion in larval tracheae.

To further investigate the roles of O-GlcNAcylation for Dp function, we analyzed the embryonic tracheal system in which aECM has an essential role during tracheal tube morphogenesis. In addition to its expected roles in the chitinous matrix, Dp also mediates the cell intercalation that produces unicellular tracheal branches by a mechanism that does not depend on the luminal chitin function 2427 . To determine whether O-GlcNAc is required for Dp function during tracheal branching, the tracheal lumen was visualized with a uorescent probe that labels chitin secreted into the luminal spaces. Analysis of Eogt M/Z embryos with the chitin-binding probe failed to detect the characteristic branch breaks seen in pio or dp mutants (Fig. 6f). Th is lack of requirement for Eogt

in chitin-independent activity of Dp is consistent with the idea that O-GlcNAc is required to modulate Dp function associated with chitin in the aECM.

Although tracheal networks in Eogt M/Z mutants appear largely intact, the tracheal tubes are slightly twisted or bended and dorsal trunks are occasionally broken ( Fig. 6f ). Since minor defects in tracheal tube morphology were also reported in dp mutants 19,we investigated the role of O-GlcNAc in Dp distribution in the tracheal lumen. In Eogt mutants, Dp staining revealed an inhomogeneous pattern in comparison to wild-type tracheal tubes, in which Dp staining is largely merged with that of luminal chitins ( Fig. 6g ). In contrast, the staining of 2A12, an uncharacterized luminal antigen, was not altered in the Eogt mutant (Fig. 6h), conrming that EOGT does not aect the secretion and distribution of unrelated aECM proteins. Th ese results suggest that O-GlcNAc is required for correct targeting of Dp into the chitinous matrix, possibly by mediating interactions with other components of the aECM. Th e precise mech-

6

NATURE COMMUNICATIONS | 2:583 | DOI: 10.1038/ncomms1591 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1591

ARTICLE

Genotype Wing blister

dplv1/+; +/+ 0% (n=101)dpolvR/+; +/+ 0% (n=103)

en-Gal4/+; UAS-EogtIR3/+ 19% (n=201)en-Gal4/dplv1; UAS-EogtIR3/+ 97% (n=89)en-Gal4/dpolvR; UAS-EogtIR3/+ 99% (n=116)en-Gal4/+; UAS-EogtIR1/+ 69% (n=171)en-Gal4/dplv1; UAS-EogtIR1/+ 100% (n=80)en-Gal4/dpolvR; UAS-EogtIR1/+ 100% (n=46)

Eogt

WT

WT

(kDa)

CBP

CBP

2A12

500

251

164 121

71

55

Wild type dpolvR/+

Eogt7.4

Eogt7.4

IB: O-GlcNAc

Dp

CBP Dp

O-GlcNAc

O-GlcNAc Dp

WT

WT

ap-Gal4/+; UAS-Eogt lR3/+

ap-Gal4/dpolvR;

UAS-Eogt lR3/+

1,048

1,048

1,048

Eogt7.4

WT

DIC Dp

Cuticle Dp

CBP 2A12

Eogt 7.4

WT

IB: Dp

IB: Tubulin

IP: Dp IB: Dp

IP: DpIB: O-GlcNAc

btl-Gal4 UAS-dpRNAi

Eogt7.4

Figure 6 | EOGT is involved in Dumpy-dependent cellmatrix interactions. ( a) Frequency of wing blisters counted in animals of indicated genotype.( b) Genetic interaction between dp and Eogt in vortex (arrow) and comma (arrowhead) phenotypes. These phenotypes were observed only in ap-Gal4/dp ovlR ; UAS-Eogt IR3 animals. Scale bar, 500 m. (c) Immunohistochemical analysis for Dp in second-instar tracheae of indicated genotypes. Cuticle autouorescence is shown in green. Arrowheads mark apical membranes. DIC, differential interference contrast image. Scale bar, 3 m. (d) Immunoblot analysis of tracheal lysates and the anti-Dp immunoprecipitants prepared from wild-type (WT) and Eogt 7.4 larvae. ( e) Immunohistochemical analysis for Dp and O-GlcNAc in second-instar tracheae of indicated genotypes. Arrowheads mark apical membranes. Scale bar, 3 m. (f) Stage-16 WT and

Eogt M/Z mutant embryos stained with a uorescent chitin-binding probe (CBP). Note that tracheal tubes of Eogt M/Z mutant are slightly bent or twisted (arrowheads). Scale bars, 100 m. (g) Close-ups of dorsal trunk (corresponding to the boxed region in ( f)) of stage-15 WT and Eogt M/Z embryos stained with anti-Dp antibody (magenta) and chitin-binding probe (green). Scale bar, 10 m. (h) Stage-15 WT and Eogt M/Z embryos were stained with 2A12 (magenta) and chitin-binding probe (green). Data are presented as described in ( g). Scale bar, 10 m. IB, immunoblotting; IP, immunoprecipitation.

anism by which O-GlcNAc mediates cell matrix interactions must await the elucidation of molecular function of Dp in the aECM.

Discussion

Th is study demonstrated that O-GlcNAcylation of protein extracellular domains is catalyzed by a novel O-GlcNAc transferase, EOGT, which localizes in the ER. O-GlcNAcylation is a pivotal intracellular modication that regulates a wide range of cellular functions 2,6. Recently, O-GlcNAc was detected on EGF-like repeats of Notch in cultured cells 9,28 , although the occurrence and biological function of O-GlcNAcylation in vivo were unknown. We have demonstrated, for the rst time, that O-GlcNAc is expressed and functions in the extracellular matrix in animals.

Genetic and biochemical analyses revealed that Eogt aects Dp function in wings, notum and larval tracheae. It was also revealed that EOGT-catalyzed O-GlcNAcylation occurs selectively on Dp in total lysates prepared from second-instar larvae. Th ese results suggest that O-GlcNAcylation of Dp is the causative factor accounting for the defective cell matrix interaction in Eogt phenotypes. However, components of aECM proteins could be altered during developmental stages and distinct molecular mechanisms might underlie the epithelial cell matrix interaction in the cuticles of dierent tissues. Th us, we cannot offi cially exclude the possibility that, under certain circumstances, O-GlcNAcylation occurs on other aECM proteins that would render additional eects on Dp-dependent cell matrix interactions. Nonetheless, O-GlcNAcylation of Dp is specically aected in the larval tracheae mutant for Eogt. We therefore propose that EOGT aects cell matrix interactions by glycosylation of Dp at least in the larval tracheae.

O-GlcNAcylation of EGF domains was initially detected on the 20th EGF domain of Notch, which is simultaneously modied

with other EGF domain-specic modications, namely O-fucose and O-glucose. Similar to Notch, Dp also contains a few putative O-fucose and O-glucose modication sites in addition to a large number of potential O-GlcNAcylation sites ( Supplementary Fig. S7 ). Nonetheless, unlike Eogt mutation, genetic inactivation of Ofut1 or rumi (responsible for O-fucosylation and O-glucosylation) did not produce the wing blistering phenotype 13,29,30. Th us, O-fucosylation and O-glucosylation, if they occur on Dp, would not be essential for Dp-dependent cell adhesion. On the other hand, Notch signalling is impaired in Ofut1 or rumi mutant by aecting the traffi cking, processing and ligand-binding ability of Notch receptors 13,3033. In contrast, our study revealed that O-GlcNAc is dispensable for most Notch receptor functions as Eogt mutants do not exhibit any apparent defects in major Notch-dependent processes, such as embryonic neurogenesis, wing margin formation and wing vein specication. Thus, O-GlcNAcylation and O-fucosylation / O-glucosylation are important for distinct protein functions and developmental processes in Drosophila. Nonetheless, there remains the possibility that these O-glycosylations may have partially redundant roles in Dp and Notch function; these may be revealed by simultaneous inactivation of Eogt with rumi or Ofut1.

In summary, our data revealed a novel biochemical mechanism for O-GlcNAcylation of extracellular protein domains in the secretion pathway and the biological function in cell matrix interactions in Drosophila . Given that Eogt genes are evolutionarily conserved and O-GlcNAcylation motifs are also conserved in several EGF-like domain-containing proteins in mammals, our ndings suggest that aberrant cell matrix or cell cell interactions caused by dysregulated O-GlcNAcylation may contribute to the aetiology of certain diseases related to the hexosamine signalling pathway.

7

NATURE COMMUNICATIONS | 2:583 | DOI: 10.1038/ncomms1591 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1591

Methods

Materials. PUGNAc was obtained from Sigma. Drosophila expression vectors encoding EGF20 (pMTBip-EGF20:V5His), Notch EGF repeat-FLAG fusion proteins (pRmHa-N-EGF:FLAG containing 36 EGF-like repeats), GFP taggedwith KDEL sequence (pMT-GFP:KDEL) and yeast expression vector encoding EGF20 (pKLAC1-EGF20:V5His) were generated in our previous studies 9,14,34. EGF20C-terminus , C-terminal subdomain of EGF20 was obtained from Biogate . EGF20C7A,C18A lacking a single disulphide bond of EGF20 was from GenScript . The following antibodies were used: mouse anti-Notch intracellular domain (C17.9C6; dilution 1:1,000; DSHB ), mouse anti-V5 (dilution 1:5,000; Invitrogen ), mouse anti--tubulin (DM1A; dilution 1:5,000; Sigma ), mouse anti-FLAG (M2; dilution 1:1,000;

Sigma ), mouse RL2 (dilution 1:1,000; Abcam ), mouse 2A12 (dilution 1:200; DSHB ),

rat anti-tropomyosin (MAC141; dilution 1:100; Barbraham Institute ) and mouse anti-O-GlcNAc (CTD110.6; dilution 1:1,000; Abcam ). Polyclonal antibodies raised in rabbit against the peptide CFPRRASENPSKTQRNEL (504 520 amino acids of EOGT) or the peptide CPWQEEPEQPKSNEN (10,755 10,768 amino acids of Dp) were obtained from MBL or GenScript , and used in a 1:500 or 1:1,000 dilution, respectively.

Plasmid constructs. Th e plasmids containing Eogt complementary DNA were obtained from the Drosophila Genomics Resource Center (DGRC; Indiana University) and the coding region (nucleotides 189 1,754) was amplied by PCR and cloned into the BglII and

XhoI sites of pRmHa expression vector (pRmHa-Eogt).

Th e pRmHA-Eogt:V5His vector was constructed, so that the sequence-encoding V5His (LESRGPFEGKPIPNPLLGLDSTRTGHHHHHH) follows aer the C-terminus of EOGT. CG9109 and CG31915 were also obtained from DGRC and cloned into pRmHa vector as described above. Using pRmHa-Eogt as a template, FLAG sequence (DYKDDDDK) was introduced by PCR following the instructions of the KOD-Plus-Mutagenesis Kit ( Toyobo ). Th e FLAG-tag was inserteda er Asp18 (pRmHA-FLAG Asp18-Eogt) or Asp23 (pRmHA-FLAG Asp23-Eogt),so that the FLAG-tag still remained on Eogt a er cleavage of the signal peptide. Both constructs behave similarly in terms of enzyme activity as well as subcellular localization. Hence, we only showed the results of pRmHA-FLAG Asp18-Eogt as a representative. FLAG-Eogt RNEL that lacks C-terminal RNEL sequence was generated by PCR as described above using pRmHa-FLAG-Eogt as a template.

For constructing expression vectors for Dp, the cDNA fragment corresponding to amino acids 9,377 9,893, 9,908 10,433 or 10,66 11,184 of Dp was amplied by PCR and cloned into the BglII/XhoI sites of pMTBip-V5His vector. These vectors were constructed so that the sequence-encoding V5-His follows a er the last amino acid of each protein. For construction of yeast expression vector encoding EGF20 mutants (pKLAC1-EGF20 GlcFuc and pKLAC1-EGF20T34A),

dp olvR and dp lv1 were obtained from the DGRC. Gal4 drivers used were ap-Gal4 (Flybase ID; Fbti0009983), tub-Gal4 (DGRC) and en-Gal4 (A. Brand). Eogt LL00272 contains a modied piggyBac transposon with splice traps 37 within the 5 -UTR, located 87 bp upstream of the Eogt translation start site. Drosophila RNAi lines UAS-Eogt IR1 (9867R-1)

UAS-Eogt IR3 (9867R-3), and UAS-dp IR (15637R-1) were obtained from the National Institute of Genetics (Mishima, Japan).

Drosophila crosses . For generation of clones of cells mutant for Eogt in the wing disc, Eogt 7.4 was crossed with FRT40A by meiotic recombination, and mutant clones were generated by crossing to y w hs-p 122 ; ubi-GFP:nls FRT40A ies and heat shocking to induce ippase-mediated mitotic recombination 38. For generation of marked clones in the adult wing disc, Eogt 7.4 ck FRT40A/CyO was used for ippase-mediated mitotic recombination. To obtain mutants lacking both maternal and zygotic Eogt (Eogt M/Z ), germline clones of Eogt 7.4 were made by crossing hsp; Eogt 7.4 FRT40A/CyO virgins to ovoD FRT40A/CyO males and heat shocking the progeny at 38 C for 1 h on two successive days during the rst and second larval instar. Then,

hs-p; Eogt 7.4 FRT40A/ovoD FRT40A female progeny that contained Eogt 7.4 germline clones were crossed to Eogt 7.4 /CyO; twi-Gal4 UAS-GFP males. All animals were raised at 25 C unless otherwise specied.

Mutation of Eogt . BG00673 contains an insertion of a P transposable element within the 5 -UTR, located 160 bp upstream of the Eogt translation start site. This P-element insertion line is homozygous-viable and fertile, and shows no visible phenotype. Th e deletions of genomic DNA of Eogt 7.4 and

in vitro mutagenesis was performed to mutate the Ser9 / Th r17 or Th r34 to Ala. For the Drosophila pUAST-Eogt construct, the BglII/XhoI fragment of Eogt prepared from pRmHa-Eogt was cloned into those sites of pUAST. Th ree transgenic lines were isolated a er injection into Drosophila. Th ey were assayed with tub-Gal4 drivers at 21C and exhibited similar phenotypes.

Lectin blotting . Dp fragments isolated with agarose beads conjugated with anti-V5 antibody ( Sigma ) were separated by SDS PAGE, electroblotted to a polyvinylidene diuoride membrane and blocked with Tris buered saline plus Tween-20 (TBST)-bovine serum albumin (BSA) (10 mM Tris HCl, pH 7.5, 150 mM NaCl, 0.05 % Tween-20 containing 5 % BSA). Th e membrane was incubated overnight at 4C with 0.5mgml1 biotinylated succinylated wheat germ agglutinin ( Vector ) diluted in TBST-BSA. A er washing four times with TBST-containing 1 M NaCl, the membrane was incubated for 30 min at room temperature with an avidin-HRP conjugate diluted in TBST-BSA. A er washing four times with TBST, the bound lectin was visualized using a chemiluminescent substrate.

Transfection and cell staining . S2 cells were cultured in Sf-900 II medium ( Invitrogen ) supplemented with 5 % fetal bovine serum. Transient transfection was performed using Cellfectin II ( Invitrogen ). Protein expression was under the control of an inducible metallothionein promoter and was induced in Sf-900 II supplemented with 0.7mM CuSO4. S2 cells were untreated or treated with PUGNAc (330 M)

for 3 days, followed by transfection with a construct expressing N-EGF:FLAG. The culture media were puried with anti-FLAG M2 beads and detected by immunoblotting. Kc cells were cultured in Schneider s medium ( Invitrogen ), and the culture media were supplemented or not with 78 mM GlcNAc or glucose and culturedfor 24 h ( Fig. 3a ). Total cell lysates or the immunoprecipitants were separated by SDSPAGE, transferred to Immobilon-P transfer membranes (Millipore), probed with CTD110.6 to detect O-GlcNAc or appropriate primary antibodies, and visualized using HRP-labeled secondary antibodies and a chemiluminescent substrate.

Tissue culture slides (eight chamber; Nunc ) were coated for at least 2 h with 0.5mgml1 of Concanavalin A (from Canavalia ensiformis; Sigma). S2 cells were plated on Concanavalin A-treated slides and allowed to spread out on the slide for 12h (ref. 35). Cells were xed for 10min with 4% formaldehyde (Polysciences Inc. ), permeabilized with PBST-BSA (PBS containing 0.1 % Triton X-100 and 1 % BSA), and incubated with appropriate antibodies in PBST-BSA.

dsRNA preparation and RNAi . RNAi for S2 cells were performed by adding dsRNA to Sf-900 II serum-free medium followed by culturing in Sf-900 II medium

supplemented with 2.5 % fetal bovine serum for additional 4 days as described previously 9,36. For the production of dsRNA for Eogt, transcription templates (nucleotides 197 1,044) were generated by PCR such that they contained T7 promoters on each end of the template. Th e primer sets used for PCR are as follows: T7-Eogt-Fw 5-TAATACGACTCACTATAGGGGCCAATCCTG CCAATACTC-3 ; T7-Eogt-Rv: 5-TAATACGACTCACTATAGGGGCGAGAA GGCCTTAAATGTG-3 , where the T7 sequence is underlined. Puried PCR products were used as the templates for in vitro transcription reactions using a Megascript RNAi kit (Ambion).

Drosophila stocks. BG00673 was obtained from Bloomington stock center.

Eogt LL00272 FRT40A,

Eogt 34.2 were generated by imprecise excision of BG00673. All the mutated alleles result in larval lethality in dierent allelic combinations. Th e lethality associated with homozygous Eogt 7.4 mutations can be rescued by ubiquitous expression of UAS-Eogt under the control of the tub-Gal4 driver.

RNAi in Drosophila . Th e RNAi phenotypes of Eogt were rescued by mouse EOGT expression, excluding the possibility of o-target eects of RNAi. btl-Gal4/UAS-dp IR larvae were raised at 28 C. All other progeny were raised at 25 C unless otherwise specied.

Tissue staining. Embryos and larval tissues were xed in 4 % paraformaldehyde and stained with antibodies as described previously 29. For phalloidin staining, embryos were devitellenized in 80 % ethanol instead of methanol. In situ hybridization in whole-mount Drosophila embryos was carried out using in vitro-transcribed Eogt (nucleotides 1971,044)29. Chitin staining was performed using FITC-conjugated chitin-binding probe (New England Biolabs).

Preparation of larval extracts for immunoprecipitation of Dp. Wild-type or Eogt 7.4 early second-instar larvae were washed twice with PBS, and then homogenized in PBS. A er extraction with a high-salt buer (8.5 % NaCl) containing 1 % dodecyl-maltoside, the insoluble materials containing larval cuticles were dissolved in lysis buer (1 % Nonidet P-40, 0.06 % SDS, 20 mM HEPES, pH 7.5). The lysates were immunoprecipitated at 4 C for 2 h using Protein G Sepharose ( GE Health-care ) and the anti-Dp antibody as described above.

Transmission electron microscopy. Th e second-instar larvae were dissected between the A6 and A7 segments, xed with 4 % paraformaldehyde and 1 % glutaraldehyde in cacodylate buer (pH 7.4) 39, and observed by transmission electron microscopy 40.

Enzyme activity assay . Membrane fraction proteins were prepared as described 34 and further puried with Ni-magnetic beads ( Promega ). O-GlcNAc transferase assay was performed in the glycosylation buer (200 mM HEPES-NaOH, pH 7.0, 1mM MnCl2, 1mgml1 BSA) containing 2 M UDP-[3H]GlcNAc (60Cimmol 1;

ARC), 1g of EGF20-V5His and affi nity-puried EOGT as an enzyme source in a volume of 20 l. A er incubation at 25 C for 30 min, the reaction sample was applied to a LC-18 SPE tube ( Supelco ), washed and eluted with 1 ml of 80 % acetonitrile, 0.052% triuoroacetic acid. Radioactivity was measured by liquid scintillation counting.

References

1. Butkinaree, C., Park, K. & Hart, G. W. O-Linked beta-N-acetylglucosamine (O-GlcNAc): extensive crosstalk with phosphorylation to regulate signaling and transcription in response to nutrients and stress . Biochim. Biophys. Acta. 1800, 96106 (2010).

8

NATURE COMMUNICATIONS | 2:583 | DOI: 10.1038/ncomms1591 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1591

ARTICLE

2. Wang, Z. et al. Extensive crosstalk between O-GlcNAcylation and phosphorylation regulates cytokinesis.Sci. Signal. 3, ra2 (2010).

3. Dentin, R., Hedrick, S., Xie, J., Yates, J. III. & Montminy , M. Hepatic glucose sensing via the CREB coactivator CRTC2 . Science 319, 14021405 (2008).4. Fujiki, R. et al. GlcNAcylation of a histone methyltransferase in retinoic-acid-induced granulopoiesis.Nature 459, 455459 (2009).

5. Yang, X. et al. Phosphoinositide signalling links O-GlcNAc transferase to insulin resistance.Nature 451, 964969 (2008).

6. Hanover, J. A., Krause, M. W. & Love, D. C. Th e hexosamine signaling pathway: O-GlcNAc cycling in feast or famine . Biochim. Biophys. Acta. 1800, 8095 (2010).

7. Kearse, K. P. & Hart, G. W. Topology of O-linked N-acetylglucosamine in murine lymphocytes.Arch. Biochem. Biophys. 290, 543548 (1991).

8. Abeijon, C. & Hirschberg, C. B. Intrinsic membrane glycoproteins with cytosol-oriented sugars in the endoplasmic reticulum . Proc. Natl Acad. Sci. USA 85, 10101014 (1988).

9. Matsuura, A.et al. O-Linked N-acetylglucosamine is present on the extracellular domain of Notch receptors . J. Biol. Chem. 283, 3548695 (2008).

10.Haltiwanger, R. S., Blomberg, M. A. & Hart, G. W. Glycosylation of nuclear and cytoplasmic proteins. Purication and characterization of a uridine diphospho-N-acetylglucosamine:polypeptide beta-N-acetylglucosaminyltransferase.J. Biol. Chem. 267, 90059013 (1992).11. Gambetta, M. C., Oktaba, K. & Muller, J. Essential role of the glycosyltransferase sxc/Ogt in polycomb repression . Science 325, 9396 (2009).

12. Sinclair, D. A. et al. Drosophila O-GlcNAc transferase (OGT) is encoded by the Polycomb group (PcG) gene, super sex combs (sxc ).Proc. Natl Acad. Sci. USA 106, 1342713432 (2009).

13. Acar, M. et al. Rumi is a CAP10 domain glycosyltransferase that modies Notch and is required for Notch signaling . Cell 132, 247258 (2008).

14. Okajima, T., Xu, A., Lei, L. & Irvine, K. D. Chaperone activity of protein O-fucosyltransferase 1 promotes notch receptor folding . Science 307, 15991603 (2005).

15.Luo, Y. & Haltiwanger, R. S. O-Fucosylation of Notch occurs in the endoplasmic reticulum.J. Biol. Chem. 280, 1128911294 (2005).

16. Shoreibah, M. et al. Isolation, characterization, and expression of a cDNA encoding N-acetylglucosaminyltransferase V . J. Biol. Chem. 268, 1538115385 (1993).

17. Zeng, Y. et al. Purication and specicity of beta1,2-xylosyltransferase, an enzyme that contributes to the allergenicity of some plant proteins . J. Biol. Chem. 272, 3134031347 (1997).

18. Prout, M., Damania, Z., Soong, J., Fristrom, D. & Fristrom, J. W. Autosomal mutations aecting adhesion between wing surfaces in Drosophila melanogaster. Genetics 146, 275285 (1997).

19.Bokel, C., Prokop, A. & Brown, N. H. Papillote and Piopio: Drosophila ZP-domain proteins required for cell adhesion to the apical extracellular matrix and microtubule organization.J. Cell Sci. 118, 633642 (2005).

20.Metcalfe, J. A. Development and complementation of lethal mutations at the dumpy locus of Drosophila melanogaster. Genet. Res. 17, 173183 (1971).

21.Brower, D. L. Platelets with wings: the maturation of Drosophila integrin biology.Curr. Opin. Cell Biol. 15, 607613 (2003).

22. Wilkin, M. B. et al. Drosophila dumpy is a gigantic extracellular protein required to maintain tension at epidermal-cuticle attachment sites . Curr. Biol. 10, 559567 (2000).

23.Metcalfe, J. A. Developmental genetics of thoracic abnormalities of dumpy mutants of Drosophila melanogaster . Genetics 65, 627654 (1970).

24.Jazwinska, A., Ribeiro, C. & Aolter, M. Epithelial tube morphogenesis during Drosophila tracheal development requires Piopio, a luminal ZP protein . Nat. Cell Biol. 5, 895901 (2003).

25.Tonning, A.et al. A transient luminal chitinous matrix is required to model epithelial tube diameter in the Drosophila trachea.Dev. Cell 9, 423430 (2005).

26. Devine, W. P. et al. Requirement for chitin biosynthesis in epithelial tube morphogenesis.Proc. Natl Acad. Sci. USA 102, 1701417019 (2005).

27. Swanson, L. E. et al. Drosophila convoluted/dALS is an essential gene required for tracheal tube morphogenesis and apical matrix organization . Genetics 181, 12811290 (2009).

28.Torres, C. R. & Hart, G. W. Topography and polypeptide distribution of terminal N-acetylglucosamine residues on the surfaces of intact lymphocytes. Evidence for O-linked GlcNAc . J. Biol. Chem. 259, 33083317 (1984).

29.Okajima, T. & Irvine, K. D. Regulation of notch signaling by o-linked fucose. Cell 111, 893904 (2002).

30.Sasamura, T.et al. Neurotic , a novel maternal neurogenic gene, encodes an O-fucosyltransferase that is essential for Notch-Delta interactions . Development 130, 47854795 (2003).

31.Bruckner, K., Perez, L., Clausen, H. & Cohen, S. Glycosyltransferase activity of Fringe modulates Notch-Delta interactions . Nature 406, 411415 (2000).

32.Okajima, T., Xu, A. & Irvine, K. D. Modulation of Notch-ligand binding by protein O-fucosyltransferase 1 and Fringe . J. Biol. Chem. 278, 4234042345 (2003).

33.Stanley, P. & Okajima, T. Roles of glycosylation in Notch signaling.Curr. Top Dev. Biol. 92, 131164 (2010).

34.Sakaidani, Y., Furukawa, K. & Okajima, T. O-GlcNAc modication of the extracellular domain of Notch receptors . Methods Enzymol. 480, 355373 (2010).

35. Rogers, S. L., Wiedemann, U., Stuurman, N. & Vale, R. D. Molecular requirements for actin-based lamella formation in Drosophila S2 cells.J. Cell Biol. 162, 10791088 (2003).

36. Kiger, A. A. et al. A functional genomic analysis of cell morphology using RNA interference.J. Biol. 2, 27 (2003).

37. Schuldiner, O. et al. piggyBac-based mosaic screen identies a postmitotic function for cohesin in regulating developmental axon pruning . Dev. Cell 14, 227238 (2008).

38.Xu, T. & Rubin, G. M. Analysis of genetic mosaics in developing and adult Drosophila tissues.Development 117, 12231237 (1993).

39.Bellen, H. J. & Budnik, V. inDrosophila Protocols (eds Ashburner, M., Hawley,S., Sullivan, B.) 175199 (Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 2000).40. Masai, I., Suzuki, E., Yoon, C. S., Kohyama, A. & Hotta, Y. Immunolocalization of Drosophila eye-specic diacylgylcerol kinase, rdgA, which is essential for the maintenance of the photoreceptor . J. Neurobiol. 32, 695706 (1997).

Acknowledgements

We thank R. Haltiwanger, P. Stanley and K. Irvine for comments on the manuscript, and the Developmental Studies Hybridoma Bank, the National Institute of Genetics, G. Hart, K. Basler, F. Schoeck, T. Bunch, and Y. Funakoshi for providing Drosophila strains and reagents. Th is work was supported in part by grants from MEXT and JSPS, the 21st Century COE Program and Global COE Program (to T.O, D.N. and T.M.), Human Frontier Science Program, Naito Foundation, Sumitomo Foundation, Takeda Science Foundation, and Kanae Foundation (to T.O.).

Author contributions

M.I. and K.M. performed the mass spectrometry analysis; Y.S., A.M. and T.O. carried out genetic experiments; Y.S. and T.N. performed biochemical analysis; E.S., Y.S. did EM analysis; T.O., D.N., T.M. and K.F. designed the experiments and analyzed data; and T.O. wrote the manuscript with the help of all authors.

Additional information

Accession codes: Th e nucleotide sequence has been submitted to the DNA databank of Japan with accession number AB675601.

Supplementary Information accompanies this paper at http://www.nature.com/ naturecommunications

Competing nancial interests: Th e authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/ reprintsandpermissions/

How to cite this article: Sakaidani, Y. et al. O-Linked-N-acetylglucosamine on extracellular protein domains mediates epithelial cell matrix interactions. Nat. Commun. 2:583 doi: 10.1038/ncomms1591 (2011).

9

NATURE COMMUNICATIONS | 2:583 | DOI: 10.1038/ncomms1591 | www.nature.com/naturecommunications

2011 Macmillan Publishers Limited. All rights reserved.