The Absence of Core Fucose Up-regulates GnT-III and Wnt Target Genes

Background: Little is known about how loss of a given glycan causes adaptive regulation of other glycosylation. Results: Deficiency in core α1,6-fucose specifically up-regulates bisecting GlcNAc by enhanced gene expression of a biosynthetic enzyme GnT-III. Conclusion: Wnt signaling pathway regulates the expression of GnT-III. Significance: Wnt-mediated GnT-III up-regulation may be an adaptive response to the loss of core fucose. Glycans play key roles in a variety of protein functions under normal and pathological conditions, but several glycosyltransferase-deficient mice exhibit no or only mild phenotypes due to redundancy or compensation of glycan functions. However, we have only a limited understanding of the underlying mechanism for these observations. Our previous studies indicated that 70% of Fut8-deficient (Fut8−/−) mice that lack core fucose structure die within 3 days after birth, but the remainder survive for up to several weeks although they show growth retardation as well as emphysema. In this study, we show that, in mouse embryonic fibroblasts (MEFs) from Fut8−/− mice, another N-glycan branching structure, bisecting GlcNAc, is specifically up-regulated by enhanced gene expression of the responsible enzyme N-acetylglucosaminyltransferase III (GnT-III). As candidate target glycoproteins for bisecting GlcNAc modification, we confirmed that level of bisecting GlcNAc on β1-integrin and N-cadherin was increased in Fut8−/− MEFs. Moreover using mass spectrometry, glycan analysis of IgG1 in Fut8−/− mouse serum demonstrated that bisecting GlcNAc contents were also increased by Fut8 deficiency in vivo. As an underlying mechanism, we found that in Fut8−/− MEFs Wnt/β-catenin signaling is up-regulated, and an inhibitor against Wnt signaling was found to abrogate GnT-III expression, indicating that Wnt/β-catenin is involved in GnT-III up-regulation. Furthermore, various oxidative stress-related genes were also increased in Fut8−/− MEFs. These data suggest that Fut8−/− mice adapted to oxidative stress, both ex vivo and in vivo, by inducing various genes including GnT-III, which may compensate for the loss of core fucose functions.

Glycans play key roles in a variety of protein functions under normal and pathological conditions, but several glycosyltransferase-deficient mice exhibit no or only mild phenotypes due to redundancy or compensation of glycan functions. However, we have only a limited understanding of the underlying mechanism for these observations. Our previous studies indicated that 70% of Fut8-deficient (Fut8 ؊/؊ ) mice that lack core fucose structure die within 3 days after birth, but the remainder survive for up to several weeks although they show growth retardation as well as emphysema. In this study, we show that, in mouse embryonic fibroblasts (MEFs) from Fut8 ؊/؊ mice, another N-glycan branching structure, bisecting GlcNAc, is specifically up-regulated by enhanced gene expression of the responsible enzyme N-acetylglucosaminyltransferase III (GnT-III). As candidate target glycoproteins for bisecting GlcNAc modification, we confirmed that level of bisecting GlcNAc on ␤1-integrin and N-cadherin was increased in Fut8 ؊/؊ MEFs. Moreover using mass spectrometry, glycan analysis of IgG 1 in Fut8 ؊/؊ mouse serum demonstrated that bisecting GlcNAc contents were also increased by Fut8 deficiency in vivo. As an underlying mechanism, we found that in Fut8 ؊/؊ MEFs Wnt/␤-catenin signaling is up-regulated, and an inhibitor against Wnt signaling was found to abrogate GnT-III expression, indicating that Wnt/␤catenin is involved in GnT-III up-regulation. Furthermore, various oxidative stress-related genes were also increased in Fut8 ؊/؊ MEFs. These data suggest that Fut8 ؊/؊ mice adapted to oxidative stress, both ex vivo and in vivo, by inducing various genes including GnT-III, which may compensate for the loss of core fucose functions.
There are over 180 glycosyltransferase genes in mammals, and limited kinds of glycosyltransferase knock-out (KO) mice show severe phenotypic abnormality, demonstrating essential roles of glycans for various physiological functions. However, most of glycosyltransferase KO mice show no or very mild phenotypic changes (1). This suggests that there exist some mechanisms of the redundancy of glycosyltransferases and compensation for the lack of the product of glycosyltransferase(s). In fact, several family members of glycosyltransferase are known to be functionally redundant (1), but so far little is known about the mechanisms by which mammals can adapt to the loss of a given glycan. ␣1,6-Fucosyltransferase (Fut8) 2 catalyzes the transfer of an ␣1,6-fucose residue from GDP-fucose to the innermost GlcNAc residue of glycans on glycoproteins (2). Purification and cDNA cloning of Fut8 from human and porcine were carried out by our group (3,4). This core fucosylation is generally recognized to be physiologically important glycosylation by the fact that Fut8 Ϫ/Ϫ mice show a semilethal phenotype, which includes severe growth retardation and emphysema-like lung destruction (5). These abnormalities are mainly caused by dysfunction due to the conformational changes of the receptors such as TGF-␤1 and EGF receptors (6,7).
Removal of the core fucose from human IgG 1 results in a significant enhancement in antibody-dependent cellular cytotoxicity (8), caused by the stronger binding of IgG 1 to Fc␥ receptor IIIa (Fc␥R IIIa) (9), whereas several studies also reported that high content of IgG 1 modified with bisecting GlcNAc, synthesized by GnT-III, leads to higher antibody-dependent cellular cytotoxicity activity (10,11). In addition, cellular functions of many glycoproteins, such as cadherins and integrins, are regulated by bisecting GlcNAc (12)(13)(14)(15)(16)(17).
Our group has been extensively studying the role of core fucose and bisecting GlcNAc as modifiers/regulators of the N-glycan (18 -20). Although the substrate specificity of Fut8 has been extensively studied, the issue of how the lack of a core fucose would affect the overall N-glycan structure remains unknown. Substrate specificity of Fut8 is strictly regulated because the addition of Gal or bisecting GlcNAc to N-glycans destroys their ability to serve as a substrate for Fut8 (21,22). Replica-exchange molecular dynamics simulations of complextype biantennary glycans revealed that the number of the major conformers is reduced upon the introduction of bisecting GlcNAc and/or core fucose (23). It was proposed that such conformer selection by bisecting GlcNAc and core fucose (23) may regulate the binding affinity of a target protein to which core fucose is added, an example of which is the TGF-␤1 receptor. However, interplay of expression/function between core fucose and bisecting GlcNAc has not been studied well so far.
In this study, we initially adopted a glycomic approach to compare the N-glycan structures between Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ MEFs. The findings indicate that overall N-glycan structures are altered in Fut8 Ϫ/Ϫ MEFs. Especially, bisected N-glycan was markedly increased in Fut8 Ϫ/Ϫ MEFs. We then found that mRNA expression of GnT-III is increased by about 3-fold together with Wnt/␤-catenin signaling as well as oxidative stress related genes, suggesting that an adaptive response may occur in Fut8 Ϫ/Ϫ mice to compensate for the absence of core fucose.

EXPERIMENTAL PROCEDURES
Cell Culture-Mouse embryonic fibroblasts (MEFs) derived from wild-type and Fut8 Ϫ/Ϫ mice (C57BL/6 genetic background) were established as described in a previous study (5). MEFs were cultured in DMEM supplemented with 10% FBS and penicillin and streptomycin at 37°C in 5% CO 2 -humidified atmosphere. In the case of the Wnt inhibition experiment, the media were supplemented with either DMSO or IWP-2 (final 15 M) (Wako Pure Chemical), and the media were changed every 24 h for 3 days.
Lectin Blotting-Each sample was separated on SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was blocked with 1% BSA in TBS-T for 30 min and incubated for 1 h with HRP-phytohemagglutinin-L4 (L4-PHA) or HRPphytohemagglutinin-E4 (E4-PHA) lectin, which were used at 1/1000 dilution. After washing with TBS-0.1% Tween 20 (TBS-T), lectin reactive proteins were detected using a 3-fold diluted SuperSignal West Dura extended duration substrate (Thermo Scientific).
In the case of biotinylated Aleuria aurantia lectin (AAL) (1/1000 dilution), the membrane was blocked with TBS-T for 30 min. Lectin reactive proteins were detected using a VECTASTAIN ABC kit (Vector Laboratories), and the blots were developed according to the manufacturer's instructions.
Glycosyltransferase Assay-MEFs, grown on 10-cm dishes to full confluence, were suspended with 60 -100 l of 10 mM Tris-HCl (pH 7.4), 0.25 M sucrose, and proteinase inhibitors and then sonicated for 5 min on ice-chilled water bath with the Bioruptor sonicator (Cosmo Bio). The lysates were centrifuged at 12,000 ϫ g for 5 min. The protein contents of the resulting supernatants were determined using a BCA protein assay (Thermo Scientific). Detailed procedures were reported in a previous study (24). The enzyme activities of Fut8, GnT-III, IV and V were determined using a pyridylamino-labeled oligosaccharide (TaKaRa pyridylamino-sugar chain 012) as an acceptor substrate. The resulting 3 l of cell lysates, 5 l of reaction buffer (200 mM MOPS-NaOH (pH 7.0), 20 mM MnCl 2 , 400 mM GlcNAc, 1% Triton X-100, 2 mg/ml BSA), 1 l of 0.1 mM acceptor substrate, and 1 l of 400 mM UDP-GlcNAc were incubated at 37°C for 4 h. The reaction was stopped by boiling for 3 min after adding 40 l of water followed by centrifugation at 20,000 ϫ g for 5 min. Resulting supernatant (10 l) was used for assay of Fut8 and GnT activities by HPLC as described previously (25).
N-Glycan Digestions-Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ MEFs (5 ϫ 10 7 cells) were washed three times with ice-cold PBS. The cell pellets were lysed with 500 l of 1% Triton X-100 in PBS containing a protease inhibitor cocktail. After centrifugation of the cell suspension at 10,000 ϫ g for 10 min, the supernatant was mixed with 3 volumes of ice-cold 95% ethanol and then incubated at Ϫ80°C for 3 h. The suspension was centrifuged at 15,000 ϫ g for 10 min. The precipitation was first reduced (10 mM DTT for 30 min at room temperature in 50 mM NH 4 HCO 3 ) and then alkylated in the dark (20 mM 2-iodoacetamide at room temperature for 30 min). This was followed by digestion with 25 l of 2 mg/ml trypsin and chymotrypsin (Nacalai Tesque) in 50 mM NH 4 HCO 3 overnight at 37°C. The digested mixtures were boiled for 5 min followed by incubating with 5 l of peptide N-glycosidase F (New England Biolabs) for 18 h at 37°C. After removing the insoluble materials by centrifugation, the supernatants were lyophilized. The digested samples were redissolved in 250 l of 5% (v/v) acetic acid for solid-phase extraction with Oasis HLB extraction cartridges. Released oligosaccharide was recovered by collecting the flow-through, which is at least 5 ml of 5% acetic acid. The released glycans were lyophilized. The amount of obtained glycans was quantified by the phenol-sulfuric acid method (26).
Permethylation of N-Glycan-To enhance the sensitivity of N-glycan analysis by mass spectrometry, permethylation of N-glycans was performed by the NaOH/DMSO slurry method (27). To prepare the NaOH slurry, small beads of NaOH (50 mg, Fluka) were mixed with 250 l of DMSO (∞Pure, Wako Pure Chemical) in an agate mortar. The slurry was added to the lyophilized 1 mg of N-glycan sample and vigorously agitated with 50 l of iodomethane (Wako Pure Chemical) for 30 min. To stop the reaction, 1 ml of ice-cold water was added to the reaction mixture, and the resulting mixture was applied to a prewashed Oasis HLB column (Waters), followed by 1 ml of acetonitrile and dried under a N 2 stream. Obtained permethylated glycan was analyzed using Bruker Ultraflex MALDI mass spectrometry.

Permethylated Glycan Analysis by MALDI-TOF-MS-
The permethylated oligosaccharides were dissolved in methanol and spotted on an MTP AnchorChip 400/384 T F plate with 10 mg/ml 2,5-dihydroxybenzoic acid and 5-methoxysalicylic acid (SDHB, Bruker) dissolved in a 0.1% (v/v) TFA and acetonitrile solution (2:1) for measurement. MS spectra were acquired on an Ultraflex MALDI-TOF/TOF (Bruker). MS n spectra were recorded on an AXIMA-QIT TOF-MS (Shimadzu Biotech). Sample and 10 mg/ml 2,3-dihydroxybenzoic acid in 0.1% TFA and acetonitrile (3:2) (Shimadzu) were applied on a Focus MALDI plate magnetic holder for Shimadzu. The measurements were carried out in the positive ion mode. The mass spectra acquired by at least 200 laser shots were accumulated, and the measurement was repeated at least three times. The peak height of the [MϩNa] ϩ ions was measured for relative semiquantitation at MS analysis. The fragmentation nomenclature for oligosaccharide by Domon and Costello (28) was used for MS n analysis.
Immunoprecipitation-The cell lysates (adjusted to be less than 1 mg/ml of protein concentration) prepared from Fut8 ϩ/ϩ , Fut8 Ϫ/Ϫ , and reintroduced MEFs were incubated with 3 g of a purified anti-mouse/rat CD29 (␤1-integrin) antibody (BioLegend) at 4°C overnight. Membrane fractions (100 -200 g) were incubated with 3 g of anti-␤1-integrin antibody (BD Biosciences) at 4°C overnight. Membrane fractions in TBS-1% Triton X-100 were incubated with anti-N-cadherin antibody (Abcam). For precipitation, 20 l of 50% slurry of protein G-agarose (GE Healthcare) was added and incubated at 4°C for 30 min. The precipitates were washed three times with PBS and analyzed by Western blotting.
Purification of IgG 1 from Mouse Serum and Analysis by MALDI-TOF MS-Serum from Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ mice (16-week-old male and female mice, ICR genetic background) were a generous gift from Dr. Jianguo Gu (Tohoku Pharmaceutical University). The amount of IgG was determined using a mouse IgG ELISA kit (Roche Applied Science). Mouse serum, containing 100 -200 g of IgG, was diluted 10-fold with phosphate buffer (pH 7.0). The mouse serum was purified on a protein G-Sepharose column (300 l). After washing the column with phosphate buffer (pH 7.0), mouse IgG was eluted with IgG elution buffer (Thermo Scientific) followed by neutralization with 1 M Tris-HCl (pH 9.0). The freeze-dried eluate was dissolved in 400 l of a solution of 6 M guanidine and 0.25 M Tris-HCl, pH 8.0, reduced with 0.13 M dithiothreitol at 56°C for 1 h and then alkylated with 0.22 M iodoacetamide for 30 min at room temperature in the dark. The reactant was applied to a NAP5 column (GE Healthcare) equilibrated with 0.05 N HCl, eluted with 1 ml of 0.05 N HCl as reported previously (29 -31). Briefly, the eluate was digested with trypsin and lysylendopeptidase (Wako Pure Chemical) at 37°C for 18 h. Obtained glycopeptides were purified using Sepharose CL4B (GE Healthcare) in n-butanol/ethanol/water (4:1:1) and 50% aqueous ethanol. The aqueous ethanol fraction was dried using a SpeedVac, and the glycopeptide samples were mixed with 10 mg/ml of 2,3-dihydroxybenzoic acid, used as a matrix, which was prepared by dissolving 50% (v/v) acetonitrile solution. The glycopeptide analysis was accomplished using a Voyager DE Pro MALDI-TOF mass spectrometer (Applied Biosystems). The measurements were carried out in positive ion and linear mode. Microarray Analysis-Poly(A)ϩ RNA was isolated from the cells using an mTRAP kit (ACTIVE MOTIF). The quality of RNA was measured using 2100 Bioanalyzer (Agilent). The RNAs were stored at Ϫ80°C until use. Briefly, total RNAs were then reverse-transcribed into cDNA, and biotinylated cDNAs were synthesized using the GeneChip 3ЈIVT Express kit (Affymetrix). DNA microarray experiments were performed using Mouse Genome 430 v2.0 (Affymetrix). The hybridization signal on the chip was scanned using a GeneChip 3000 7G scanner and processed by GeneChip operating software (GCOS) Ver. 1.4 (Affymetrix). The DNA microarray expression profiles were compared between Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ MEFs.
Quantitative PCR-Total RNAs were isolated from MEFs using TRI Reagent (Molecular Research Center, Inc.) according to the manufacturer's instructions. Three g of the RNAs was reverse-transcribed with random hexamers using a SuperScript III first-strand synthesis system for RT-PCR (Invitrogen) according to the manufacturer's protocol. The expression levels of the target genes were measured in duplicate and normalized by the corresponding ribosomal RNA expression levels. For real-time quantitative PCR assay, GnT-III, ␤-catenin, ribosomal RNA primers and probes, Assay-on-Demand gene expression products, and cDNAs were added to TaqMan universal PCR master mix (Applied Biosystems). The probes for GnT-III and ␤-catenin were labeled with FAM at its 5Ј-end and with quencher minor groove binder at its 3Ј-end. The probe for ribosomal RNA was labeled with VIC at its 5Ј-end and with quencher TAMRA at its 3Ј-end. The cDNAs were amplified by one cycle at 50°C for 2 min, one cycle at 95°C for 10 min, and 40 cycles at 95°C for 15 s and 60°C for 1 min in a total volume of 20 l using an ABI PRISM 7900HT sequence detection system (Applied Biosystems).
Preparation of a Membrane and Cytosolic Fraction from MEFs-To prepare membrane and cytosolic fractions, MEFs were harvested with 1 ml of PBS containing a protease inhibitor cocktail and homogenized with a potter homogenizer. The obtained homogenates were centrifuged at 100,000 ϫ g for 33 min at 4°C. The supernatant was collected as the cytosolic fraction. The precipitate was resuspended in 1 ml of 1% Triton X-100 in PBS containing a protease inhibitor cocktail followed by rotation at 4°C overnight and then centrifuged at 100,000 ϫ g for 23 min at 4°C. The supernatant was collected as the membrane fraction. The concentration of protein was determined using a BCA protein assay (Thermo Scientific).

N-Glycan Structures Were Markedly Altered in Fut8
Ϫ/Ϫ MEFs-To investigate how loss of core fucosylation affects overall N-glycan structures, we carried out a series of lectin blot analyses for MEFs derived from Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ mice. First, the application of AAL lectin, which preferentially recognizes fucose (32) (Fig. 1A), verified the loss of core fucose in lysates of Fut8 Ϫ/Ϫ MEFs (Fig. 1B, left). The bands at 118 and 70 kDa (shown as asterisk) in the Fut8 Ϫ/Ϫ MEFs were nonspecific bands and were detected by the avidin-horseradish peroxidase conjugate (32). A marked increase in bisecting GlcNAc in the Fut8 Ϫ/Ϫ MEFs was detected by E4-PHA blotting (Fig. 1B, middle). An overall decrease in the signal intensities of L4-PHA, a ␤1,6-GlcNAc branch in N-glycan, was observed (Fig. 1B, right). This could be explained by the fact that GnT-III activity competes with that of a ␤1,6-GlcNAc branch-synthesizing enzyme GnT-V (33). It should be noted, however, that several exceptionally strong signals were detected with L4-PHA lectin in the lysates of Fut8 Ϫ/Ϫ MEFs.
On the assumption that a specific GnT activity would be altered in Fut8 Ϫ/Ϫ MEFs, we measured the activities of a series of GnT enzymes (GnT-III, IVa,b, and -V) in Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ MEFs. We confirmed the complete lack of Fut8 activity in Fut8 Ϫ/Ϫ MEFs. We detected a marked increase (ϳ8-fold) in GnT-III activity in Fut8 Ϫ/Ϫ MEFs ( Fig. 2A), although the GnT-V activity in Fut8 Ϫ/Ϫ MEFs was decreased to about ϳ60%. Furthermore, real-time quantitative PCR analysis revealed that GnT-III mRNA was increased by ϳ3-fold, and the mRNA level of GnT-V was decreased to 30% (Fig. 2B), suggesting that the change in glycan structure is controlled at transcription level in GnT-III and -V. The mRNA level of GnT-IVa in Fut8 Ϫ/Ϫ was increased 2-fold as compared with those in Fut8 ϩ/ϩ MEFs. In contrast, the activity of GnT-IVa,b was not altered in Fut8 Ϫ/Ϫ MEFs. This can be attributed to the alteration in the mRNA level of GnT-IVb. To address the effects of the expression of Fut8 on expression of GnT-III, we reintroduced the Fut8 gene into Fut8 Ϫ/Ϫ MEFs. As shown in Fig. 2C, the mRNA level of GnT-III was down-regulated in the reintroduced MEFs.
These alterations in glycosylation were investigated by a comparative glycomic analysis in which N-glycans were enzymatically released from the cell lysates of Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ MEFs. The resulting free N-glycans were then permethylated to facilitate their analysis by MALDI-TOF MS. The acquired mass spectra of the N-glycans are shown in Fig. 3A. All signals corresponding to mono-fucosylated glycans detected in Fut8 ϩ/ϩ MEFs completely disappeared in Fut8 Ϫ/Ϫ MEFs. As shown in Fig. 3B, the relative proportions of the high mannose to total N-glycans decreased by ϳ80% in Fut8 Ϫ/Ϫ MEFs as compared with those in Fut8 ϩ/ϩ MEFs. The relative ratios of bisecting GlcNAc in each branched structure are difficult to compare between WT and KO MEFs. However, as shown in Fig. 3A, in WT MEFs bisected biantennary sugar chains were found at m/z 2111, 2315, 2489, and 2820, whereas in KO MEFs 1662, 1907, 2315, 2519, 2723, and triantennary sugar chain 2765 were found. All these peaks were assigned by MS 2 or MS 3 analysis.
The structural assignment of bisecting GlcNAc-containing N-glycans observed in MS analysis was performed based on subsequent MS/MS analysis data using AXIMA-QIT TOF MS (Fig. 3C). The sodiated ion at m/z 2081 obtained from Fut8 ϩ/ϩ MEFs was subjected to MS 2 analysis in Fig. 3C (panel i). We observed the product ion at m/z 444.0 (B/Y/Z and/or C/Z/Z) at MS 3 (Fig. 3C (panel ii)), which contains the bisecting GlcNAc, produced from the precursor ion at m/z 1562.6 (Fuc1Hex3HexNAc3). The precursor ion is composed of two possible fragments as shown in Fig. 3C (panel i). The ion at m/z 444 is previously reported as a key product ion bearing a bisecting GlcNAc containing glycan (34).
We next performed an MS/MS analysis of the sodiated ion at m/z 2315 from Fut8 Ϫ/Ϫ MEFs at the MS 2 stage. The ion at m/z 1388.9, which is composed of Hex3HexNAc3, consists of three possible isomeric structures (Fig. 3C (panel iii)). In the m/z 1389 fragment in the MS 3 spectra (Fig. 3C (panel iv)), the ion at m/z 426.5 (Man-GlcNAc) was observed, which corresponds to the loss of a branching hexose and bisecting GlcNAc.
IgG 1 Derived from Fut8 Ϫ/Ϫ Mice Serum Contained Increased Bisecting GlcNAc-Because we anticipated that an increase in bisecting GlcNAc would be also observed in vivo, we next performed glycopeptide structural analyses of serum IgG 1 derived from Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ mice using MALDI-TOF MS. We confirmed that all of the IgG 1 glycans in Fut8 Ϫ/Ϫ mice lacked a core fucose unit (Fig. 4A, lower), whereas IgG 1 glycans obtained from Fut8 ϩ/ϩ mice were fully modified with core fucose (Fig.  4A, upper). Mizuochi et al. (35) previously reported that most of IgGs from mice were core fucosylated and contained no bisecting GlcNAc. Most of glycopeptides in Fut8 Ϫ/Ϫ were non-core fucosylated forms of the structures observed in Fut8 ϩ/ϩ . Therefore, although the ions at m/z 2926.1 in Fut8 ϩ/ϩ and at m/z 2925.8 in Fut8 Ϫ/Ϫ showed close mass numbers, we assigned the ion at m/z 2925.8 in Fut8 Ϫ/Ϫ as the non-core fucosylated form of the m/z 3071.5 (G1FS) fragment in Fut8 ϩ/ϩ .
The glycan profiles are summarized in Fig. 4B. In the case of IgG 1 from Fut8 ϩ/ϩ mice, bisecting GlcNAc was not detectable. In Fut8 Ϫ/Ϫ mice, the ratio of simple biantennary glycan (G2) in IgG 1 was significantly reduced as compared with its core fucosylated form (G2F) in Fut8 ϩ/ϩ mice, and a marked increase in bisecting GlcNAc-containing glycans, such as G0N and G1N, was clearly observed. Actually, bisected N-glycans were not detectable in IgG 1 from Fut8 ϩ/ϩ mice, whereas 9.7% of the N-glycan contained a bisecting GlcNAc in IgG 1 from Fut8 Ϫ/Ϫ mice (Fig. 4C). These data clearly show that the up-regulation of bisecting GlcNAc was also observed in Fut8 Ϫ/Ϫ mice.
Change in the Glycosylation of ␤1-Integrin and N-cadherin-We and others previously reported that there are several target proteins for GnT-III such as integrins and cadherins (12)(13)(14). In addition, a recent study reported that integrin-dependent cell migration and intracellular signaling were down-regulated in the absence of core fucose (36). Therefore, we expected that integrin and/or cadherin would be a target for core fucose/ bisecting GlcNAc and contain an increased level of bisecting N-glycan in Fut8 Ϫ/Ϫ MEFs. Western blot analysis of the immunoprecipitated ␤1-integrin from both Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ MEFs showed double bands, and the upper band of ␤1-integin, corresponding to the mature form, migrated faster in Fut8 Ϫ/Ϫ MEFs (Fig. 5A, upper). This difference was not cancelled by sialidase treatment (Fig. 5B), indicating that the glycan structure of the ␤1-integrin other than sialylation constitutes the difference between Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ MEFs. Lectin blot analysis with AAL and E4-PHA showed that the upper band of ␤1-integrin (from Fut8 Ϫ/Ϫ MEFs) lacks core fucose and has a higher level of bisecting GlcNAc (Fig. 5A, lower). We unexpectedly observed a marked increase in the L4-PHA epitope on ␤1-integrin, suggesting that the competition of GnT-III and GnT-V is somehow dependent on the type of glycoprotein, although GnT-III and GnT-V compete with each other for the common substrate (37,38). In the Fut8-reintroduced MEFs, the AAL signal was recovered in the upper band of ␤1-integrin (Fig.  5C, middle). The E4-PHA signal on ␤1-integrin from Fut8-reintroduced MEFs was reduced as compared with Fut8 Ϫ/Ϫ MEFs (Fig. 5C, lower). These results are consistent with the mRNA levels of GnT-III (Fig. 2C, right).
Because a major cadherin component in MEFs is N-cadherin instead of E-cadherin (39), glycosylation status of N-cadherin was examined in both types of cells (Fig. 5D). We found enhanced E4-PHA signals and reduced L4-PHA in the immunoprecipitated N-cadherin from Fut8 Ϫ/Ϫ MEFs. These data show that ␤1-integrin and N-cadherin are highly modified with bisecting GlcNAc by the lack of their core fucose.
Increase in ␤-Catenin Levels and Down-regulation of Phosphorylation-Our data showed that the absence of core fucose up-regulates bisecting GlcNAc level through increased expression of GnT-III gene. To explore the underlying mechanism of GnT-III gene up-regulation, we carried out DNA microarray analysis of Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ MEFs. We found that Wnt-responsive genes were significantly up-regulated in Fut8 Ϫ/Ϫ MEFs, including the growth arrest-specific 1 gene (12.1-fold increase) and the Frizzled homolog 6 gene (2.3-fold increase) ( Table 1), suggesting that Wnt/␤-catenin signaling is enhanced in Fut8 Ϫ/Ϫ MEFs. In fact, the level of ␤-catenin was increased in Fut8 Ϫ/Ϫ MEFs in terms of protein content as well as mRNA levels as judged by real-time quantitative PCR and Western blotting (Fig. 6, A and C). The degradation of ␤-catenin by the proteasome strictly depends upon its phosphorylation status (40). To determine the phosphorylation level of ␤-catenin, we carried out Western blotting, the results of which indicated that the phosphorylation level of ␤-catenin was decreased (Fig. 6B) in Fut8 Ϫ/Ϫ MEFs. The result suggests that the N-cadherin-␤-catenin complex is not sufficiently degraded by the ubiquitin/proteasome pathway and accumulates in the cytosol and nuclei (Fig. 6C), although N-cadherin/␤-catenin complex level was not changed (Fig. 6D).
Wnt Signal Inhibitor (IWP-2) Reduced mRNA Level of GnT-III in MEFs-We speculated that enhanced Wnt/␤-catenin signaling contributed to the up-regulation of GnT-III gene. To confirm whether or not Wnt signaling is involved in the up-regulation of GnT-III, IWP-2, an inhibitor of the O-acyltransferase, porcupine, which palmitoylates Wnt, was added to the culture medium, and the level of GnT-III was determined. Indeed, IWP-2 decreased the mRNA levels of ␤-catenin to 67% of the control in Fut8 Ϫ/Ϫ MEFs (Fig. 6, A and C). In addition, expression level of GnT-III was also suppressed to 70% of the mock sample in Fut8 Ϫ/Ϫ MEFs, whereas IWP-2 treatment did not have an effect on GnT-V expression (Fig. 6A). This suggests that the change in the Wnt/␤-catenin signal is highly associated with the up-regulation of GnT-III. Because we also found the reduction of GnT-III expression in Fut8 ϩ/ϩ MEFs, Wnt signaling pathway generally up-regulates the GnT-III expression.

DISCUSSION
In this study, our glycomic analysis using glycans released from whole lysates of Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ MEFs revealed an

Absence of Core Fucose Up-regulates GnT-III
increase in bisected N-glycans in Fut8 Ϫ/Ϫ MEFs. These data prompted us to investigate whether or not some kinds of compensation for or redundancy of glycan functions may exist in these mice due to the absence of core fucose. We then found a significant increase in bisected N-glycan content in Fut8 Ϫ/Ϫ MEFs, and that appears to be due to a markedly higher expression of GnT-III mRNA in Fut8 Ϫ/Ϫ MEFs. In the case of Fut8 Ϫ/Ϫ MEFs, to which Fut8 was reintroduced, the mRNA level of GnT-III was down-regulated, confirming that the expression of Fut8 affects the expression of GnT-III. Microarray analysis with Fut8 ϩ/ϩ and Fut8 Ϫ/Ϫ cells revealed the up-regulation of Wnt target genes. It should be noted that Wnt inhibitor IWP-2, which prevents Wnt secretion, leading to the reduction of cellular level of ␤-catenin, abrogates GnT-III expression especially in Fut8 Ϫ/Ϫ cells. This transcriptional regulation may play a key role in the adaptive response to the loss of core fucose. A recent study by Chen et al. (41) shows that Fut8 expression is up-regulated during epithelial-mesenchymal transition, a critical process for malignant transformation of tumor, in the ␤-catenin/LEF-1-dependent pathway (41). It is interesting that both GnT-III expression and Fut8 expression were regulated by ␤-catenin/LEF1 pathway. It is known that E-cadherin/␤-catenin-dependent cell-cell adhesion induces GnT-III expression (42). We also found that N-cadherin instead of E-cadherin is expressed in MEFs, but the levels of N-cadherin as well as the N-cadherin-␤-catenin complex were comparable in the two types of cells, regardless of the increase in bisected N-glycans in N-cadherin from Fut8 Ϫ/Ϫ MEFs (Figs. 5D and 6D). In a microarray analysis, expression of a number of genes other than Wnt target genes was altered by Fut8 deficiency, which might partly explain the adaptive response of Fut8 Ϫ/Ϫ mice. In the microarray analysis, we were not able to detect any up-regulation in GnT-III, probably because glycosyltransferase expression, including that for GnT-III, is too low to permit a satisfactory quantitative measurement to be made, as reported previously (43,44).
In Fut8 Ϫ/Ϫ MEFs as well as in Fut8 Ϫ/Ϫ mice, the absence of core fucose induces oxidative stress as judged by up-regulation of various oxidative stress-related genes including a major transcription factor Nrf2 (Table 1). Funato et al. (45) show that oxidative stress induces up-regulation of Wnt target genes. This suggests that deficiency in Fut8 might cause oxidative stress and that up-regulation of GnT-III might be a response to reduce cellular oxidative stress. In fact in Fut8 Ϫ/Ϫ mice, 70% of the mice die within 3 days after birth but ϳ30% survive, suggesting that an adaptive response may have occurred. Adaptive regulation may proceed in some cases of glycosyltransferase KO mice as reported (46 -48).
As depicted in Fig. 7, in the case of Fut8 Ϫ/Ϫ MEFs, the "Wnt on" situation occurs under oxidative stress conditions. Nrf2 and Wnt are up-regulated, and phosphorylation of ␤-catenin is inhibited. ␤-Catenin was then accumulated in cytosol and nuclei and cooperated with LEF1 to activate GnT-III gene.
This study may open a new insight into the elucidation of adaptive response in vivo in terms of glycan functions. Clarification of such compensation or redundancy mechanism would reveal how glycan functions are maintained in vivo.