mutational and functional analysis of large in a novel cho ...glycobiology/glycoomancc20311.pdf ·...

Glycobiology vol. 19 no. 9 pp. 971–986, 2009doi:10.1093/glycob/cwp074Advance Access publication on May 21, 2009

Mutational and functional analysis of Large in a novel CHO glycosylation mutant

Jennifer T Aguilan2, Subha Sundaram2, Edward Nieves3,and Pamela Stanley1,2

2Department of Cell Biology; and 3Laboratory of Macromolecular Analysisand Proteomics, Albert Einstein College Medicine, New York, NY 10461,USA

Received on November 3, 2008; revised on May 12, 2009; accepted on May15, 2009

Inactivating mutations of Large reduce the functional gly-cosylation of α-dystroglycan (α-DG) and lead to mus-cular dystrophy in mouse and humans. The N-terminaldomain of Large is most similar to UDP-glucose glucosyl-transferases (UGGT), and the C-terminal domain is relatedto the human i blood group transferase β1,3GlcNAcT-1. Theamino acids at conserved motifs DQD+1 and DQD+3 in theUGGT domain are necessary for mammalian UGGT ac-tivity. When the corresponding residues were mutated toAla in mouse Large, α-DG was not functionally glycosy-lated. A similar result was obtained when a DXD motif inthe β1,3GlcNAcT-1 domain was mutated to AIA. Therefore,the first putative glycosyltransferase domain of Large hasproperties of a UGGT and the second of a typical glyco-syltransferase. Co-transfection of Large mutants affectedin the different glycosyltransferase domains did not leadto complementation. While Large mutants were more lo-calized to the endoplasmic reticulum than wild-type Largeor revertants, all mutants were in the Golgi, and only verylow levels of Golgi-localized Large were necessary to gen-erate functional α-DG. When Large was overexpressed inldlD.Lec1 mutant Chinese hamster ovary (CHO) cells whichsynthesize few, if any, mucin O-GalNAc glycans and no com-plex N-glycans, functional α-DG was produced, presumablyby modifying O-mannose glycans. To investigate mucin O-GalNAc glycans as substrates of Large, a new CHO mutantLec15.Lec1 that lacked O-mannose and complex N-glycanswas isolated and characterized. Following transfection withLarge, Lec15.Lec1 cells also generated functionally glycosy-lated α-DG. Thus, Large may act on the O-mannose, com-plex N-glycans and mucin O-GalNAc glycans of α-DG.

Keywords: α-dystroglycan/CHO mutants/DXD/laminin/Large/mutagenesis

Introduction

α-Dystroglycan (α-DG) is the component of the dystrophin gly-coprotein complex that links extracellular matrix proteins in-cluding laminin to the intracellular cytoskeleton to maintainskeletal muscle function (Ibraghimov-Beskrovnaya et al. 1992;

1To whom correspondence should be addressed: Tel: +1-718-430-3346; Fax:+1-718-430-8574; e-mail: [email protected]

Barresi and Campbell 2006). Abnormal glycosylation of α-DGdisrupts binding to laminin and other proteins of the extracellularmatrix resulting in congenital muscular dystrophies termed dys-troglycanopathies (Martin and Freeze 2003; Endo 2007; Mooreand Hewitt 2008). Mutations in glycosyltransferase, putativeglycosyltransferase, or chaperone genes that mediate glycosy-lation of α-DG lead to the dystroglycanopathies termed Walker–Warburg syndrome (Kanoff et al. 1998; Beltran-Valero de Bern-abe et al. 2002; Jimenez-Mallebrera et al. 2003; Schachter et al.2004; van Reeuwijk et al. 2007), muscle–eye–brain disease(Yoshida et al. 2001; Manya et al. 2003), and Fukuyama con-genital muscular dystrophy (Toda et al. 2003). Mutations inthe human LARGE gene are the basis of the dystroglycanopa-thy termed MDC1D (Longman et al. 2003; Brockington et al.2005). The mouse Large ortholog is 97.8% identical to humanLARGE, and a deletion in mouse Large is the basis of the mydmouse (Grewal et al. 2001; Grewal and Hewitt 2002) that de-velops a muscular dystrophy similar to MDC1D (Grewal et al.2005).

The O-mannose glycans implicated in the binding ofα-DG to extracellular matrix proteins including laminin(Ervasti and Campbell 1993; Seifert et al. 2000;McDearmon et al. 2006) are absent or altered in the dystro-glycanopathies. The minimal glycan unit is a tetrasaccharideNeuAcα2,3Galβ1,4GlcNAcβ1,2Man linked to Ser or Thr inthe mucin domain of α-DG (Smalheiser and Kim 1995; Chibaet al. 1997; Smalheiser et al. 1998; Breloy et al. 2008). TheO-linked Man is transferred from Dol-P-Man by protein O-mannosyltransferases POMT1 and POMT2 in the endoplasmicreticulum (ER), and then β1,2GlcNAc is transferred from UDP-GlcNAc by POMGnT1 in the Golgi compartment (Endo 1999,2003; Manya et al. 2004). There are six β1,4GalTs and severalα2,3SiaTs that may transfer Gal and sialic acid, respectively,but to date mutations in these genes have not been causally re-lated to muscular dystrophy. The simple O-mannose glycan isthought to be further extended by LARGE, a putative glyco-syltransferase, because the absence of LARGE leads to hypo-glycosylation of α-DG and loss of α-DG functions in a mannersimilar to deficiencies of POMT1, POMT2, or POMGnT1. Inthe myd mouse which has a deletion in the Large gene, α-DG is hypoglycosylated and does not bind to laminin or themonoclonal antibodies (mAbs) IIH6 or VIA4.1 (Ibraghimov-Beskrovnaya et al. 1992; Grewal et al. 2001; Grewal and Hewitt2002). Similarly, in the human disease MDC1D, α-DG is hy-poglycosylated and does not bind to laminin (Longman et al.2003).

LARGE is a glycoprotein of 756 amino acids and isproposed to be a dual-domain glycosyltransferase. It is a type IItransmembrane protein that is a resident of the Golgi complex(Brockington et al. 2005; Grewal et al. 2005) and containsfour DXD motifs. Based on sequence similarities in the CAZydatabase, CAZy family 8 is related to the N-terminal domain

c© The Author 2009. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: [email protected] 971

at University of G

eorgia Libraries, Serials D

epartment on F

ebruary 15, 2011glycob.oxfordjournals.org

Dow

nloaded from

http://glycob.oxfordjournals.org/

J T Aguilan et al.

of LARGE and CAZy family 49 is related to the C-terminaldomain. The family 8 domain of LARGE is most similar tothe C-terminal portion of the catalytic domain of mammalianUDP-glucose:glycoprotein glucosyltransferases (UGGT)(Patnaik and Stanley 2005) that belong to CAZy family GT24which is related to family GT8. The sequences of UGGTsfrom human and rat are 60–70% identical over a 1500-aa-longstretch, with significant homology in the ∼300-aa-longC-terminal region (Arnold et al. 2000; Tessier et al. 2000).The N-terminal portion of mouse Large is ∼25% identical and40% similar to UGGTs. Moreover, the second DXD domainof mouse Large (DQD at aa 334–336) and the amino acids atDQD+1 and DQD+3 are conserved, and mutational studieshave shown that the latter two amino acids are necessaryfor UGGT glucosyltransferase activity (Arnold et al. 2000;Tessier et al. 2000). Mutation of either of the DXD motifsin the UGGT domain to NNN inactivates Large, although itremains localized to the Golgi (Brockington et al. 2005). TheC-terminal glycosyltransferase domain of Large is similar toβ1,3GlcNAcT-1 (sometimes termed β3GlcNAcT-6; Peyrardet al. 1999; Patnaik and Stanley 2005) which transfers GlcNActo form polylactosamine units (Sasaki et al. 1997). Thefourth DXD domain of mouse Large (DID at aa 563–565)has sequence similarity to, and aligns with, the DVD motifconserved in the family of β3GlcNAcTs. Mutation of this DXDto NNN caused LARGE to localize to the ER, and functionalanalysis was not pursued (Brockington et al. 2005).

Sugars not required for Large to generate functional α-DGthat binds to laminin and to the monoclonal antibodies (mAb)IIH6 and VIA4.1 have been identified by exoglycosidase studies(Combs and Ervasti 2005) and experiments in Chinese hamsterovary (CHO) cell glycosylation mutants. CHO mutants that donot transfer CMP-sialic acid (Lec2) or UDP-Gal (Lec8) into theGolgi, or that synthesize little GDP-Fuc (Lec13) (Patnaik andStanley 2006) nevertheless generate good levels of functionalα-DG in the presence of Large (Patnaik and Stanley 2005).Therefore, none of these is a major sugar transferred by Largenor is any one of them required as a major substrate of Large.Sequential digestion of α-DG from various sources with exogly-cosidases has shown that the binding of mAbs IIH6 and VIA4.1and laminin-1 is enhanced after removal of sialic acid, β-linkedGal or β-linked GlcNAc (Chiba et al. 1997; Combs and Ervasti2005), and therefore these sugars are not required in the glycansof functional α-DG.

Determining the biochemical activities of LARGE is im-portant because overexpression of LARGE in cultured fibrob-lasts from patients with different dystroglycanopathies gener-ates functional α-DG, even in cells from dystroglycanopathiescaused by mutations in genes other than LARGE (Barresi et al.2004). Thus, understanding the properties of LARGE is keynot only to understanding the basis of MDC1D, but also fordesigning therapeutic approaches to ameliorating the effects ofthe dystroglycanopathies in general. CHO mutants have proveduseful in this regard by showing that mouse Large may modifycomplex N-glycans on α-DG (Patnaik and Stanley 2005). MucinO-glycans were also implicated indirectly because overexpres-sion of core 2 GlcNAcT enhanced the functional glycosylationof α-DG by Large (Patnaik and Stanley 2005). Here, we showthat two mutations that specifically inactivate UGGTs also inac-tivate Large, and that the fourth DXD motif of Large is requiredfor activity. We also show that Large generates functionally gly-

Fig. 1. Diagram of Large and mutant constructs. (A) Mouse Large is 756amino acids including a transmembrane domain (TM, aa 12-34), coiled coildomain 1 (CC1, aa 55–90), coiled coil domain 2 (CC2, aa 483–496), and fiveputative N-glycosylation sites (∗). There are four DXD motifs at aa 242–244(DTD, not shown), aa 334–336 (DQD), aa 438–440 (DED, not shown), and aa563–565 (DID). The DQD, DQD+1(I), and DQD+3(N) residues areconserved in the C-terminal domain of UGGTs. The DTD and DQD are alsospaced similarly in Large and UGGTs. The DID at aa 563–565 is conserved inmouse and human β3GlcNAcT-1 (or iGnT1), which initiates extension ofpolylactosamine units. (B) Constructs of Large and Large mutants withMycHis tag (�) in the pIRES-GFP vector.

cosylated α-DG in ldlD.Lec1 CHO cells that express few, ifany, mucin O-GalNAc glycans and no complex N-glycans, andin a new CHO mutant termed Lec15.Lec1, that has few, if any,O-mannose glycans and no complex N-glycans.

Results

Mutational analysis of LargeThere are two domains with similarities to glycosyltransferasesin human and mouse Large (Figure 1A). Of the four DXD motifsin Large, only the second and the fourth are predicted to bewithin a catalytic domain involved in binding a nucleotide sugardonor. Amino acids at +1 and +3 from the second DXD domain(DQD, aa 334–336) are conserved in Large and UGGTs and areknown to be important for UGGT enzymatic activity (Arnoldet al. 2000; Tessier et al. 2000; Arnold and Kaufman 2003).The fourth DXD domain (DID, amino acids 563–565) alignswith a DVD motif in the family of β3GlcNAcTs. Site-directed

972

at University of G


epartment on F


Dow

nloaded from


Mutational analysis and substrates of Large

Fig. 2. Mutations that inactivate Large. (A) Lec8 cells were transiently transfected with the pIRES-eGFP vector (GFP) or wild-type (WT) Large, Large mutants, orrevertants (-R). Lysates were subjected to Western analysis, and blots were cut and probed separately with mAb IIH6, or anti-Myc mAb, or anti-β-DG. The lamininoverlay assay was performed on the same amount of the same cell lysates. (B) Co-transfection of two Large constructs together was performed as shown andlysates analyzed as in (A).

mutagenesis of these motifs was performed (Figure 1B), andthe function and activity of Large mutants were investigated.Because the Large cDNA is very long, confirmation that eachsite-directed mutation was responsible for a change in Largeactivity was obtained by reverting the mutant cDNA to wildtype. Large constructs were transiently transfected into CHOand Lec8 mutant CHO cells to test the ability of Large mutantsto functionally glycosylate α-DG using Western analysis withthe IIH6 mAb, and a laminin overlay assay.

As observed previously (Patnaik and Stanley 2005), over-expression of Large in Lec8 (Figure 2A) or CHO cells (notshown) resulted in functional glycosylation of α-DG detectedby mAb IIH6 as a broad smear from ∼150 kDa to 250 kDa.This signal was absent in Lec8 cells transfected with a greenfluorescent protein (GFP) vector control (Figure 2A). The ex-pression level of Large in different transfectants was equivalentas determined by anti-Myc mAb. Laminin-1 binding was per-formed on the same lysates by an overlay assay and gave similarresults revealing functional glycosylation of α-DG by Large inLec8 (Figure 2A). β-Dystroglycan detected by the monoclonalantibody 43DAG1/8D5 was used as a loading control. Mutationof the fourth DXD motif in the β3GlcNAcT-1 domain from DIDto AID (D563A) gave Large activity comparable to wild-type(WT) Large as determined by IIH6 Western analysis and lamininbinding (Figure 2A). A similar result was obtained for the IIH6signal with the DID to DIA mutant (D565A) (Figure 2A). How-ever, the double mutant DID to AIA showed a dramatic re-duction in the ability to functionally glycosylate Large. Therevertants, AID-R, DIA-R, and AIA-R, exhibited similar activ-ities to wild-type Large based on IIH6 and laminin-1 overlayassays (Figure 2A).

The two amino acids conserved in mammalian UGGTs closeto the second DXD domain were mutated to Ala in mouseLarge DQD+1 (I337A) and DQD+3 (N339A). Both of thesechanges resulted in a very marked reduction in the functionalglycosylation of α-DG (Figure 2A). The corresponding rever-tants (DQD+1-R and DQD+3-R) showed essentially wild-typeLarge activity based on IIH6 Western analysis and laminin-1 overlay assays (Figure 2A). Therefore, Large has properties

predicted for a UGGT. To determine if the two glycosyltrans-ferase domains of Large could complement each other, mutantsaffected in the UGGT domain alone were co-transfected withmutants affected in the β3GlcNAcT-1 domain alone. The co-transfection of constructs DQD+3 (N339A) and AID gave ac-tivity similar to AID alone (Figure 2B). Therefore, the presenceof inactive Large DQD+3 did not significantly reduce the ac-tivity of the AID mutant, nor did the wild-type UGGT domainin the AID mutant rescue the DQD+3 mutant phenotype. Sim-ilarly, co-transfection of the DQD+3 and AIA mutant Largeconstructs failed to show complementation (Figure 2B). How-ever, very weak complementation was observed when the twomutants, DQD+1 (I337A) and AIA, were co-transfected. Nodominant negative effect was observed when the inactive mu-tants were co-transfected with wild-type Large. This providesgood evidence that in order to functionally glycosylate α-DG,both of the glycosyltransferase domains of Large must be activein the same molecule.

Sub-cellular localization of Large mutantsSince the Western analysis of Large mutants showed a reduction(AID and DIA) or loss (AIA, DQD+1, and DQD+3) of Largefunction, it was important to determine their location in thecell. Constructs of wild-type Large or Large mutants were tran-siently transfected into parent CHO cells, and co-localization ex-periments were performed by immunofluorescence microscopy(Figure 3A). Since all constructs were made using the bicistronicpIRES-eGFP vector, diffuse GFP background signals were ob-served in the green channel but did not interfere with identify-ing cells binding Alexafluor-conjugated antibodies (Figure 3A).Using secondary antibodies conjugated to the red AlexaFluor568 dye against Myc-tagged Large, and the nuclear stain DAPI,transfected cells could be clearly distinguished. Rabbit poly-clonal antibodies to TGN 38, a Golgi marker, and protein disul-fide isomerase (PDI), a marker of endoplasmic reticulum (ER),were controls. Immunofluorescent images of ∼100 independenttransfectant cells were visually assessed to determine the sub-cellular localization of Large proteins (Figure 3B).

973

at University of G


epartment on F


Dow

nloaded from


J T Aguilan et al.

Fig. 3. Subcellular localization of Large and Large mutants. (A) Wild-type Large and Large mutants with C-terminal Myc were detected using anti-Myc mAb andsecondary antibody conjugated to AlexaFluor 568 (red). DAPI (blue) was used for nuclear staining. The TGN38 Golgi marker and PDI ER markers were detectedusing secondary Ab conjugated to AlexaFluor488 (green). First row: localization of WT Large and TGN38. Second row: localization of WT Large and PDI. Thirdrow: Large showing enrichment in both the Golgi and ER compartments. Fourth row: Large AIA mutant with reticular-like staining merged with PDI. (B) Percenttransfectants expressing wild-type Large or Large mutants and revertants in Golgi, Golgi and ER (Golgi/ER) or ER (n ≥ 100 for each).

Wild-type Large was localized primarily to the Golgi or bothGolgi and ER (Golgi/ER) (Figure 3A and B), as reported pre-viously (Brockington et al. 2005). All the Large mutants wereexpressed in the Golgi complex, but to a lesser extent thanwild-type Large or Large revertants. Less than 20% of wild-type Large was mainly in the ER, whereas for Large mutants,>20% of cells exhibited ER localization (Figure 3A and B).However, all mutant Large proteins showed substantial Golgiand Golgi/ER staining. In the β1,3GlcNAcT-1 domain Largemutants, about 24% of AIA, 57% of AID, or 79% of DIAtransfectants exhibited Golgi or Golgi/ER localization. For theUGGT domain mutants, about 40% of DQD+1 and DQD+3mutants were localized to the Golgi or Golgi/ER (Figure 3Aand B). A titration experiment showed that this amount of Largeshould be sufficient to functionally glycosylate α-DG (see be-low). Importantly, all UGGT domain revertants were localizedin the Golgi, similar to wild-type Large (Figure 3B). Thus, thelack of activity of the DQD+1 and DQD+3 mutants suggeststhat the first glycosyltransferase domain of Large indeed has aUGGT-like catalytic activity.

To investigate the amount of wild-type Large necessaryto functionally glycosylate α-DG, Large cDNA of the usualamount (4 μg) was titrated to 0.1 μg and transfected into CHO orLec8 cells (Figure 4A and B). Functional glycosylation of α-DGwas observed even when only 0.1 μg Large cDNA was trans-fected. Similar results were obtained for the laminin-1 overlayassay. Therefore, Large mutants introduced using 4 μg cDNAand localized to Golgi or Golgi/ER in about 24% cells wouldgive easily observable IIH6 and laminin signals, if they wereindeed functionally active.

Fig. 4. Titration of Large cDNA. (A) Decreasing amounts of Large cDNAsupplemented to a total of 4 μg cDNA with pIRES-eGFP cDNA weretransfected into CHO, and lysates were analyzed by Western analysis usingmAb IIH6, stripped, and re-probed with anti-Myc mAb. The laminin overlayassay was performed on a second portion of the same cell lysates. (B) Thesame experiment was performed with Lec8 cells.

Glycosylation of α-DG in ldlD.Lec1 CHO cells expressingLargeIt is known that Large modifies complex N-glycans whenoverexpressed in CHO cells (Patnaik and Stanley 2005) and

974

at University of G


epartment on F


Dow

nloaded from



Fig. 5. Effects of Large on α-DG in ldlD.Lec1 cells. (A) VVL binding to blots of cell lysates of ldlD.Lec1 grown in alpha-MEM with 10% dialyzed fetal bovineserum without exogenous sugars for 4 days and Lec8 and CHO grown in alpha-MEM with 10% fetal bovine serum. (B) DSA lectin binding to cell lysates fromCHO, Lec8, and ldlD.Lec1 grown as in (A). (C) PNA binding to cell lysates from Lec2 and ldlD.Lec1 grown in serum-free medium (30 min exposure).(D) Western analysis of cell lysates from ldlD.Lec1 and Lec2 cells transfected with Large or pIRES-GFP vector and grown in serum-free medium. Blots were cutand probed separately with mAbs IIH6, antiMyc, and anti-β-DG. The laminin overlay assay was performed using another portion of the same cell lysates.

suggestive evidence was obtained that Large modifies O-mannose glycans missing from Lec15 cells (Rojek et al. 2007;Patnaik and Stanley 2005). To investigate further, we exam-ined the generation of functionally glycosylated α-DG by Largein the CHO mutant ldlD.Lec1. The ldlD.Lec1 mutant hasa UDP-Gal/UDP-GalNAc-4-epimerase deficiency and conse-quently does not synthesize mucin O-glycans unless galac-tose and N-acetylgalactosamine are available from the medium(Kingsley and Krieger 1984; Kingsley, Kozarsky, Hobbie, et al.1986; Kingsley, Kozarsky, Segal, et al. 1986; Kozarsky et al.1986, 1988). ldlD.Lec1 double mutant cells are also defective inthe Mgat1 gene (Ravdin et al. 1989). It was reported previouslythat Large does not modify the oligomannosyl N-glycans syn-thesized in the absence of Mgat1 (Patnaik and Stanley 2005), andldlD.Lec1 cells synthesize few, if any, mucin O-glycans whengrown in the ITS medium with low concentrations of dialyzedfetal bovine serum (Ravdin et al. 1989) or serum-free medium(Rojek et al. 2006). Under these circumstances, if Large causesfunctional glycosylation of α-DG in ldlD.Lec1 cells, it shouldbe modifying mainly, if not solely, the O-mannose glycans ofα-DG.

ldlD.Lec1 cells were grown in the serum-free (CHO-S-SFMII) medium for 3 days prior to transfection with Largeor pIRES-EGFP control. To confirm the glycans expressed byldlD.Lec1 cells under these conditions, lectin blot analyses wereperformed using Datura stramonium (DSA) or Vicia villosa(VVL) or peanut agglutinin (PNA) (Figure 5A–C). CHO cellsthat express polylactosamine served as a positive control forDSA binding, while Lec8 cells that do not transfer Gal servedas a negative control. On the other hand, Lec8 cells served as a

positive control for VVL binding to terminal GalNAc residuesthat are essentially missing from CHO cells since the addi-tion of Gal to GalNAc blocks VVL binding. In the case ofPNA, Lec2 cells provided a positive control since they do notadd sialic acid to glycoconjugates. ldlD.Lec1 cells showed verylow or no binding to VVL (Figure 5A) or DSA (Figure 5B)or PNA (Figure 5C) as expected. The low signal detected inthe PNA blot was after a >30 min exposure and may reflectnonspecific background binding or O-glycans synthesized fromthe generation of UDP-Gal and UDP-GalNAc by scavengerpathways. The ldlD mutation is not known but in vitro as-says revealed no UDP-Gal/UDP-GalNAc-4-epimerase activity(Kingsley, Kozarsky, Hobbie, et al. 1986). Therefore, ldlD.Lec1cells synthesized very few, if any, core 1 mucin O-GalNAcglycans (T-antigen), or polylactosamine-containing N-glycans.Nevertheless, ldlD.Lec1 cells expressed functionally glycosy-lated α-DG in the absence of exogenous Gal and GalNAc,suggesting that the modification is on O-mannose glycans(Figure 5D). Proteoglycans are not modified by Large(Smalheiser and Kim 1995) but an as yet unidentified glycanacceptor cannot be ruled out.

Isolation of Lec15.Lec1 CHO cellsIndirect evidence has implicated mucin-type O-GalNAc glycansas substrates of Large (Patnaik and Stanley 2005). To obtainfurther evidence, we developed Lec15.Lec1 CHO cells withfew, if any, O-mannose glycans. Lec15 cells have a mutation inthe Dpm2 gene that largely prevents Dol-P-Man synthesis (Stollet al. 1992; Maeda et al. 1998, 2000). C-Mannosylation is greatly

975

at University of G


epartment on F


Dow

nloaded from


J T Aguilan et al.

Table I. Lectin resistance of new isolatesa

L-PHA ConA WGA Ricin

Lec1.3c >200 3 30 0.4Lec15.B4 5 16 2 0.05Colony 1A >200 3 40 0.5Colony 1F >200 3 40 0.4Colony 1O >200 3 40 0.5Colony 2A >200 3 8 0.1Colony 2D >200 3 20 0.2Colony 2J >200 3 20 0.2

aLectin concentrations that killed 90% of the cells are in μg/mL.

Table II. Complementation analysis

Cell lines fused No. WGA-resistant coloniesa

0 2 10Lec1 × Lec15 808 0 0Lec1 × Lec15.1A 928 872 408Lec1 × Lec15.1F 1056 1,176 944Lec1 × Lec15.1O 1360 936 848

aSurviving colonies per 10 cm dish in 0, 2, or 10 μg/mL WGA.

reduced while O-mannosylation is abrogated in Lec15 cells(Maeda and Kinoshita 2008). The Lec15.Lec1 double mutantcan synthesize only oligomannosyl N-glycans due to a mutationin the Mgat1 gene. To obtain this double mutant, Lec15.1 CHOcells (Pu et al. 2003) were selected for resistance to 30 μg/mLPhaseolus vulgaris agglutinin (L-PHA). Resistant colonies thathad also acquired hypersensitivity to concanavalin A (Con A)were analyzed for resistance to plant lectins in comparison withLec1 and Lec15 CHO cells. Colonies 1A, 1F, and 1O were mostsimilar to Lec1 in their pattern of lectin resistance (Table I), aspredicted for a Lec15.Lec1 phenotype.

To determine if these isolates possessed a mutation in theMgat1 gene by genetic complementation analysis, colonies 1A,1O, and 1F (all proline auxotrophs) were fused with Gat−Lec1CHO cells (glycine, adenosine, and thymidine auxotroph) us-ing polyethylene glycol, and hybrids were tested for resistanceto wheat germ agglutinin (WGA). Lec15 × Lec1 hybrids werekilled at <3 μg/mL WGA, similar to parent CHO cells, con-firming that the Lec15 and Lec1 mutations affect different genesand complement as expected. By contrast, when colonies 1A,1F, and 1O were crossed with Gat−Lec1 CHO cells, the hy-brids were resistant to WGA (Table II), suggesting noncomple-mentation and a mutation in the Mgat1 gene. Hence, colonies1A, 1F, and 1O are all Lec15.Lec1 double mutants. Consistentwith this, each putative Lec15.Lec1 mutant had very little orno GlcNAcT-I activity, although they had equivalent β1,4GalTactivity (Table III). Lec15.Lec1.1A showed the least GlcNAcT-Iactivity and was the line chosen for molecular analysis.

RT-PCR of total RNA from Lec15.Lec1.1A cells identified amutation within the coding region of the Mgat1 gene close to theATG. A T-insertion was observed after nucleotide 22 causing aframe shift and introducing a premature stop codon at amino acid41 (121–123 bp). This mutation was confirmed by sequencinggenomic DNA isolated from Lec15.Lec1.1A. Lec1.3C genomicDNA was the same as parental CHO DNA in that region ofthe Mgat1 gene. In addition, two silent or synonymous mu-tations at A21C and A444G were found in the Mgat1 gene ofLec15.Lec1 cDNA. The second silent mutation, but not the first,

Table III. GlcNAcT-I activity of Lec15.Lec1 cells

GlcNAcT-I β1,4GalT-1Cell line (nmol/h/mg protein) (nmol/h/mg protein)

CHO 2.2 13.6Lec1.3c <0.1 15.1Lec15 2.6 16.2Lec15.Lec1.1A <0.1 18.8Lec15.Lec1.1F 0.3 23.8Lec15.Lec1.1O 1.2 22.2

was also found in Lec1.3C previously (Chen and Stanley 2003).In conclusion, a truncated, inactive form of GlcNAcT-I compris-ing 44 N-terminal amino acids is predicted to be translated inLec15.Lec1.1A cells. For simplicity, Lec15.Lec1.1A is termedLec15.Lec1 in this paper.

N- and O-glycans of Lec15.Lec1 CHO cellsThe N- and O-glycans released from Lec1 and Lec15.Lec1glycoproteins by N-glycanase (PNGase F) and chemicalβ-elimination, respectively, were permethylated and ana-lyzed by MALDI-TOF mass spectrometry. Lec15.Lec1 andLec1 had similar N-glycan profiles consistent with oligo-mannosyl N-glycans being predominant, and Man5GlcNAc2being the major N-glycan (supplementary Figure S1).Core 1 O-glycans corresponding to NeuAcHexHexNAc-ol (m/z895) consistent with SA-Gal-GalNAc-ol and the desialylatedform HexHexNAc-ol (m/z 534) consistent with Gal-GalNAc-ol were detected in both Lec15.Lec1 and Lec1 O-glycans(supplementary Figure S1). Interestingly, other possible O-glycan structures (Kang et al. 2008) such as Hex3HexNAc-ol(m/z 942), Hex4HexNAc-ol (m/z 1147), Hex3HexNAcHexNAc-ol (m/z 1187) were also detected. The tetrasaccharide,NeuAcHexHexNAcHex-ol (m/z 1099), consistent with SA-Gal-GlcNAc-Man-ol (Smalheiser et al. 1998) was not de-tected. However, an ion with m/z 738 which may correspondto Hex2HexNAc-ol (Gal-Gal-GalNAc-ol) mucin O-glycan, oran isobaric structure HexHexNAcHex-ol (Gal-GlcNAc-Man-ol)(Smalheiser et al. 1998) was detected in both Lec15.Lec1 andLec1 O-glycans. The latter may arise from the loss of sialic acidfrom NeuAcHexHexNAcHex-ol (SA-Gal-GlcNAc-Man-ol)(Smalheiser et al. 1998; Jang-Lee et al. 2007; Breloy et al. 2008).These possibilities were further investigated by MS/MS andMSn analyses. Mass spectra were interpreted and ions were as-signed using the Glycoworkbench program (Ceroni et al. 2008).

MS/MS analysis of the precursor ion m/z 738 generated B, Y,and C product ions and cross-ring cleavages of X ions. The O-glycans corresponding to product ions m/z 463.2, 445.2, 357.1,316.1, 227.1 were detected in Lec15.Lec1 O-glycans, consis-tent with the mucin O-glycan HexHexHexNAc-ol (Figure 6A).Importantly, B and Z ions of m/z 486.2 and 275.1, arising fromfragmentation of HexHexNAcHex-ol (m/z 738), were not de-tected in Lec15.Lec1 O-glycans (Figure 6A). By contrast, m/z738 from Lec1 O-glycans showed the expected mixture of thetwo isobaric structures with product ions of m/z 621.0, 592.1,520.2, 486.2, 463.2, 445.1, 430.1, 316.1, and 275.1 detected(Figure 6B).

The precursor ion, m/z 738, from Lec15.Lec1 O-glycans wasfurther analyzed using LTQ-ESI-MSn which gave daughter ionsdetectable up to MS6 (supplementary Figure S2). However,fragment ions which gave significant structural information

976

at University of G


epartment on F


Dow

nloaded from



Fig. 6. MS/MS analysis of precursor ion m/z 738 from Lec15.Lec1 and Lec1.(A) MALDI-TOF/TOF analysis of the precursor ion m/z 738 from theMALDI-TOF spectrum of Lec15.Lec1 permethylated O-glycans (seesupplementary Figure S1) with fragment ions assigned usingGlycoworkbench. (B) MALDI-TOF/TOF analysis of the precursor ion m/z 738from the MALDI-TOF spectrum of Lec1 permethylated O-glycans (seesupplementary Figure S1). Fragment ions show a mixture of isobaric structurescorresponding to Hex2HexNAc-ol (e.g., Gal-Gal-GalNAc-ol) mucinO-glycans and HexHexNAcHex-ol (e.g., Gal-GlcNAc-Man-ol). Sialic acid(♦); Gal (�); GlcNAc (�); Man (�).

were detected only up to MS4. Precursor ion, m/z 738, sub-jected to MS2 generated m/z 520.27, 463.27, 445.27, 316.27,284.18 product ions with the assignments shown in supplemen-tary Figure S2. Since m/z 520 ion may correspond to eitherHexHexNAc-ol (Gal-GalNAc-ol) or HexNAcHex-ol (GlcNAc-Man-ol), m/z 520 was selected for MS3. Fragmentation of them/z 520 precursor ion showed more product ions correspondingto a mucin O-GalNAc glycan, including m/z 316 which gave riseto m/z 298 when selected for MS4. In addition, when m/z 463detected from MS2 was subjected to a collision energy of 33%,product ions consistent with mucin O-glycans were obtained(supplementary Figure S2). The combined data suggest that thepredominant O-glycans present in Lec15.Lec1 glycoproteinsare mucin O-glycans. No product ions consistent with m/z 738from Lec15.Lec1 O-glycans being Gal-GlcNAc-Man-ol weredetected (supplementary Figure S2).

Cell surface expression of α-DG in Lec15.Lec1 cellsTo determine whether the glycosylation changes expressed byLec15.Lec1 CHO cells affected α-DG trafficking, cell surfaceexpression of endogenous α-DG was examined using the anti-α-DG mAb 6C1 (Figure 7A). Lec15.Lec1 cells bound 6C1 sim-ilarly to parent CHO and Lec15 cells. Interestingly, Lec1 and

Fig. 7. Cell surface expression of endogenous α-DG and effects of Large onα-DG in Lec15.Lec1 cells. (A) Flow cytometry using primary antibody 6C1anti-cranin (α-DG) and secondary antibody FITC-conjugated to anti-mouseIgG+A+M (black line), secondary Ab alone (gray profile), and primaryantibody alone (dark gray line). (B) Cell lysates of Lec8, Lec15, Lec1, andLec15.Lec1 cells expressing Large or pIRES-GFP vector control and probedusing primary mAb IIH6, anti-Myc, or anti-β-DG followed byHRP-conjugated goat anti-mouse IgG+M Ab. The laminin overlay assay wasperformed using another portion of the same cell lysates. (C) Another portionof the same cell lysates was analyzed before and after treatment with PNGaseF. PNGase F digestion was confirmed by stripping the membrane and probingwith anti-NCAM, and expression levels of Large by stripping the samemembrane again and probing with anti-Myc mAb.

977

at University of G


epartment on F


Dow

nloaded from


J T Aguilan et al.

ldlD.Lec1 cells bound slightly more, and Lec8 cells bound sig-nificantly more 6C1 mAb than parent CHO. This suggests thateither the latter mutants express more α-DG at the cell sur-face or that complex N-glycans and mucin O-glycans inhibit6C1 mAb binding. Nevertheless, it was apparent that any dif-ferences in functional glycosylation of α-DG between Lec15and Lec15.Lec1 cells would not be due to their differential cellsurface expression of α-DG.

Glycosylation of α-DG in Lec15.Lec1 CHO cells expressingLargeLec15.Lec1 cells were transfected with Large or pIRES-GFPas the vector control, and lysates were analyzed by Westernblotting with mAb IIH6 for comparison with Lec1, Lec15, andLec8 lysates (Figure 7B). Laminin-1 overlay assays showedthat IIH6-positive samples also bound laminin. It is apparentthat Large modified α-DG in Lec15.Lec1 cells. Expressionlevels of Large detected by anti-Myc mAb varied somewhatdespite equal loading of the protein based on β-DG levels.This is probably due to small differences in transfection ef-ficiencies. Functionally glycosylated α-DG from Lec15.Lec1was not susceptible to PNGase F treatment, similar to Lec1(Figure 7C), providing evidence that Large modified mainly O-glycans (or potentially some unidentified glycan) of Lec15.Lec1cells. By contrast, α-DG from Lec15 cells was highly suscepti-ble to PNGase F, indicating that Large modified N-glycans to agreater extent than O-glycans on α-DG in Lec15 cells. This wassimilar to α-DG from Lec8 cells which lost most IIH6 bindingactivity after PNGase F treatment (Figure 7C). PNGase F ac-tivity was confirmed by re-probing with an anti-NCAM mAb.The shift in molecular weight after PNGase F was more pro-nounced in N-CAM from Lec15 because the immature N-glycanMan5GlcNAc2-PP-dolichol that accumulates in Lec15 cells isconverted to complex N-glycans (Maeda and Kinoshita 2008)which serve as substrates for polysialic acid transfer.

Further evidence that Large modified mucin O-GalNAc gly-cans on α-DG in Lec15.Lec1 cells was obtained by overexpress-ing Large in the presence of benzyl-O-GalNAc (BzOGalNAc),a competitive inhibitor of mucin O-glycan synthesis. Followingtreatment with 10 mM BzOGalNAc beginning 24 h before trans-fection, a reduction in IIH6 signal was observed in Lec15.Lec1cells and this signal was resistant to removal by N-glycanase(data not shown). The effect of benzyl 2-acetamido-3,4-di-o-acetyl-2-deoxy-α-D-galactopyranoside (AcBzOGalNAc) wasalso assessed because acetylated inhibitors are taken up by cellsmore efficiently (Sarkar et al. 1995, 1997). The reduction infunctional glycosylation of α-DG by Large in Lec15.Lec1 cellstreated with AcBzOGalNAc was marked, and proportionatelymuch greater than for α-DG from treated Lec8 cells (Figure 8A).While a slight decrease in cell surface expression of α-DG wascaused by treatment with AcBzOGalNAc (Figure 8B), this didnot differ between Lec8 and Lec15.Lec1 cells. The combineddata suggest that Large modifies mainly mucin O-GalNAc gly-cans in Lec15.Lec1 cells.

Discussion

In this paper, we provide new insights into the properties ofthe putative glycosyltransferase Large by mutational analysesand by analyzing Large functions in ldlD.Lec1 CHO cells and a

Fig. 8. Effects of AcBenzyl-O-GalNAc on the functional glycosylation ofα-DG by Large. (A) Lec8 and Lec15.Lec1 cells were incubated for 24 h in thepresence or absence of 1 mM AcBzOGalNAc and transfected with Large forthe next 24 h. Western analysis of lysates was performed using mAbs IIH6,anti-Myc, and anti-β-DG. (B) Lec8 and Lec15.Lec1 cells were incubated 24 hin the presence and absence of 1 mM AcBzOGalNAc, transfected with Largeand incubated another 24 h before preparation for flow cytometry. The profilesshow binding to primary antibody 6C1 mAb and secondary antibodyFITC-conjugated to anti-mouse IgG+A+M. Secondary Ab alone (gray).

novel CHO mutant Lec15.Lec1. Brockington et al. (2005) pre-viously showed that changing either of two DXD motifs in theUGGT domain of human LARGE to NNN destroyed the abilityof LARGE to functionally glycosylate α-DG. Interestingly, theintroduction of three Asn residues did not affect the localizationof either LARGE mutant to the Golgi complex. These obser-vations are consistent with both DXD motifs in the N-terminaldomain of LARGE being potential sites of nucleotide sugarbinding (Bourne and Henrissat 2001). Our previous analysisrevealed that the N-terminal portion of mouse Large is similarto the C-terminal region of ER-resident UGGTs (Patnaik andStanley 2005) and contains the conserved DQD as well as aminoacids at positions DQD+1 and DQD+3 known to be necessaryfor UGGT glucosyltransferase activity. In human UGGT, mu-tation of DQD+1 (L) or DQD+3 (N) to Ala reduces activityto background levels, despite appropriate localization to the ER(Arnold et al. 2000; Tessier et al. 2000). Here, we generated thecorresponding mutations in Large and compared their activityto revertants derived from the respective mutant.

Both the Large DQD+1 and DQD+3 mutants were well ex-pressed but IIH6 immunoreactivity and laminin-1 binding toα-DG were dramatically reduced. Revertants had similar activ-ity to wild-type Large and were also predominantly localizedto the Golgi. By contrast, both DQD+1 and DQD+3 mutantLarge were localized more to the ER. Nevertheless, immunoflu-orescence microscopy showed that ∼40% of the DQD+1 and>30% of the DQD+3 Large mutant-expressing cells had Largein the Golgi. This amount of Golgi-localized Large would givea robust IIH6 signal if a mutant was active because even as

978

at University of G


epartment on F


Dow

nloaded from



little as 100 ng (or 2.5%) of the 4 μg Large cDNA was able tofunctionally glycosylate α-DG (Figure 4). However, neither ofthese mutants functionally glycosylated α-DG. Thus, the firstglycosyltransferase domain of Large has properties related toUGGTs and is predicted to transfer α1,3Glc, α1,3GlcNAc, orα1,3GlcUA to α-DG. While members of the UGGT family mayalso catalyze the transfer of α1,3Gal, Large is predicted not totransfer Gal because it functionally glycosylates α-DG in Lec8CHO cells that have no Gal in their glycoconjugates (Patnaikand Stanley 2005). The Large UGGT-like activity may act onα-DG in the ER.

We also mutated the DID at amino acids 563–565 in theβ1,3GlcNAcT-1 domain. Mutation of this domain to NNN inhuman LARGE causes localization to the ER and further anal-ysis was not pursued (Brockington et al. 2005). We found thatmutation of DID to AID or DIA in mouse Large gave IIH6 im-munoreactivity and laminin-1 binding of α-DG. Compared towild type and revertants, the AID and DIA Large mutants were∼55% localized to the Golgi or Golgi/ER. A much more dra-matic effect was achieved by mutating DID to AIA. Large withthe AIA mutations was similar to DQD+1 or DQD+3 mutantsand essentially lacked the ability to functionally glycosylate α-DG. Large/AIA was localized to the Golgi or Golgi/ER in ∼25%of transfectants and thus was well positioned to act on α-DG.The fact that it did not provides good evidence that disruptionof the DID at aa 563–565 leads to the inactivation of Large andis consistent with this DID being required for nucleotide sugarbinding.

Wiggins and Munro (1998) identified a consensus sequencein which the pair of aspartate residues in a functional DXDis flanked by four hydrophobic residues (h) on the N-terminalside with the third being an aromatic residue and a hydropho-bic residue on the C-terminal side (hhhhDxDxh). This motifis found in numerous glycosyltransferases (Busch et al. 1998;Wiggins and Munro 1998). In mouse, human, and dog Large, theflanking aa for DQD at amino acids 334–336 are STLADQDIFand for DID at amino acids 563–565 are MFLSDIDFL, withbold residues conforming to the consensus. DXD motifs maybe required for nucleotide sugar binding and/or for catalysis(Gulberti et al. 2003). For example, the DDD motif in GM2synthase (Li et al. 2001) is required for enzyme activity butis not necessary for UDP-GalNAc binding, unlike the DVD inClostridial cytotoxin which is essential for both donor and sub-strate binding (Busch et al. 1998). Mutation of the spaced pairof invariant Asp residues was required to completely block glu-cosyltransferase activity of the Clostridial cytotoxins unlike forthe α1,3mannosyltransferase of yeast or Fringe which requiresmutation of only one Asp to block enzyme activity (Busch et al.1998; Wiggins and Munro 1998; Munro and Freeman 2000).It is also interesting to note that although glycosaminoglycans(GAGs) are not the primary glycans involved in laminin bindingby α-DG (Smalheiser and Kim 1995), we found that the DDDmotif of GlcAT1 aligns with the DID motif of the β1,3GlcNAcT-1 domain of Large. When the DDD at amino acids 194–196 ofGlcAT1 were mutated one at a time (Gulberti et al. 2003), theADD and EDD mutants produced only ∼15% reduction in ac-tivity whereas DAD, DDA, and double mutants gave ∼85%reduction, suggesting that the Asp at position 194 is not so im-portant for transferase activity. This may also be the case in

Large as the AID and DIA mutants retained substantial activ-ity. It took removal of both Asp residues in the AIA mutant toinactivate Large.

The mutational analyses reported here and those ofBrockington et al. (2005) targeted only one glycosyltransferasedomain of Large at a time. We investigated whether therecould be intermolecular complementation if two mutants ofLarge mutated in separate glycosyltransferase domains wereco-expressed. When the Large mutants DQD+3 (UGGT do-main) and AIA (β1,3GlcNAcT-1 domain) were co-transfected,no functional glycosylation of α-DG was observed. Similarly,when the AID (slightly active) and DQD+3 (inactive) Large cD-NAs were co-transfected, no enhanced IIH6 immunreactivity orlaminin binding were observed. Only very weak complemen-tation was observed when the mutants DQD+1 and AIA wereco-transfected. Therefore, it would appear that the two glyco-syltransferase activities of Large must be present in the samemolecule for endogenous Large to be active. In addition, nodominant negative effect was observed when the inactive mu-tants were co-transfected with wild-type Large.

The second goal of this analysis was to further investigatethe range of substrates that may be acted on by Large usingCHO glycosylation mutants. These mutants provide a qualita-tive but not quantitative approach to identifying glycans thatmay be used by Large. We showed previously (Patnaik andStanley 2005), and in this paper, that Large functionally gly-cosylates the complex N-glycans of α-DG. To investigate theextent to which Large modifies O-mannose glycans on α-DG,Large was overexpressed in the ldlD.Lec1 CHO mutant cell line.This double mutant expresses oligomannosyl N-glycans and O-mannose glycans. It appears to lack mucin glycans due to theloss of UDP-Gal/UDP-GalNAc 4-epimerase activity (Kingsley,Kozarsky, Hobbie, et al. 1986) but it is not established that thisstems from a null mutation. In fact, there may be a very low levelof O-GalNAc glycans in ldlD.Lec1 if the mutation is hypomor-phic or circumvented in other ways. Nevertheless, Large wasexpected to modify mainly the O-mannose glycans on α-DG inldlD.Lec1 cells. When Large was overexpressed in ldlD.Lec1,α-DG was functionally glycosylated albeit at a rather low level.These data suggest that the O-mannose glycans of α-DG maynot be a preferred substrate of overexpressed Large. While thepresence of potentially novel glycan substrates as substrates forLarge cannot be ruled out, only O-mannose, N- and O-GalNAcglycans have been found to date on α-DG in mammals (Manyaet al. 2007) and implicated in mAb IIH6 and laminin-1 bind-ing, which in turn require functional Large. In addition, it isof note that Large is a glycoprotein which carries N-glycans.Thus, changes in N-glycan structure may theoretically affectLarge activity. However, Large expressed in ldlD.Lec1 versusLec15.Lec1, both of which synthesize only oligomannosyl N-glycans and therefore should express Large with equivalent ac-tivity, gives rise to different levels of functionally glycosylatedα-DG (Figure 8), consistent with the mucin O-GalNAc glycansof Lec15.Lec1 being better substrates than the O-mannose gly-cans of ldlD.Lec1.

The O-mannose glycans modified by Large are also necessaryfor LCMV virus to bind to α-DG (Imperiali et al. 2005; Rojeket al. 2007). Evidence for this was obtained using Lec15 CHOmutant cells mutated in the Dpm2 gene (Maeda et al. 1998). The

979

at University of G


epartment on F


Dow

nloaded from


J T Aguilan et al.

Lec15 defect reduces the levels of Dol-P-Man and thus is a hy-pomorphic rather than a null mutation which nevertheless resultsin markedly reduced transfer of N-glycans to protein and inhibitsO-mannose to a greater extent than C-mannose glycan synthesis(Maeda and Kinoshita 2008). In this paper, we used the Lec15.1mutant (Pu et al. 2003) to generate a new mutant, Lec15.Lec1,that allowed us to investigate whether Large modifies mucin O-GalNAc glycans. In-depth characterization of Lec15.Lec1 cellsrevealed that the Mgat1 gene is nonfunctional due to an insertionmutation. Based on MALDI-TOF and ESI mass spectrometryanalyses, Lec15.Lec1 cells synthesize oligomannosyl N-glycanswhich are not modified by Large and mucin O-GalNAc glycansbut no detectable O-mannose glycans. We have shown previ-ously that increased expression of functionally glycosylated α-DG is observed when core 2 N-acetylglucosaminytransferase 1is overexpressed with Large in Lec15 CHO mutant cells (Patnaikand Stanley 2005). Lec15.Lec1 provided a more direct approachto determining if Large modifies O-GalNAc glycans. In fact,the laminin-binding glycans generated by Large on α-DG fromLec15.Lec1 cells were resistant to PNGase F digestion, and it islikely that O-GalNAc is the sugar acted on by Large since Galand sialic acid are not necessary for Large to function (Patnaikand Stanley 2005). Consistent with this is the significant inhi-bition of functional glycosylation of α-DG by AcBzOGalNAc(Figure 8A).

Previous studies have identified glycoforms of α-DG withmucin O-GalNAc glycans that react with VVL and perform dis-tinct functions in skeletal muscle (McDearmon et al. 2001). Sia-lylated core 1 oligosaccharides of α-DG in myotubes or musclecells mediate laminin-induced AChR clustering (McDearmonet al. 2003). Mammalian brain α-DG affinity-purified on VVLwas resolved into populations containing or lacking sulfated glu-curonic acid epitopes which bind preferentially to laminin 10/11unlike skeletal muscle α-DG, suggesting that tissue-specificglycosylation modifies the laminin binding specificity of α-DG (McDearmon et al. 2006). The fact that Large functionswell in Lec1, Lec2, Lec8, Lec13, ldlD.Lec1, and Lec15.Lec1CHO glycosylation mutants suggests that Large transfers a sugar(α1,3Glc, α1,3GlcNAc, α1,3GlcUA) by the UGGT-like domainand β1,3GlcNAc or β1,3GalNAc from the β1,3GlcNAcT-1-likedomain as discussed above, to an N-acetylhexosamine residue(GlcNAc or GalNAc) at the 6 position to initiate the formationof a glycan polymer (Figure 9).

Lec15.Lec1 is a cell line well suited for investigating howLarge bypasses defects in POMT1 or POMT2 or POMGnT1, aspreviously shown in myoblast and fibroblast cells from patientshaving the corresponding dystroglycanopathies (Barresi et al.2004). Thus, the range of α-DG glycoforms from Lec15.Lec1cells overexpressing Large and the intensity of the IIH6 sig-nal were greater compared to α-DG from Lec1 cells express-ing Large, whereas they might be expected to be similar. InLec15.Lec1 cells, there are no O-mannosyl glycans with bulkyGal or sialic acid groups that might cause steric hindrance to O-GalNAc glycans as a substrate of Large in the mucin domain ofα-DG (Figure 9). This may also be the case for Lec8 N-glycanswhich are more strongly acted on by Large than N-glycans inwild-type CHO cells which carry Gal and sialic acid (Patnaikand Stanley 2005 and this paper). This may also explain the dif-ferential binding of the anti-α-DG mAb 6C1 to CHO and Lec8cells (Figure 7A).

Fig. 9. Large modifies different mammalian glycans. The diagram shows amodel of α-DG with the range of glycans it is proposed to carry whensynthesized in CHO cells or the mutants Lec8, ldlD.Lec1, or Lec15.Lec1.Large generates functionally glycosylated α-DG when produced in each ofthese cell lines. In CHO and Lec8 cells, Large-modified N-glycans aresensitive to treatment with PNGase F. Large-modified N-glycans are notsensitive to PNGase F in ldlD.Lec1 or Lec15.Lec1 and thus Large modifiescomplex N-glycans but not oligomannosyl N-glycans. In ldlD.Lec1 cells, themajor substrate of Large is O-mannose glycans because O-GalNAc glycans arenot synthesized and oligomannosyl N-glycans are not a substrate. InLec15.Lec1 cells, the major substrate of Large is O-GalNAc glycans becauseO-mannose glycans are not synthesized, and oligomannosyl N-glycans are nota substrate. We propose that Large transfers either a Glc, GlcNAc, or GlcUA inα1,3-linkage to the hydroxyl at C6 of GlcNAc or GalNAc. The C6 hydroxyl isproposed because Large modifies glycans in which GlcNAc or GalNAc issubstituted at the C2, C3, and/or C4 positions.

Material and methods

Mouse Large cDNA expression constructs and site-directedmutagenesisThe full-length Large mouse cDNA generated previ-ously in pcDNA 3.1/Myc-His(-)B (Patnaik and Stanley2005) was amplified by PCR using primers PS829 (5′-AGACTCGAGGCCACCATGCTGGGAATCTGCAGAGGGAAA, forward, with an XhoI site) and PS830 (5′-CCGGGCCCGGGTCAATGATGATGATGATGATGGTC, reverse, withXmaI, SmaI, and ApaI sites), purified by agarose gel elec-trophoresis, digested with XhoI and XmaI and cloned intopIRES2-EGFP (Clontech Laboratories, Mountain View, CA).This was the template used for subsequent site-directed muta-genesis by PCR using the Quick ChangeTM XL Site-DirectedMutagenesis Kit (Stratagene, La Jolla, CA). Point mutationswere made in the second and fourth DXD domains of Large.Each mutant was subsequently reverted back to wild type bysite-directed mutagenesis of the respective mutant. Mutations

980

at University of G


epartment on F


Dow

nloaded from



were confirmed by sequencing of each strand. The followingprimers were used to generate Large mutants and revertants:

DQD+1 – I > A mutant (DQD+1)

Forward 5′-GCTGACCAGGATGCGTTCAATGCTGTTATCAAACAAAACCCC-3′

Reverse 5′-GGGGTTTTGTTTGATAACAGCATTGAACGCATCCTGGTCAGC-3′

DQD+1 – A > I revertant (DGD+1-R)

Forward 5′-GCTGACCAGGATATTTTCAATGCTGTTATCAAACAAAACCCC-3′

Reverse 5′-GGGGTTTTGTTTGATAACAGCATTGAAAATATCCTGGTCAGC-3′

DQD+3 – N>A mutant (DQD+3)

Forward 5′-GCTGACCAGGATATTTTCGCGGCTGTTATCAAACAAAACCCC-3′

Reverse 5′-GGGGTTTTGTTTGATAACAGCCGCGAAAATATCCTGGTCAGC-3′

DQD+3 – A > N revertant (DQD+3-R)

Forward 5′-GCTGACCAGGATATTTTCAATGCTGTTATCAAACAAAACCCC-3′

Reverse 5′-GGGGTTTTGTTTGATAACAGCATTGAAAATATCCTGGTCAGC-3′

DID to AID mutant (AID)

Forward – 5′-CCCTACATGTTCTTATCTGCAATTGACTTCCTGCCTATG-3′

Reverse – 5′-CATAGGCAGGAAGTCAATTGCAGATAAGAACATGTAGGG-3′

AID to DID Revertant (AID-R)

Forward – 5′-CCCTACATGTTCTTATCTGACATTGACTTCCTGCCTATG-3′

Reverse – 5′-CATAGGGACGAAGTCAATGTCAGATAAGAACATGTAGGG-3′

DID to DIA mutant (DIA)

Forward 5′-CCCTACATGTTCTTATCTGACATTGCATTCCTGCCTATG-3′

Reverse 5′-CATAGGCAGGAATGCAATGTCAGATAAGAACATGTAGGG-3′

DIA to DID revertant (DIA-R)

Forward 5′-CCCTACATGTTCTTATCTGACATTGACTTCCTGCCTATG-3′

Reverse 5′-CATAGGGACGAAGTCAATGTCAGATAAGAACATGTAGGG-3′

DID to AIA mutant (AIA)

Forward 5′-CCCTACATGTTCTTATCTGCAATTGCATTCTGCCTATG-3′

Reverse 5′-CATAGGCAGGAATGCAATTGCAGATAAGAACATGTAGGG-3′

AIA to DID revertant (AIA-R)

Forward 5′-CCCTACATGTTCTTATCTGACATTGACTTCCTGCCTATG-3′

Reverse 5′-CATAGGGACGAAGTCAATGTCAGATAAGAACATGTAGGG-3′

Immunofluorescence microscopyCHO and Lec8 cells growing in suspension in α-MEM contain-ing 10% FBS were plated on poly-L-lysine-coated coverslipsat 1 × 105 cells and incubated for 24 h at 37◦C in 5% CO2.Cells were transfected with Large-Myc cDNA constructs usingLipofectamine 2000. After 16 h at 37◦C in 5% CO2, cells werewashed, fixed with 3% paraformaldehyde, blocked, and semi-permeabilized with bovine serum albumin and saponin (Sigma,St. Louis, MO) in phosphate buffered saline (PBS) with cal-cium and magnesium as described (Chiu et al. 2002; Mukherjeeet al. 2007). One set of duplicate cover slips was double-stainedwith an anti-Myc mAb (1:5) and an anti-TGN38 rabbit poly-clonal antibody (1:200; a gift from Dr. Dennis Shields, AlbertEinstein College of Medicine). A second set of duplicate coverslips was stained with anti-Myc mAb (1:5) and an anti-proteindisulfide isomerase (PDI) rabbit polyclonal antibody (1:200;Stressgen, Ann Arbor, MI). The cells were then incubated withAlexa Fluor 568- (1:150; Molecular Probes, Eugene, OR) or488- (1:100; Invitrogen) conjugated secondary antibodies. DAPI(1:1000) was used for nuclear staining. Cells were mountedusing Fluoromount(tm) antifade (Southern Biotech, Birming-ham, AL). Immunofluorescent images were digitally acquiredon an EclipseTM TE300 microscope (Nikon) using a 60×objective.

Isolation of the CHO double mutant Lec15.Lec1To obtain Lec15.Lec1 double mutant CHO cells, Lec15.1 CHOcells (Pu et al. 2003) were selected for resistance to 30 μg/mLPhaseolus vulgaris agglutinin (L-PHA). Resistant colonies thathad also acquired sensitivity to Concanavalin A (Con A) aroseat a frequency of ∼105. They were analyzed for resistance toplant lectins in comparison with Lec1 and Lec15 cells as de-scribed previously (Stanley et al. 1975; Stanley and Siminovitch1976, 1977). Three colonies with a lectin-resistance phenotypemost resembling Lec1 were selected for complementation anal-ysis and assay of GlcNAcT-I activity using Man5GlcNAc2Asnas a substrate and β4GalT-1 using GlcNAc as a substrate asdescribed previously (Chen et al. 2001; Chen and Stanley2003).

For complementation analysis, Lec15 or Lec15.Lec1 cellswhich carried the Pro− auxotrophic mutation were fused withGat−Lec1 cells. Controls were co-culture of the respective celllines without inducing fusion by polyethylene glycol. Hybridcells were selected in medium lacking proline, glycine, adeno-sine, and thymidine in the presence of 0, 1, 2, and 10 μg/mLwheat germ agglutinin (WGA). Colonies of survivors werecounted after ∼10 days of monolayer culture.

Mgat 1 gene sequencingTotal RNA from 107 Lec15.Lec1.1A or Lec1.3C cells (Chenand Stanley 2003) were prepared using 1 mL of Trizol reagent(Invitrogen, Carlsbad, CA) according to the manufacturer’sprotocol. The RNA was precipitated with 0.5 mL cold iso-propanol on ice for 10 min and centrifuged at 12,000 × gfor 10 min at 4◦C. The RNA was washed with 0.5 mL cold75% EtOH and centrifuged at 5000 × g for 10 min at 4◦C.Aliquots of 1 μg total were treated with 1 U of Dnase (In-vitrogen) in 10 μL of reaction buffer and nuclease-free waterfor 12 min at room temperature (RT); 1 μL 25 mM EDTAwas added, and the mixture was incubated for 10 min at

981

at University of G


epartment on F


Dow

nloaded from


J T Aguilan et al.

65◦C. After cooling, oligodT12−−18 (1 μL) and 10 mM dNTPmix (1 μL) were added; the mixture was heated to 65◦C for5 min and cooled on ice for at least 1 min. Reverse transcriptaseSuperscriptTM III RT (1 μL; Invitrogen) was added to half thetubes in the 5× first strand buffer (4 μL), 0.1M DTT (1 μL),RNaseOUTTM (1 μL) to a final volume of 20 μL. Control tubescontained nuclease-free water in place of RT. The samples wereincubated at 25◦C for 5 min, 50◦C for 50 min, and 70◦C for15 min. Treatment with RNase H (1 μL; Invitrogen) was at37◦C for 20 min. For PCR amplification, degenerate primers(Chen and Stanley 2003) PS150, forward, 5′-CCA AGC TTCCTC CCK GYG GGG GCC AGG, and PS154, reverse, 5′-GGCRGA GCC CAG RAR GGA MAG GCA GGW GCT, were used.Reactions consisted of 1× High Fidelity buffer (Invitrogen),2 mM MgSO4, 10 nmoles dNTP mix, 10 pmoles of each primer,5 U High Fidelity Platinum Taq DNA Polymerase (Invitrogen),2.5 U Pfu DNA Polymerase (Stratagene), cDNA (4 μL) andwater in a 50 μL volume. The hot-start PCR had 40 cycles ofamplification through denaturation at 94◦C for 30 s, annealingat 56◦C for 30 s, and extension at 72◦C for 90 s. For sequencingthe following primers were used:

PS 150 forward (CGAAGCTTCCTCCCCKGYGGGGGCCAGG)

PS 1552 reverse (GCCAGCCAAGGCCAGGAAA)

PS 1154 reverse (CCGCAGTCCTGGCTGACAAT)

PS 1205 reverse (GGCTGGCAGGGTCATCATCATCGATA)

β-Actin was used to assess genomic DNA contamination, andthe quality of the cDNA in a reaction with a 10× PCR reactionbuffer (Invitrogen; 5 μL); 1.5 mM MgCl2 (1.5 μL), 10 nmolesdNTP mix, 10 pmoles each primers PS1135 (5′-GTG GGC CGCTCT AGG CAC CA) and PS1136 (5′-TGG CCT TAG GGT TCAGGG GG), 5 U Nu-taq Taq DNA polymerase, 2 μL cDNA, andwater in a 50 μL volume. Forty cycles of amplification wereperformed through denaturation at 94◦C for 30 s, annealing at55◦C for 30 s, and extension at 72◦C for 50 s.

PCR of genomic DNA (gDNA)Lec15.Lec1 and Lec1.3c cells were grown in monolayer cultureand DNAZOLTM (GibcoBRL Life Technologies, Carlsbad, CA)was used to isolate gDNA from ∼107 cells according to themanufacturer’s instructions. The gDNA was precipitated in 0.5mL 100% ethanol at room temperature for 5–10 min, washedthree times with 1 mL 75% ethanol, air-dried, and dissolved in300 μL 8 mM NaOH. Prior to PCR amplification, the pH ofthe gDNA solution was adjusted to pH 8.4 with 0.1 M HEPES.The same conditions and primers described above were used forPCR amplification of the Mgat1 gene.

Cell culture and transfectionParent CHO cells and glycosylation-defective mutant cell lines,Pro−Lec1.3C, Gat−Lec1.1N (Chen and Stanley 2003), andPro−Lec8.D3 (Oelmann et al. 2001), were maintained in alpha-modified minimal Eagle’s medium (Invitrogen) with 10% fetalbovine serum (Gemini, West Sacramento, CA) in suspensionat 37◦C in 10 mL rollers. ldlD.Lec1 cells (Ravdin et al. 1989)were maintained in monolayer culture using the same medium at37◦C or at 34◦C for Pro−Lec15.B4 (Lec15.1 subtype; Beck et al.

1990; Maeda et al. 1998; Pu et al. 2003) and Lec15.Lec1 cells,in a 5% CO2 incubator. For transient transfection, cells weregrown in monolayer using the conditions mentioned above ex-cept for ldlD.Lec1 cells which were grown in medium contain-ing 10% dialyzed fetal bovine serum (Invitrogen) or serum-freemedium (CHO-S-SFM II; (Gibco) Invitrogen) or alpha minimalEagle’s medium supplemented with ITS (Invitrogen) for at least48 h prior to transfection. Transient transfection was performedon cells grown either on 6-well dishes or 10 cm plates usingLipofectamine 2000 (Invitrogen) or Fugene 6 (Roche AppliedScience, Indianapolis, IN) following manufacturer’s protocols.

Electrophoresis of cell lysates and immunoblottingAt 24–48 h after transfection, cell lysates were prepared using50–110 μL of cell lysis buffer (1× Tris-buffered saline [TBS]pH 7.5, 1.5% Triton-X100 [Sigma], and CompleteTM EDTA-free protease inhibitor cocktail [Roche, Indianapolis, IN]). Forelectrophoresis, approximately 40 μg of protein were loadedon either 4–20% or 7.5% Tris-HCl gradient polyacrylamidegels (BioRad or prepared fresh) at 10–30 mA for 2 h. Proteinswere transferred to polyvinylidene difluoride (PVDF) mem-brane overnight at 50 mA in a transfer buffer containing 5%methanol. For immunoblotting, membranes were blocked with3–5% nonfat dry milk (Carnation) in 1× Tris buffered salinewith 0.05% Tween-20 (TBST), pH 7.5, and incubated in eitherthe same solution or 3% bovine serum albumin (Fraction V,Sigma) solution containing primary antibody for 1–2 h. Mem-branes were washed with 1× TBST for 30 min and incu-bated in 3–5% nonfat dry milk in TBST containing an anti-mouse horseradish peroxidase (HRP)-conjugated secondaryantibody (IgG+IgM; Pierce) for 1 h at 1:10,000–20,000 di-lution. Membranes were washed thoroughly for 30 min, incu-bated in SuperSignalTM West Pico chemiluminiscence reagent(Pierce), and exposed to photographic film (Eastman Kodak orDenville Scientific). The mouse monoclonal antibody anti-α-dystroglycan IIH6 C4 (Upstate Biotechnology-Millipore, Bil-lerica, MA) was used at a 1:1000 dilution. The mouse mon-oclonal antibody anti-Myc AB164 and AB166 (9E10) wasused at 1:200 dilution. Mouse anti-NCAM OB11 (Sigma) wasused at 1:500 dilution. The mouse monoclonal antibody, anti-β-DG (43DAG1/8DG) (Novocastra Laboratories Ltd., Newcastle,UK), was used as a loading control at a 1:200 dilution.

Laminin overlay assayProteins transferred to PVDF membranes were blocked in 5%nonfat dry milk in the laminin binding buffer (LBB; 10 mMethanolamine, 140 mM NaCl, 1 mM MgCl2, 1 mM CaCl2). Af-ter 2 h at room temperature, the membrane was incubated inLBB containing 5 μg/mL EHS Laminin I (Roche) (1:1000) at4◦C overnight. The membrane was washed repeatedly with LBBbefore incubation in rabbit polyclonal anti-Laminin I (Sigma)(1:1500) for 1 h at 4◦C. The membrane was washed thor-oughly incubated in horseradish peroxidase (HRP)-conjugatedanti-rabbit secondary antibody (Zymed) for 1 h at 4◦C. Themembrane was washed, incubated in SuperSignalTM West Pico(Pierce, Rockford, IL) chemiluminisence reagent, and exposedto photographic film (Eastman Kodak or Denville Scientific,Matuchen, NJ).

982

at University of G


epartment on F


Dow

nloaded from



Lectin blot analysisTo determine lectin binding, proteins transferred to PVDF mem-branes were blocked using 3–5% nonfat dry milk in TBST for2 h at room temperature. For lectin binding, membrane wassubsequently incubated in the blocking buffer containing lectinat 10–20 μg/mL. The following lectins were used: biotiny-lated Datura stramonium (DSA), Vicia villosa (VVL) lectins,and Peanut agglutinin (PNA) (Vector Laboratories, Burlingame,CA). For biotinylated lectins, the membrane was washed re-peatedly with TBST and then incubated in a blocking buffer-containing HRP-conjugated streptavidin (Vector Laboratories)(1:10,000) for 1 h at room temperature. The membrane waswashed well, incubated in Super Signal West Pico (Pierce)chemiluminisence reagent, and exposed to photographic film(Eastman Kodak or Denville Scientific).

Release of N- and O-glycansApproximately 0.5 to 1 × 108 cells were lysed using 0.5%Triton-X 100, centrifuged at 8000 × g, and the supernatant col-lected. Approximately 2 mg lysate was reduced and denaturedwith β-mercaptoethanol and 10% SDS in 300 μL of sodiumphosphate buffer, pH 7.2, at 95◦C for 10 min. The N-glycanswere released by adding 500 U of PNGase F (New EnglandBiolabs, Ipswich, MA) and incubated at 37◦C overnight. An ad-ditional 500 U was added and digestion was continued for 6–8 h.N-Deglycosylated proteins were precipitated by adding 900 μLof cold 67% ethanol, stored at −20◦C for 16–18 h, and cen-trifuged at 14,000 rpm for 10 min. The pellet was washed twicewith 300 μL of cold ethanol, and supernatant was collected,pooled, and evaporated in a speed vac. The dried material wasresuspended in 13 mM acetic acid and incubated at room tem-perature for 2 h and evaporated in a speed-vac. To further removedetergent contaminants, 10 ethyl acetate extractions were per-formed (Yeung et al. 2008). The material was redissolved in 100μL of 100 mM sodium phosphate buffer, pH 7.1, and loaded on a400 μL bed volume column of Extracti-gel (Thermo-Scientific)pre-washed four times with the same buffer. The column waswashed three times and eluted with 3 volumes (400 μL) of thesame buffer and loaded onto a PGC column (Thermo-Scientific,Franklin, MA) pre-conditioned with 80% acetonitrile in 0.1%trifluoroacetic acid (TFA; Pierce) and water. The column wasfurther washed with 20 mL of Chrommosolve (HPLC gradewater, Sigma), eluted with 3 mL 50% acetonitrile in 0.1% TFA,and concentrated using a speed-vac.

To release O-glycans, the N-deglycosylated protein was re-suspended in 200 μL 0.05 M NaOH/1 M NaBH4 and incubatedat 45◦C for 16 h. Processing of the O-glycans was performedaccording to the methods of Morelle and Michalski (2007) andfurther purified by the method described above for N-glycans.Released N- and O-glycans were permethylated according tothe methods of Kang et al. (2005, 2007, 2008).

Mass spectrometry analysesThe permethylated N- and O-glycans were redissolved in 10μL of 2.5 mM sodium acetate in 50% methanol, and 1 μLof sample was spotted on a MALDI-target plate (AppliedBiosystems, Foster City, CA) followed by 1 μL of matrixand dried under vacuum. Matrices used were dihydroben-zoic acid (DHB; Fluka, St. Louis, MO) at 10 mg/mL in 1mM sodium acetate in 50% methanol or D-arabinosazone

(3 mg/mL) in 75% ethanol (Chen et al. 1997). Profiling ofN- and O-glycans was performed using an Applied BiosystemsVoyager DE-STR MALDI-TOF mass spectrometer, and massspectra were acquired in the reflector and positive ion mode.For MS/MS analysis, samples were prepared in the same way,the Applied Biosystems 4800 Proteomics Analyzer (AB4800)MALDI-TOF/TOF instrument was used and mass spectra wereacquired in the reflector and positive ion mode. Prior to MSn

analysis, samples were resuspended in 1 mM NaOH in 50%methanol and introduced by direct infusion into a linear ion trapmass spectrometer, LTQ-ESI-MS (Thermo Finnigan), at a flowrate of 0.75 μL per min. MSn analysis was performed in thepositive ion mode (Aoki et al. 2008). For MSn analyses, 10%,25%, and 33% collision energy (CE) were applied for fragmen-tation by collision-induced dissociation (CID) of the selectedprecursor ions. Interpretation of mass spectra was performedusing Glycoworkbench (Ceroni et al. 2008).

Flow cytometry analysisCells were plated in 10 cm plates with alpha-modified mini-mal Eagle’s medium (Invitrogen) with 10% fetal bovine serum(Gemini) except for ldlD.Lec1 which was grown in serum-free(CHO-S-SFM II) medium ((Gibco) Invitrogen) and incubated at37◦C for 24 h. After washing with 10 mL phosphate buffer saline(PBS), cells were detached using 5 mL of enzyme free cell dis-sociation buffer, PBS-based (Chemicon, Temecula, CA). Cellswere resuspended, counted, and 5 × 105 cells used for flow cy-tometry analysis. After centrifugation, the cell pellet was washedtwice with the FACS binding buffer (PBS with 1% bovineserum albumin (BSA) and 0.05% Na azide) and incubatedwith a 6C1 anti-cranin (dystroglycan) monoclonal antibody(Chemicon, Millipore) in 0.1 mL of PBS with 1% BSA inPBS for 20 min at 4◦C. The cells were washed twice with theFACS binding buffer and incubated with 0.1 mL of FACS bind-ing buffer containing FITC-conjugated anti-mouse IgG+A+M(Zymed). After 20 min at 4◦C, the cells were washed twice and7-amino actinomycin (7-AAD; BD Pharminogen, San Digeo,CA) was added; the cells were resuspended in 0.5 mL PBS/1%BSA and subjected to flow cytometric analysis using a FACScanor FACSCalibur (BD Biosciences, Rockville, MD) instrument.Raw data were analyzed using FlowJo software (Tree Star Inc.,Ashland, OR) after 7-AAD-positive cells were gated out.

Treatment with benzyl-O-galNAc and acetylatedbenzyl-O-galNAcCells were plated at 1.7 × 106 cells per 10 cm plate in α-MEMcontaining 10% FBS and incubated for 24 h at 37◦C. BzO-GalNAc (2 mM or 10 mM) (Sigma) was added at least 3 hprior to transfection. In another experiment, 1 mM acetylatedbenzyl galNAc (Carbosynth, San Francisco, CA) dissolved in0.5% dimethylsulfoxide (DMSO) was added 24 h before trans-fection. Transient transfection with Large cDNA (6 μg) wasperformed using Fugene 6TM (Roche Applied Sciences) for 16 hfollowing the manufacturer’s protocol. Cells were lysed, proteinconcentration was determined, and cell lysates were analyzedby Western blot using the antibodies mentioned above. Alter-natively, cells were washed and prepared for flow cytometryfollowing the method described above. The primary antibodyused was 6C1 monoclonal antibody (Chemicon, Millipore) and

983

at University of G


epartment on F


Dow

nloaded from


J T Aguilan et al.

the secondary antibody used was anti-mouse (R-Phycoerythrin)PE (Jackson Immuno Research, West Grove, PA).

Supplementary Data

Supplementary data for this article is available online athttp://glycob.oxfordjournals.org/.

Funding

The National Cancer Institute (NIH RO1 36434 to P.S.); theMuscular Dystrophy Association (J.T.A.); the Albert EinsteinCancer Center (PO1 13330).

Acknowledgements

The authors wish to thank Dr. Santosh Patnaik for his con-tributions in the planning of the mutagenesis experiments andDrs. Reto Muller, Shaeri Mukherjee, and Dennis Shields fortheir assistance and helpful advice.

Conflict of interest statement

None declared.

Abbreviations

α-DG, α-dystroglycan; BSA, bovine serum albumin; CE, col-lision energy; CHO, Chinese hamster ovary; Con A, Con-canavalin A; DSA, Datura stramonium agglutinin; FBS, fetalbovine serum; GFP, green fluorescent protein; GlcAT1, glu-curonic acid transferase; L-PHA, Phaseolus vulgaris; PBS,phosphate buffered saline; PDI, protein disulfide isomerase;PNA, peanut agglutinin; PNGase F, N-glycansase or peptide-N-glycosidase F; PVDF, polyvinylidine difluoride; RT-PCR, re-verse transcriptase polymerase chain reaction; TBS, tris buffersaline; TBST, tris buffer saline with Tween 20; VVL, Viciavillosa lectin; WGA, wheat germ agglutinin.

References

Aoki K, Porterfield M, Lee SS, Dong B, Nguyen K, McGlamry KH, TiemeyerM. 2008. The diversity of O-linked glycans expressed during Drosophilamelanogaster development reflects stage- and tissue-specific requirementsfor cell signaling. J Biol Chem. 283:30385–30400.

Arnold SM, Fessler LI, Fessler JH, Kaufman RJ. 2000. Two homologues encod-ing human UDP-glucose:glycoprotein glucosyltransferase differ in mRNAexpression and enzymatic activity. Biochemistry. 39:2149–2163.

Arnold SM, Kaufman RJ. 2003. The noncatalytic portion of human UDP-glucose: Glycoprotein glucosyltransferase I confers UDP-glucose bindingand transferase function to the catalytic domain. J Biol Chem. 278:43320–43328.

Barresi R, Campbell KP. 2006. Dystroglycan: From biosynthesis to pathogen-esis of human disease. J Cell Sci. 119:199–207.

Barresi R, Michele DE, Kanagawa M, Harper HA, Dovico SA, Satz JS, MooreSA, Zhang W, Schachter H, Dumanski JP et al. 2004. LARGE can function-ally bypass alpha-dystroglycan glycosylation defects in distinct congenitalmuscular dystrophies. Nat Med. 10:696–703.

Beck PJ, Gething MJ, Sambrook J, Lehrman MA. 1990. Complementing mutantalleles define three loci involved in mannosylation of Man5-GlcNAc2-P-P-dolichol in Chinese hamster ovary cells. Somat Cell Mol Genet. 16:539–548.

Beltran-Valero de Bernabe D, Currier S, Steinbrecher A, Celli J, van BeusekomE, Van Der Zwaag B, Kayserili H, Merlini L, Chitayat D, Dobyns WB et al.2002. Mutations in the O-mannosyltransferase gene POMT1 give rise to thesevere neuronal migration disorder Walker–Warburg syndrome. Am J HumGenet. 71:1033–1043.

Bourne Y, Henrissat B. 2001. Glycoside hydrolases and glycosyltrans-ferases: Families and functional modules. Curr Opin Struct Biol. 11:593–600.

Breloy I, Schwientek T, Gries B, Razawi H, Macht M, Albers C, Hanisch FG.2008. Initiation of mammalian O-mannosylation in vivo is independent ofa consensus sequence and controlled by peptide regions within and up-stream of the alpha-dystroglycan mucin domain. J Biol Chem. 283:18832–18840.

Brockington M, Torelli S, Prandini P, Boito C, Dolatshad NF, Longman C,Brown SC, Muntoni F. 2005. Localization and functional analysis of theLARGE family of glycosyltransferases: Significance for muscular dystro-phy. Hum Mol Genet. 14:657–665.

Busch C, Hofmann F, Selzer J, Munro S, Jeckel D, Aktories K. 1998. A commonmotif of eukaryotic glycosyltransferases is essential for the enzyme activityof large clostridial cytotoxins. J Biol Chem. 273:19566–19572.

Ceroni A, Maass K, Geyer H, Geyer R, Dell A, Haslam SM. 2008. GlycoWork-bench: A tool for the computer-assisted annotation of mass spectra of gly-cans. J Proteome Res. 7:1650–1659.

Chen P, Baker AG, Novotny MV. 1997. The use of osazones as matrices forthe matrix-assisted laser desorption/ionization mass spectrometry of carbo-hydrates. Anal Biochem. 244:144–151.

Chen W, Stanley P. 2003. Five Lec1 CHO cell mutants have distinct Mgat1gene mutations that encode truncated N-acetylglucosaminyltransferase I.Glycobiology. 13:43–50.

Chen W, Unligil UM, Rini JM, Stanley P. 2001. Independent Lec1ACHO glycosylation mutants arise from point mutations in N-acetylglucosaminyltransferase I that reduce affinity for both substrates.Molecular consequences based on the crystal structure of GlcNAc-TI. Bio-chemistry. 40:8765–8772.

Chiba A, Matsumura K, Yamada H, Inazu T, Shimizu T, Kusunoki S, KanazawaI, Kobata A, Endo T. 1997. Structures of sialylated O-linked oligosaccha-rides of bovine peripheral nerve alpha-dystroglycan. The role of a novelO-mannosyl-type oligosaccharide in the binding of alpha-dystroglycan withlaminin. J Biol Chem. 272:2156–2162.

Chiu R, Novikov L, Mukherjee S, Shields D. 2002. A caspase cleavage fragmentof p115 induces fragmentation of the Golgi apparatus and apoptosis. J CellBiol. 159:637–648.

Combs AC, Ervasti JM. 2005. Enhanced laminin binding by alpha-dystroglycanafter enzymatic deglycosylation. Biochem J. 390:303–309.

Endo T. 1999. O-Mannosyl glycans in mammals. Biochim Biophys Acta.1473:237–246.

Endo T. 2003. Finding of O-mannosyl glycan in mammals and congenital mus-cular dystrophies due to glycosylation defects. Yakugaku Zasshi. 123:825–835.

Endo T. 2007. Dystroglycan glycosylation and its role in alpha-dystroglycanopathies. Acta Myol. 26:165–170.

Ervasti JM, Campbell KP. 1993. A role for the dystrophin–glycoprotein complexas a transmembrane linker between laminin and actin. J Cell Biol. 122:809–823.

Grewal PK, Hewitt JE. 2002. Mutation of Large, which encodes a putativeglycosyltransferase, in an animal model of muscular dystrophy. BiochimBiophys Acta. 1573:216–224.

Grewal PK, Holzfeind PJ, Bittner RE, Hewitt JE. 2001. Mutant glycosyltrans-ferase and altered glycosylation of alpha-dystroglycan in the myodystrophymouse. Nat Genet. 28:151–154.

Grewal PK, McLaughlan JM, Moore CJ, Browning CA, Hewitt JE. 2005. Char-acterization of the LARGE family of putative glycosyltransferases associ-ated with dystroglycanopathies. Glycobiology. 15:912–923.

Gulberti S, Fournel-Gigleux S, Mulliert G, Aubry A, Netter P, Magdalou J,Ouzzine M. 2003. The functional glycosyltransferase signature sequence ofthe human beta 1,3-glucuronosyltransferase is a XDD motif. J Biol Chem.278:32219–32226.

Ibraghimov-Beskrovnaya O, Ervasti JM, Leveille CJ, Slaughter CA, SernettSW, Campbell KP. 1992. Primary structure of dystrophin-associated gly-coproteins linking dystrophin to the extracellular matrix. Nature. 355:696–702.

Imperiali M, Thoma C, Pavoni E, Brancaccio A, Callewaert N, Oxenius A.2005. O-Mannosylation of alpha-dystroglycan is essential for lymphocyticchoriomeningitis virus receptor function. J Virol. 79:14297–14308.

984

at University of G


epartment on F


Dow

nloaded from



Jang-Lee J, Curwen RS, Ashton PD, Tissot B, Mathieson W, Panico M, Dell A,Wilson RA, Haslam SM. 2007. Glycomics analysis of Schistosoma mansoniegg and cercarial secretions. Mol Cell Proteomics. 6:1485–1499.

Jimenez-Mallebrera C, Torelli S, Brown SC, Feng L, Brockington M, SewryCA, Beltran-Valero De Bernabe D, Muntoni F. 2003. Profound skeletalmuscle depletion of alpha-dystroglycan in Walker–Warburg syndrome. EurJ Paediatr Neurol. 7:129–137.

Kang P, Mechref Y, Klouckova I, Novotny MV. 2005. Solid-phase permethy-lation of glycans for mass spectrometric analysis. Rapid Commun MassSpectrom. 19:3421–3428.

Kang P, Mechref Y, Kyselova Z, Goetz JA, Novotny MV. 2007. Comparativeglycomic mapping through quantitative permethylation and stable-isotopelabeling. Anal Chem. 79:6064–6073.

Kang P, Mechref Y, Novotny MV. 2008. High-throughput solid-phase per-methylation of glycans prior to mass spectrometry. Rapid Commun MassSpectrom. 22:721–734.

Kanoff RJ, Curless RG, Petito C, Falcone S, Siatkowski RM, Pegoraro E.1998. Walker–Warburg syndrome: Neurologic features and muscle mem-brane structure. Pediatr Neurol. 18:76–80.

Kingsley DM, Kozarsky KF, Hobbie L, Krieger M. 1986. Reversible defects inO-linked glycosylation and LDL receptor expression in a UDP-Gal/UDP-GalNAc 4-epimerase deficient mutant. Cell. 44:749–759.

Kingsley DM, Kozarsky KF, Segal M, Krieger M. 1986. Three types of lowdensity lipoprotein receptor-deficient mutant have pleiotropic defects in thesynthesis of N-linked, O-linked, and lipid-linked carbohydrate chains. J CellBiol. 102:1576–1585.

Kingsley DM, Krieger M. 1984. Receptor-mediated endocytosis of low den-sity lipoprotein: Somatic cell mutants define multiple genes required forexpression of surface-receptor activity. Proc Natl Acad Sci USA. 81:5454–5458.

Kozarsky K, Kingsley D, Krieger M. 1988. Use of a mutant cell line to study thekinetics and function of O-linked glycosylation of low density lipoproteinreceptors. Proc Natl Acad Sci USA. 85:4335–4339.

Kozarsky KF, Brush HA, Krieger M. 1986. Unusual forms of low densitylipoprotein receptors in hamster cell mutants with defects in the receptorstructural gene. J Cell Biol. 102:1567–1575.

Li J, Rancour DM, Allende ML, Worth CA, Darling DS, Gilbert JB, MenonAK, Young WW Jr. 2001. The DXD motif is required for GM2 synthaseactivity but is not critical for nucleotide binding. Glycobiology. 11:217–229.

Longman C, Brockington M, Torelli S, Jimenez-Mallebrera C, Kennedy C,Khalil N, Feng L, Saran RK, Voit T, Merlini L et al. 2003. Mutations in thehuman LARGE gene cause MDC1D, a novel form of congenital musculardystrophy with severe mental retardation and abnormal glycosylation ofalpha-dystroglycan. Hum Mol Genet. 12:2853–2861.

Maeda Y, Kinoshita T. 2008. Dolichol-phosphate mannose synthase: Structure,function and regulation. Biochim Biophys Acta. 1780:861–868.

Maeda Y, Tanaka S, Hino J, Kangawa K, Kinoshita T. 2000. Human dolichol-phosphate-mannose synthase consists of three subunits, DPM1, DPM2, andDPM3. Embo J. 19:2475–2482.

Maeda Y, Tomita S, Watanabe R, Ohishi K, Kinoshita T. 1998. DPM2 regulatesbiosynthesis of dolichol phosphate-mannose in mammalian cells: Correctsubcellular localization and stabilization of DPM1, and binding of dolicholphosphate. Embo J. 17:4920–4929.

Manya H, Chiba A, Yoshida A, Wang X, Chiba Y, Jigami Y, Margolis RU, EndoT. 2004. Demonstration of mammalian protein O-mannosyltransferase ac-tivity: Coexpression of POMT1 and POMT2 required for enzymatic activity.Proc Natl Acad Sci USA. 101:500–505.

Manya H, Sakai K, Kobayashi K, Taniguchi K, Kawakita M, Toda T, Endo T.2003. Loss-of-function of an N-acetylglucosaminyltransferase, POMGnT1,in muscle–eye–brain disease. Biochem Biophys Res Commun. 306:93–97.

Manya H, Suzuki T, Akasaka-Manya K, Ishida HK, Mizuno M, Suzuki Y,Inazu T, Dohmae N, Endo T. 2007. Regulation of mammalian protein O-mannosylation: Preferential amino acid sequence for O-mannose modifica-tion. J Biol Chem. 282:20200–20206.

Martin PT, Freeze HH. 2003. Glycobiology of neuromuscular disorders. Gly-cobiology. 13:67R–75R.

McDearmon EL, Combs AC, Ervasti JM. 2001. Differential Vicia villosa agglu-tinin reactivity identifies three distinct dystroglycan complexes in skeletalmuscle. J Biol Chem. 276:35078–35086.

McDearmon EL, Combs AC, Ervasti JM. 2003. Core 1 glycans on alpha-dystroglycan mediate laminin-induced acetylcholine receptor clustering butnot laminin binding. J Biol Chem. 278:44868–44873.

McDearmon EL, Combs AC, Sekiguchi K, Fujiwara H, Ervasti JM. 2006. Brainalpha-dystroglycan displays unique glycoepitopes and preferential bindingto laminin-10/11. FEBS Lett. 580:3381–3385.

Moore CJ, Hewitt JE. 2008. Dystroglycan glycosylation and muscular dystro-phy. Glycoconj J.

Morelle W, Michalski JC. 2007. Analysis of protein glycosylation by massspectrometry. Nat Protoc. 2:1585–1602.

Mukherjee S, Chiu R, Leung SM, Shields D. 2007. Fragmentation of the Golgiapparatus: An early apoptotic event independent of the cytoskeleton. Traffic.8:369–378.

Munro S, Freeman M. 2000. The notch signalling regulator fringe acts in theGolgi apparatus and requires the glycosyltransferase signature motif DXD.Curr Biol. 10:813–820.

Oelmann S, Stanley P, Gerardy-Schahn R. 2001. Point mutations identified inLec8 Chinese hamster ovary glycosylation mutants that inactivate both theUDP-galactose and CMP-sialic acid transporters. J Biol Chem. 276:26291–26300.

Patnaik SK, Stanley P. 2005. Mouse large can modify complex N- and mucinO-glycans on alpha-dystroglycan to induce laminin binding. J Biol Chem.280:20851–20859.

Patnaik SK, Stanley P. 2006. Lectin-resistant CHO glycosylation mutants. Meth-ods Enzymol. 416:159–182.

Peyrard M, Seroussi E, Sandberg-Nordqvist AC, Xie YG, Han FY, Frans-son I, Collins J, Dunham I, Kost-Alimova M, Imreh S et al. 1999. Thehuman LARGE gene from 22q12.3-q13.1 is a new, distinct member ofthe glycosyltransferase gene family. Proc Natl Acad Sci USA. 96:598–603.

Pu L, Scocca JR, Walker BK, Krag SS. 2003. A single point mutation resultingin an adversely reduced expression of DPM2 in the Lec15.1 cells. BiochemBiophys Res Commun. 312:555–561.

Ravdin JI, Stanley P, Murphy CF, Petri WA Jr. 1989. Characterization of cellsurface carbohydrate receptors for Entamoeba histolytica adherence lectin.Infect Immun. 57:2179–2186.

Rojek JM, Spiropoulou CF, Campbell KP, Kunz S. 2006. Characterization ofthe cellular receptors for the South American hemorrhagic fever virusesJunin, Guanarito, and Machupo. Virology. 349:476–491.

Rojek JM, Spiropoulou CF, Campbell KP, Kunz S. 2007. Old World and clade CNew World arenaviruses mimic the molecular mechanism of receptor recog-nition used by alpha-dystroglycan’s host-derived ligands. J Virol. 81:5685–5695.

Sarkar AK, Fritz TA, Taylor WH, Esko JD. 1995. Disaccharide uptake andpriming in animal cells: Inhibition of sialyl Lewis X by acetylated Galbeta 1→4GlcNAc beta-O-naphthalenemethanol. Proc Natl Acad Sci USA.92:3323–3327.

Sarkar AK, Rostand KS, Jain RK, Matta KL, Esko JD. 1997. Fucosylationof disaccharide precursors of sialyl Lewis X inhibit selectin-mediated celladhesion. J Biol Chem. 272:25608–25616.

Sasaki K, Kurata-Miura K, Ujita M, Angata K, Nakagawa S, Sekine S,Nishi T, Fukuda M. 1997. Expression cloning of cDNA encoding a hu-man beta-1,3-N-acetylglucosaminyltransferase that is essential for poly-N-acetyllactosamine synthesis. Proc Natl Acad Sci USA. 94:14294–14299.

Schachter H, Vajsar J, Zhang W. 2004. The role of defective glycosylation incongenital muscular dystrophy. Glycoconj J. 20:291–300.

Seifert J, Ogawa T, Kurono S, Ito Y. 2000. Syntheses of alpha-dystroglycanderived glycosyl amino acids carrying a novel mannosyl serine/threoninelinkage. Glycoconj J. 17:407–423.

Smalheiser NR, Haslam SM, Sutton-Smith M, Morris HR, Dell A. 1998. Struc-tural analysis of sequences O-linked to mannose reveals a novel Lewis Xstructure in cranin (dystroglycan) purified from sheep brain. J Biol Chem.273:23698–23703.

Smalheiser NR, Kim E. 1995. Purification of cranin, a laminin binding mem-brane protein. Identity with dystroglycan and reassessment of its carbohy-drate moieties. J Biol Chem. 270:15425–15433.

Stanley P, Caillibot V, Siminovitch L. 1975. Selection and characterization ofeight phenotypically distinct lines of lectin-resistant Chinese hamster ovarycell. Cell. 6:121–128.

Stanley P, Siminovitch L. 1976. Selection and characterization of Chinese ham-ster ovary cells resistant to the cytotoxicity of lectins. In Vitro. 12:208–215.

Stanley P, Siminovitch L. 1977. Complementation between mutants of CHOcells resistant to a variety of plant lectins. Somatic Cell Genet. 3:391–405.

Stoll J, Cacan R, Verbert A, Krag SS. 1992. Lec15 cells transfer glucosylatedoligosaccharides to protein. Arch Biochem Biophys. 299:225–231.

985

at University of G


epartment on F


Dow

nloaded from


J T Aguilan et al.

Tessier DC, Dignard D, Zapun A, Radominska-Pandya A, Parodi AJ, Berg-eron JJ, Thomas DY. 2000. Cloning and characterization of mam-malian UDP-glucose glycoprotein: Glucosyltransferase and the develop-ment of a specific substrate for this enzyme. Glycobiology. 10:403–412.

Toda T, Kobayashi K, Takeda S, Sasaki J, Kurahashi H, Kano H, TachikawaM, Wang F, Nagai Y, Taniguchi K, et al. 2003. Fukuyama-type congeni-tal muscular dystrophy (FCMD) and alpha-dystroglycanopathy. CongenitAnom (Kyoto). 43:97–104.

van Reeuwijk J, Grewal PK, Salih MA, Beltran-Valero de Bernabe D,McLaughlan JM, Michielse CB, Herrmann R, Hewitt JE, Stein-brecher A, Seidahmed MZ, et al. 2007. Intragenic deletion in the

LARGE gene causes Walker–Warburg syndrome. Hum Genet. 121:685–690.

Wiggins CA, Munro S. 1998. Activity of the yeast MNN1 alpha-1,3-mannosyltransferase requires a motif conserved in many other families ofglycosyltransferases. Proc Natl Acad Sci USA. 95:7945–7950.

Yeung YG, Nieves E, Angeletti RH, Stanley ER. 2008. Removal of deter-gents from protein digests for mass spectrometry analysis. Anal Biochem.382:135–137.

Yoshida A, Kobayashi K, Manya H, Taniguchi K, Kano H, Mizuno M, Inazu T,Mitsuhashi H, Takahashi S, Takeuchi M et al. 2001. Muscular dystrophy andneuronal migration disorder caused by mutations in a glycosyltransferase,POMGnT1. Dev Cell. 1:717–724.

986

at University of G


epartment on F


Dow

nloaded from


mutational and functional analysis of large in a novel cho ...glycobiology/glycoomancc20311.pdf ·...

Documents