O-Fucose Monosaccharide of Drosophila Notch Has a Temperature-sensitive Function and Cooperates with O-Glucose Glycan in Notch Transport and Notch Signaling Activation*

Background: The requirement of O-fucose monosaccharide on Notch is not fully understood. Results: Loss of O-fucose monosaccharide on Notch caused temperature-sensitive loss of Notch signaling. Conclusion: O-Fucose monosaccharide of Notch has a temperature-sensitive function and cooperates with O-glucose glycan in Notch signal activation. Significance: Our findings elucidate how different forms of glycosylation on a protein influence protein functions. Notch (N) is a transmembrane receptor that mediates the cell-cell interactions necessary for many cell fate decisions. N has many epidermal growth factor-like repeats that are O-fucosylated by the protein O-fucosyltransferase 1 (O-Fut1), and the O-fut1 gene is essential for N signaling. However, the role of the monosaccharide O-fucose on N is unclear, because O-Fut1 also appears to have O-fucosyltransferase activity-independent functions, including as an N-specific chaperon. Such an enzymatic activity-independent function could account for the essential role of O-fut1 in N signaling. To evaluate the role of the monosaccharide O-fucose modification in N signaling, here we generated a knock-in mutant of O-fut1 (O-fut1R245A knock-in), which expresses a mutant protein that lacks O-fucosyltransferase activity but maintains the N-specific chaperon activity. Using O-fut1R245A knock-in and other gene mutations that abolish the O-fucosylation of N, we found that the monosaccharide O-fucose modification of N has a temperature-sensitive function that is essential for N signaling. The O-fucose monosaccharide and O-glucose glycan modification, catalyzed by Rumi, function redundantly in the activation of N signaling. We also showed that the redundant function of these two modifications is responsible for the presence of N at the cell surface. Our findings elucidate how different forms of glycosylation on a protein can influence the protein's functions.

O-fut1 and Pofut1 are apparently globally required for normal N signaling in Drosophila and mammals, respectively, in that their mutant phenotypes are similar to those caused by disrupted N signaling (16 -18). In contrast, mutants of genes encoding Fng family proteins reduce N signaling in only a subset of contexts that involve N signaling (15, 19 -21). Therefore, the GlcNAc modification by Fng, which regulates the interaction between N and its ligands, Delta (Dl) or Serrate, is required in a tissue-specific manner for various tissue boundary formations (22).
O-Fut1 has two O-fucosyltransferase activity-independent functions (23)(24)(25)(26). First, it is an endoplasmic reticulum (ER) resident protein and functions as a N-specific chaperon (23). Second, overexpressed O-Fut1 promotes the endocytosis of the N cell nonautonomously (25). Thus, it has been difficult to determine whether O-Fut1's O-fucosyltransferase activity or its nonenzymatic functions (or both) are required in N signaling. To address this issue, Okajima et al. (23) generated a genomic fragment, including a mutant O-fut1 locus, O-fut1 R245A , which produces an O-Fut1 variant carrying an amino acid substitution in the binding site for GDP-fucose. O-Fut1 R245A lacks O-fucosyltransferase activity in vitro, although it reportedly retains its N-specific chaperon activity (23). Most of the defects associated with the absence of O-fut1 are rescued by introducing the O-fut1 R245A genomic locus, except for the phenotypes that are probably due to the disruption of Fng functions, which are tissue-specific and depend on the O-fucosylation of N (24). Based on these results, it was proposed that the O-fucose monosaccharide modification of N does not have a specific function in N signaling, although it is required for the Fng-dependent activation of N signaling. However, it was also reported that the ability of the O-fut1 R245A genomic locus to rescue the O-fut1 null allele differs among transgenic lines, probably because the insertion site of the transgene affects the transcription efficiency of the integrated O-fut1 R245A locus (23,24).
To overcome this problem, here we generated a knock-in mutant of O-fut1 R245A (O-fut1 R245A knock-in ) in Drosophila melanogaster. We used O-fut1 R245A knock-in and other mutants affecting the O-fucosylation of Notch's EGF-like repeats to examine the specific role of the monosaccharide O-fucose in N signaling. We also studied the genetic interaction between O-fut1 R245A knock-in and rumi, which encodes an O-glucosyltransferase that adds O-glucose to the EGF-like repeats of N (27)(28)(29)(30). Based on the presented results, we propose a model showing how multiple O-glycosylation sites in EGF-like repeats might affect the function of N.
Generation of the O-fut1 R245A knock-in Fly-O-fut1 R245A knock-in is a knock-in mutation generated by a homologous recombination method described previously (40,41). Two genomic fragments covering the O-fut1 locus, referred to as the left arm and right arm, were PCR-amplified. The left arm (5005 bp) was amplified using the primers 5Ј-CAACCAAGCAGGGCCAAT-CCCA-3Ј and 5Ј-AATTTCTTATAGTCATATAAATACAA-AATA-3Ј, and it included the region from 4560 bp upstream of the start of the O-fut1 5ЈUTR to 188 bp downstream of the end of the O-fut1 3ЈUTR. The right arm (4996 bp) was amplified using the primers 5Ј-TCTTTTAGCTTTAATTCTTAAAAA-GGATTT-3Ј and 5Ј-CCGAATCGGCGACCCAGTAAAC-3Ј, and it included the region from 189 bp downstream of the end of the O-fut1 3ЈUTR to 5115-bp downstream of the end of the O-fut1 3ЈUTR. The left arm fragment was inserted into the AscI site of the pT7 Blue vector (Novagen), and the right arm fragment was inserted between the SphI and NotI sites of the pT7 Blue vector. The resulting constructs were pT7 Blueϩleft arm and pT7 Blueϩright arm. To introduce a base substitution that would result in the amino acid replacement of arginine (Arg) at the 245th amino acid with alanine (Ala), an overlap extension PCR was performed using pT7 Blueϩleft arm and two primers, 5Ј-CATCTGGCCAACGGTATCGATTGGGTG-3Ј and 5Ј-A-CCGTTGGCCAGATGAATGCCCAAAAA3Ј. The right arm and mutated left arm were excised and cloned into an ends-out homologous recombination vector, pW25, with a selectable marker, white (40,41). This construct was introduced into the Drosophila genome by P-element-mediated transformation (41). Using the transgenic line obtained, homologous recombination was performed as described previously (40,41). Briefly, pW25 contains two lox sites, which make it feasible to remove the white marker by Cre-mediated recombination (41). The white marker was removed as described previously (41), and the resulting lines were maintained as O-fut1 R245A knock-in . The O-fut1 locus of the O-fut1 R245A knock-in line was sequenced, and the mutation was confirmed.

Comparison of the O-Fucose and O-Glucose Sites in Various Notch
Receptors-The Swiss-Prot and KEGG GENES databases were searched using the Motif search program or EGF-like repeats containing the consensus sequence for O-glucosylation between the first and second cysteines, and for O-fucosylation between the second and third cysteines (8,13).

O-Fucosyltransferase Activity of O-fut1 R245A knock-in Is
Negligible in Vivo-In the previously reported Drosophila O-fut1 R245A mutant, arginine is replaced by alanine at the 245th amino acid position of the deduced O-Fut1 protein (Fig. 1A) (23). This amino acid substitution is located in the GDP-fucosebinding motif and largely abolishes the O-fucosyltransferase activity in vitro (Fig. 1A) (23,55,56). We first sought to confirm that the O-fucosyltransferase activity of O-fut1 R245A knock-in was also negligible in vivo. O-fut1 R245A knock-in homozygotes survived until the third-instar larval stage (data not shown). In the wild-type wing discs of the third-instar larvae, N signaling is activated at the border between fng-expressing and fng-nonex-pressing regions, which correspond to the dorsal and ventral (D/V) compartments, respectively. Therefore, the expression of wingless (wg), a downstream gene of N signaling, is activated along the D/V compartment boundary (D/V boundary) (  modification of the O-fucose on NЈs EGF-like repeats by Fng (20). We found that wg expression was abolished in the wing discs isolated from O-fut1 R245A knock-in homozygotes at the third instar cultured at 25°C (Fig. 1C), probably because the fng function was disrupted in the absence of NЈs O-fucosylation. As reported previously, misexpressed fng driven by dpp-Gal4 at 25°C resulted in the ectopic activation of wg in the ventral compartment of the wild-type wing, where endogenous fng is not expressed (n ϭ 99) (Fig. 1, D and DЈ). However, when we misexpressed fng driven by dpp-Gal4 in the wing discs of O-fut1 R245A knock-in homozygotes cultured at 25°C, no wg expression was induced in any case examined (n ϭ 52) (Fig. 1, E and EЈ). We confirmed that the expression pattern of dpp-Gal4 was almost the same in the wing discs of wild-type and O-fut1 R245A knock-in homozygotes at this temperature ( O-fut1 is a maternal-effect gene (17). Therefore, to observe the phenotypes associated with the O-fut1 R245A knock-in mutation, we needed to remove the maternal contribution of the wild-type O-fut1 gene from the female germ line. However, these females are heterozygous for O-fut1 R245A knock-in , because O-fut1 R245A knock-in is a recessive lethal mutation (data not shown). Therefore, we obtained embryos homozygous for O-fut1 R245A knock-in from females carrying germ line clones homozygous for O-fut1 R245A knock-in , using a previously described method (17,58). In this study, we call these O-fut1 R245A knock-in Homozygotes Obtained from Females Carrying the O-fut1 R245A knock-in Homozygous Germ Line Show a Temperature-sensitive Neurogenic Phenotype-In the embryonic central nervous system, the number of neuroblasts segregated from the neuroectoderm is determined by "lateral inhibition" through N signaling (59). A disruption of N signaling causes a failure of lateral inhibition, resulting in neuronal hyperplasia, known as the neurogenic phenotype (59). We studied the nervous system development in O-fut1 R245A knock-in m/z embryos at 25 and 30°C, in case O-fut1 R245A had a temperature-sensitive property. Neurons were detected by anti-Elav antibody staining (Fig. 2, A-J). At 25 and 30°C, the wild-type flies develop and reproduce normally, and their embryonic nervous system is also normal (Fig. 2, A, B, and K). In addition, the nervous system was normal in most of the O-fut1 R245A knock-in m/z embryos at 25°C (Fig. 2, C and K; 1/38 embryos showed the neurogenic phenotype). However, at 30°C, the O-fut1 R245A knock-in m/z embryos showed a highly penetrant neurogenic phenotype (Fig. 2, D and K, 15/15 embryos showed the neurogenic phenotype). This temperature-sensitive neurogenic phenotype has not been reported in previous studies using the transgenic O-fut1 R245A genomic locus (23,24).
To confirm that the temperature-sensitive neurogenic phenotype was not due to the degradation of O-Fut1 R245A at 30°C, we compared the protein levels of O-Fut1 and its derivative in wild-type and O-fut1 R245A knock-in homozygous cells, respectively. We generated genetic mosaic wing discs composed of somatic clones homozygous for O-fut1 4R6 or  (24). The level of the O-Fut1 protein derivative in the O-fut1 R245A knock-in homozygous cells was slightly lower than that of the wild-type O-Fut1 protein in wild-type cells at 25°C (Fig. 3, B-BЉ). However, no further reduction was observed at 30°C (Fig. 3, B-CЉ), suggesting that the temperature-sensitive phenotype of O-fut1 R245A knock-in was not caused by a decrease in O-Fut1 protein. Note that an anti-O-Fut1 antibody that can detect the endogenous O-Fut1 protein on Western blots was not available, so we were unable to confirm this conclusion by quantitative biochemical approaches.
A null mutation of O-fut1, O-fut1 4R6 , is known to result in a neurogenic phenotype even at 25°C (17). Therefore, the largely normal phenotype of the O-fut1 R245A knock-in m/z embryos at this temperature may be attributable to an enzyme activityindependent function of O-Fut1.
O-Fucose Monosaccharide on the EGF-like-repeats of N Has a Temperature-sensitive Role in N Signaling, Which Is Independent of Its Further GlcNAc Modification by Fng-There are two possible explanations for the temperature-sensitive property of the O-fut1 R245A knock-in mutation. First, the monosaccharide O-fucose modification of N may function in a temperaturesensitive manner. Second, the activity of the O-fut1 R245A knock-in gene and/or its product may be temperature-sensitive. To distinguish between these possibilities, we examined whether mutations of other genes that abolish the O-fucosylation of N also show a temperature-sensitive neurogenic phenotype.
We observed embryos homozygous for Gmd H78 or Gmer SH lacking their maternal contributions (Gmd H78 m/z and Gmer SH m/z). We found that most of the Gmd H78 m/z (2/19 embryos showed the neurogenic phenotype) and Gmer SH m/z (0/26 embryos showed the neurogenic phenotype) embryos had a normal nervous system at 25°C (Fig. 2, E, G, and K). However, as found in the O-fut1 R245A knock-in m/z embryos, at 30°C the neurogenic phenotype was highly penetrant in embryos with either the Gmd H78 m/z (18/18 neurogenic embryos) or the Gmer SH m/z (23/25 neurogenic embryos) mutation (Fig. 2, F, H, and K). In these embryos, various fucose modifications, including the fucosylation of N-glycan, should be abolished. However, the collective results obtained from the O-fut1 R245A knock-in m/z, Gmd H78 m/z, and Gmer SH m/z embryos indicated that the requirement for the monosaccharide O-fucose modification of N might be temperature-sensitive.
A contribution of fng to lateral inhibition has not been reported (22). However, to exclude the possibility that the temperature-sensitive neurogenic phenotype was caused by the lack of GlcNAc modification of O-fucose by Fng, we generated embryos homozygous for fng 13 and lacking its maternal contribution (fng 13 m/z). The fng 13 m/z embryos did not show the neurogenic phenotype at either 25°C (n ϭ 14) or 30°C (n ϭ 15) in any case examined (Fig. 2, I-K). This finding indicated that the absence of GlcNAc modification by Fng is irrelevant to the temperature-sensitive neurogenic phenotype observed in the O-fut1 R245A knock-in m/z, Gmd H78 m/z, and Gmer SH m/z embryos. Together, these findings indicate that the O-fucose monosaccharide modification of N is essential for N signaling during lateral inhibition at 30°C but not at 25°C.
O-Fucose Monosaccharide Modification of N Is Generally Required to Activate N Signaling at 30°C-We next examined whether the O-fucose monosaccharide modification of N is generally required for N signaling at 30°C. For this analysis, we generated somatic mosaic clones homozygous for O-fut1 R245A knock-in in the wing discs of third-instar larvae, using the FLP/FRT system (Fig. 4, B-CЈ). The expression of cut along the D/V boundary of wild-type wing discs is induced by N signaling (Fig. 4A). In the mosaic wing discs, cut was also ectopically expressed along the boundaries of the O-fut1 R245A knock-in somatic homozygous clones (marked by the absence of GFP) located in the dorsal compartment, at 25°C (Fig. 4, B and BЈ). This phenotype was similar to that of somatic mosaic clones homozygous for fng 13 (Fig. 4, D-EЈ) at 25 and 30°C (62). Therefore, the absence of monosaccharide O-fucose had a similar defect in cut activation at 25°C as did the absence of GlcNAc on O-fucose. This result is consistent with the previous conclusion that O-fucose monosaccharide does not have a specific function in N signaling at 25°C (24). However, at 30°C, the ectopic expression of cut along the boundaries of somatic mosaic clones homozygous for O-fut1 R245A knock-in was abolished ( various developmental contexts at 30°C. We also noted that the monosaccharide O-fucose was cell-autonomously required for the activation of N signaling at 30°C (Fig. 4, C and CЈ), which is consistent with the idea that the O-fucose monosaccharide attached to N is essential for NЈs activity at this temperature.
To obtain more evidence for this idea, we examined the formation of sensory organ precursors (SOPs) in the wing discs of third-instar larvae homozygous for O-fut1 R245A knock-in . A single SOP is selected from each proneural cluster in the wing disc, where SOPs arise in a well defined pattern at the third instar (63). In wild-type wing discs, the SOP formation pattern, observed by anti-Senseless (Sens) staining, was essentially the same from 18 to 30°C (Fig. 4, F-HЈ). In the wing discs of O-fut1 R245A knock-in homozygotes, SOPs did not form along the D/V boundary, because this SOP formation depends on the expression of wg, which is induced by the fng-dependent activation of N signaling (Fig. 4, I-K) (62). At 18 and 25°C, the number of SOPs elsewhere in the wing discs increased slightly (Fig. 4, I-JЈ). However, these SOPs markedly increased in these wing discs at 30°C (Fig. 4, K and KЈ). These results suggest that lateral inhibition was disrupted in the wing discs of O-fut1 R245A knock-in homozygotes in a temperature-sensitive manner, as found in the embryonic central nervous system.
We also examined the SOP formation in the wing discs in the absence of Gmer function. A trans-heterozygote of Gmer SH and a deletion mutant uncovering the Gmer locus survived until the third-instar larval stage, when we could isolate the wing discs to study SOP formation (33). At 25°C, the number of SOPs increased slightly in these wing discs, although they did not form along the D/V boundary due to the absence of N signaling there (Fig. 4, L and LЈ). Nevertheless, the SOPs, except for those located at the D/V boundary, noticeably increased at 30°C (Fig.  4, M and MЈ). We also confirmed that the SOP number was not affected in somatic mosaic clones homozygous for fng 13 in the wing discs of third-instar larvae, even at 30°C (Fig. 4, N-NЉ). Based on these results, we concluded that the monosaccharide form of O-fucose is generally required for N signaling at 30°C, although we could not exclude the possibility that some exceptions exist.  5). In wild-type wing discs, the overexpression of N FL , N ⌬E , or NICD resulted in the ectopic induction of cut at 30°C (Fig. 5, A, C, and E). However, the overexpression of N FL failed to induce ectopic cut expression in the somatic clones homozygous for O-fut1 R245A knock-in at 30°C (Fig. 5B)  overexpression of N ⌬E or NICD still induced the ectopic expression of cut in somatic mosaic clones homozygous for O-fut1 R245A knock-in at 30°C (Fig. 5, D and F) R245A knock-in in the wing discs of third-instar larvae (Fig. 6). In epithelial cells in wild-type wing discs, most of the N protein is detected in the sub-apical region and the adherence junctions at 25°C, although it is also detected in intracellular vesicles (26,64).
To detect total cellular N protein, we stained permeabilized wing discs with an antibody against NЈs intracellular domain (Fig. 6, A-BЈ), which showed N distribution at the optical plane corresponding to the middle of the epithelial cells at 25°C (Fig.  6, A and AЈ) and 30°C (Fig. 6, B and BЈ). At 25°C, there were no marked abnormalities in the intracellular N distribution in somatic mosaic clones homozygous for O-fut1 R245A knock-in (Fig. 6, A and AЈ); however, at 30°C, N accumulated inside the cell (Fig. 6, B and BЈ) although other membrane proteins, such as Dl and DE-cadherin (DE-Cad), were distributed normally under the same conditions (Fig. 6, E-EЉ).
To visualize N protein delivered to the epithelial cell surface at the sub-apical region, nonpermeabilized wing discs carrying

analysis of co-localization between N protein and various markers of intracellular compartments in O-fut1 R245A knock-in or O-fut1 4R6 mutant cells, or O-fut1 R245A knock-in and rumi double-mutant cells
Somatic mosaic clones homozygous or double homozygous for the indicated mutants were induced in wing discs of larvae cultured at 25 or 30°C, as indicated. Control cells heterozygous for O-fut1 R245A knock-in were in the wing discs with somatic mosaic clones of O-fut1 R245A knock-in homozygotes. The percentage of vesicles positive for each marker that were also positive for N is shown. The ER was detected with two markers, Boca and Pdi, at two different optical planes, sub-apical (SA) and medial. The ER showed an indistinct and shapeless structure at the sub-apical plane. N protein distribution with Boca or Pdi is indicated by ϩ (co-localizing) or Ϫ (not co-localizing). The MARCM system was used to express UAS-Lamp-HRP. All the N-positive vesicles and vesicles positive for each marker were counted in 32 ϫ 32-m square areas of representative Z-series confocal images. Three to six independent images were analyzed for each experiment. Asterisks show statistically significant results (p Ͻ 0.05), compared with control. Standard errors are shown, following Ϯ symbol. ND means not determined. somatic mosaic clones homozygous for O-fut1 R245A knock-in were stained with an antibody against the extracellular domain of N (Fig. 6, C-DЈ) (65). The amount of cell-surface N at the sub-apical plane was the same at 25 or 30°C in cells homozygous for O-fut1 R245A knock-in under this condition (Fig. 6, C-DЈ). Thus, even at 30°C, N delivery to the cell surface was not disrupted in cells homozygous for O-fut1 R245A knock-in , although N accumulated within these cells (Fig. 6, B and BЈ). However, intracellular and cell-surface N protein did not accumulate in somatic mosaic clones homozygous for fng 13 at 25 or 30°C, suggesting that the intracellular accumulation of N could not be attributed to the GlcNAc modification of O-fucose (data not shown).
We next determined the intracellular compartments in which N accumulated in the somatic mosaic clones homozygous for O-fut1 R245A knock-in at 30°C. We identified the ER (Boca and Pdi), Golgi (GM130), early endosomes (Hrs), late endosomes (Rab7), recycling endosomes (Rab11), and lysosomes (Lamp) by immunostaining (25,36,51,52,54). In wingdisc epithelial cells, the ER was detected either as an indistinct and shapeless structure (Fig. 6, F-FЉ and H-HЉ) or as dots (Fig.  6, G-GЉ and I-IЉ). The former was detected mostly in the cells' apical regions (Fig. 6, F-FЉ and H-HЉ), and the latter in more basal regions (Fig. 6, G-GЉ and I-IЉ). These structures were recognized by antibodies against two ER-marker proteins, Boca (Fig. 6, F-GЉ) and Pdi (Fig. 6, H-IЉ). To analyze the co-localization of N and markers for various intracellular compartments quantitatively, the percentage of vesicles positive for each marker that was also positive for N is shown in Table 1. At 30°C, the distribution of accumulated N protein overlapped partly with the indistinct and shapeless ER in somatic mosaic clones homozygous for O-fut1 R245A knock-in (white arrowhead in Fig. 6, F-FЉ, and H-HЉ, and Table 1), although hardly any N protein was detected in this structure in O-fut1 R245A knock-in /ϩ (control) cells (Fig. 6, F-FЉ, and H-HЉ, and Table 1). However, there was no increase in N-protein level in the dot-structure ER in the cells homozygous for O-fut1 R245A knock-in , compared with this ER in O-fut1 R245A knock-in /ϩ cells at 30°C (white arrowheads in Fig. 6, G-GЉ, and I-IЉ, and Table 1). Therefore, N accumulation in the indistinct and shapeless ER structures but not the dot-shaped ER structures may coincide with the loss of N signaling activity. Although N is reported to accumulate in the ER in somatic clones homozygous for O-fut1 4R6 , a null mutant of O-fut1 (17), there are conflicting reports (23,25). Here, we found that accumulated N partially overlapped with indistinct, shapeless ER structures, although there was no change of the amount of N in dot-shaped ER at 25 or 30°C (Fig. 7, A-Dٞ, and Table 1). This may account for the previous discrepancy in the ER accumulation of N reported in O-fut1 4R6 mutant cells. Nonpermeabilized staining did not detect N protein at the sub-apical cell-surface region in O-fut1 4R6 mutant epithelial cells at 25 or 30°C; this agreed with previously reported findings at 25°C (data not shown) (23,25).
We also observed that N protein was detected more often in early endosomes (23 Ϯ 2.8) and lysosomes (21 Ϯ 3.4) in somatic mosaic clones homozygous for O-fut1 R245A knock-in at 30°C, compared with those at 25°C or with control cells (13 Ϯ 2.4 and 4.4 Ϯ 0.59, respectively) (Fig. 7, E-FЉ and Table 1). These differences were statistically significant (t test, p Ͻ 0.05). In contrast, there was no marked difference in N distribution in the Golgi, late endosomes, or recycling endosomes of these cells at either temperature (Fig. 7, G-IЉ and Table 1). These data suggest that the O-fucose monosaccharide modification of N may be required for specific step(s) of NЈs trafficking at 30°C but not at 25°C. In addition, the distinct defects in NЈs trafficking found in O-fut1 R245A knock-in mutant cells versus O-fut1 4R6 mutant cells suggest that the nature of the defect in these two mutants is different (Table 1). These observations further support our idea that the temperature-sensitive phenotype of O-fut1 R245A knock-in is not due to a temperature-dependent loss of O-fut1 activity, associated with, for example, the breakdown of the O-Fut1 R245A mutant protein (Fig. 3). To detect possible alterations in the N protein in the O-fut1 R245A knock-in mutant, including changes in its stability and processing, we detected the N protein in O-fut1 R245A knock-in homozygous wing discs from the third-instar larvae cultured at 25 and 30°C, by Western blot (Fig. 8). However, we did not observe any marked differences in the amount or size of the N protein between the mutant and the wild-type wing discs (compare lanes 1 and 2 and lanes 3 and 4 in Fig. 8).

O-Fucose and O-Glucose Modifications of N Function
Redundantly-The EGF-like repeats of N have consensus sequences not only for O-fucose but also for O-glucose modifications (27). O-Glucose is added by Rumi, a protein O-glucosyltransferase in Drosophila (27,28). In the absence of rumi function, temperature-sensitive phenotypes associated with the loss of N signaling, which are similar to those of O-fut1 R245A knock-in , are observed (27,28). In somatic mosaic clones homozygous for rumi 44 , a null mutation, N accumulated intracellularly, which was also reminiscent of a defect in O-fut1 R245A knock-in homozygous cells. Therefore, we thought that the O-fucose monosaccharide modification of N might function redundantly with the O-glucose glycan modification of N in N signaling. To test this possibility, we generated a rumi and O-fut1 R245A knock-in double mutant and observed SOP formation in the wing discs of third-instar larvae. As described anti-Notch  above, the number of SOPs increased at 30°C in wing discs homozygous for rumi 44 , but it was normal at 18 and 25°C (Fig.  9, A-CЈ). This phenotype is very similar to that of O-fut1 R245A knock-in wing discs (Fig. 4, I-KЈ). However, the number of SOPs increased in the wing discs homozygous for rumi 44 and O-fut1 R245A knock-in even at 18 and 25°C, suggesting that N  signaling was abolished under these conditions (Fig. 9, D-EЈ).
These results suggest that the O-fucose monosaccharide and O-glucose glycans have a redundant function during lateral inhibition that controls the number of SOPs in the wing discs. Similar redundancy in the function of the O-fucose monosaccharide and O-glucose glycans was observed in the activation of N signaling along the D/V boundary of the wing disc at the third instar. The expression of cut along the D/V boundary disappeared at 30°C in somatic mosaic clones homozygous for rumi 44 (Fig. 10, B-BЉ), but cut's expression was normal at 25°C (Fig. 10, A-AЉ). The phenotype of the somatic mosaic clones homozygous for rumi 44 at 30°C was very similar to that of the somatic mosaic clones homozygous for O-fut1 R245A knock-in at 30°C (Fig. 4, C and CЈ). However, in somatic mosaic clones homozygous for rumi 44 and O-fut1 R245A knock-in , cut expression was abolished even at 25°C (Fig. 10, C-CЉ).
The O-fucose monosaccharide and O-glucose glycans on the EGF-like repeats of N also collaborated to increase the amount of N at the sub-apical plasma membrane. As reported previously, slight accumulations of N protein were found in cells homozygous for rumi 44 alone, at 25°C (Fig. 10, D-DЉ), although the amount of cell-surface N at the sub-apical region, detected by nonpermeabilized staining, was not altered in these cells (Fig. 10, E-EЉ) (27). However, the N-protein accumulations were more prominent in intracellular compartments in somatic mosaic clones homozygous for rumi 44 and O-fut1 R245A knock-in (Fig. 10, F-FЉ). In these double mutant cells, the accumulated N partially co-localized with ER markers (Fig.  10, G-GЉ, and Table 1). Curiously, nonpermeabilized staining for N revealed that the cell-surface N was diminished in these cells (Fig. 10, H-HЉ), but it was unaffected in cells homozygous for either O-fut1 R245A knock-in (Fig. 6, C-DЈ) or rumi 44 (Fig. 10, (27). Therefore, O-fucose monosaccharide and O-glucose glycans have redundant roles, and in their absence N was not delivered to or stably maintained at the plasma membrane.

DISCUSSION
The functions of the monosaccharide O-fucose on the EGFlike repeats of N have been vague. In Drosophila, the monosaccharide O-fucose was proposed to serve primarily as an acceptor for Fng, with O-Fut1 being required as a specific chaperon for N (23,24). However, in mammalian cells, the O-fucose modification was reported to be required for the optimal activation of N signaling under certain conditions (55). More recently, Fng-independent functions of fucosylation in Drosophila N signaling were reported, although these effects are relatively mild or cell type-specific (66,67). In this study, we revealed that the O-fucose monosaccharide, independent of GlcNAc modification by Fng, is essential for N signaling at 30°C in various developmental contexts in Drosophila. Moreover, in the absence of modifications with both O-fucose monosaccharide and O-glucose glycan, N signaling was abolished even at 25°C, although the loss of either modification alone caused this phenotype only at 29 -30°C (27). The loss of N signaling coincides with a severe reduction of N at the sub-apical plasma membrane and an accumulation of N in a particular fraction of the ER (Figs. 10, F-GЉ and 11A). These results suggest that O-fucose monosaccharide and O-glucose glycan modifications of N have redundant and collaborative roles in the activation of N signaling and the proper localization of N at the sub-apical plasma membrane (Fig. 11A). However, at this point we could not distinguish between whether N failed to reach the plasma membrane or it failed to be stably maintained on the membrane under this double mutant condition. Drosophila melanogaster Bombyx mori 5 1 0 1 5 2 0 2 5 3 0

Human Notch2
Mouse Notch2 Rat Notch2 The biochemical roles of the O-fucose monosaccharide and O-glucose glycan modifications are still elusive. However, a previous NMR study revealed that the fucose moiety directly functions as a "bridge" in the formation of an antiparallel ␤-sheet in EGF-like repeat 12 of mouse Notch1, stabilizing its structure through an interaction between the fucose and the peptide backbone of the EGF-like repeat (68). Therefore, we speculate that structural destabilization of each EGF-like repeat may affect the folding of the full-length N (Fig. 11B). Deficient N folding may explain our observation that N accumulated in the ER of cells homozygous for O-fut1 R245A knock-in at 30°C (Fig.  11B). More severe folding defects may occur in the absence of both O-fucose monosaccharide and O-glucose glycan modifications (Fig. 11B). It was previously shown that O-Fut1 and Rumi may be involved in quality control of the N protein, because they both only modify properly folded EGF-like repeats (69 -71). Therefore, it is possible that full-length N with EGFlike repeats lacking O-fucose monosaccharide and O-glucose glycan modifications may escape from its chaperon, leading to misfolding of the full-length N. However, at this point, the mechanism by which the potential destabilization of EGF-like repeats influences the global structure of full-length N is unknown. Moreover, it was previously shown that disrupting the O-fucosylation site in the 12th EGF-like repeat of mouse Notch1 does not affect the amount of cell-surface Notch1, but instead it reduces Notch1's activation, suggesting that the O-fucose modification of the 12th EGF-like repeat does not play a role in Notch1's folding (72). Therefore, it is possible that the O-fucose modification on other EGF-like repeats is required for the folding of full-length N and that the O-fucose modification of N has multiple functions in N signaling.
Considering that the O-fucose monosaccharide and O-glucose glycan modifications of the EGF-like repeats may function to increase the heat stability of full-length N, it is tempting to speculate that some sites for these modifications are specifically conserved in animals with higher body temperatures (homeothermic) but not in those with lower temperatures (poikilothermic). However, a comparison of the consensus sequences for these modifications failed to find a correlation between body temperature and the evolutionary conservation of these sites (Fig. 12).
A number of protein motifs contain multiple glycosylation sites for different types of glycan modifications (73). In this study, we found that the O-fucose monosaccharide and O-glucose glycan modifications of the EGF-like repeats of N have a redundant role, possibly in the folding of N. Thus, our results provide a case in which different types of glycosylation occurring at a protein motif may play a common and collaborative role.