Xylosylation of the Notch receptor preserves the balance between its activation by trans-Delta and inhibition by cis-ligands in Drosophila

The Drosophila glucoside xylosyltransferase Shams xylosylates Notch and inhibits Notch signaling in specific contexts including wing vein development. However, the molecular mechanisms underlying context-specificity of the shams phenotype is not known. Considering the role of Delta-Notch signaling in wing vein formation, we hypothesized that Shams might affect Delta-mediated Notch signaling in Drosophila. Using genetic interaction studies, we find that altering the gene dosage of Delta affects the wing vein and head bristle phenotypes caused by loss of Shams or by mutations in the Notch xylosylation sites. Clonal analysis suggests that loss of shams promotes Delta-mediated Notch activation. Further, Notch trans-activation by ectopically overexpressed Delta shows a dramatic increase upon loss of shams. In agreement with the above in vivo observations, cell aggregation and ligand-receptor binding assays show that shams knock-down in Notch-expressing cells enhances the binding between Notch and trans-Delta without affecting the binding between Notch and trans-Serrate and cell surface levels of Notch. Loss of Shams does not impair the cis-inhibition of Notch by ectopic overexpression of ligands in vivo or the interaction of Notch and cis-ligands in S2 cells. Nevertheless, removing one copy of endogenous ligands mimics the effects of loss shams on Notch trans-activation by ectopic Delta. This favors the notion that trans-activation of Notch by Delta overcomes the cis-inhibition of Notch by endogenous ligands upon loss of shams. Taken together, our data suggest that xylosylation selectively impedes the binding of Notch with trans-Delta without affecting its binding with cis-ligands and thereby assists in determining the balance of Notch receptor’s response to cis-ligands vs. trans-Delta during Drosophila development.

Introduction Notch signaling is a juxtacrine signaling pathway broadly used during animal development and tissue homeostasis [1]. Both Notch and its ligands are type I transmembrane proteins containing multiple epidermal growth factor-like (EGF) repeats, which are involved in Notchligand interactions and are the sites of several O-linked sugar modifications [2,3]. Interaction of the Delta/Serrate/Lag-2 (DSL) ligands on the surface of the signal-sending cell and the Notch receptor on the signal-receiving cell initiates Notch trans-activation and results in the induction of Notch downstream targets [4,5]. In contrast, binding of Notch and ligands expressed in the same cell can result in cis-inhibition of the pathway [6][7][8][9][10]. The balance between the cis and trans interactions of Notch with ligands is thought to determine whether each cell assumes a signal-sending or a signal-receiving role with regard to a given ligand [11,12].
The Notch ligands Delta and Serrate function redundantly in several contexts during fly development [13]. However, there are developmental processes in which Delta and Serrate show non-redundant roles [14][15][16][17]. For example, although Serrate plays a minor, fully redundant role during wing vein formation, Delta is the ligand primarily involved in wing vein development [13,15,18]. In this context, both trans-activatory and cis-inhibitory interactions between Delta and Notch are important for demarcation of the boundary between vein and intervein tissues [15,18]. Specifically, high levels of Delta in the central provein cells are thought to cis-inhibit Notch and allow the cells to assume the wing vein fate. Delta from central provein cells also trans-activates Notch in neighboring cells to prevent the adoption of vein fate and to establish the wing-intervein boundary. The haploinsufficient wing vein phenotypes exhibited by Notch +/and Delta +/animals and their mutual suppression in Notch +/-; Delta +/double-heterozygous flies [19] further indicate that both cis-inhibition and trans-activation of Notch by Delta are required for proper wing vein formation, and that altering the relative expression levels of these two proteins can tip the balance between cis-and transinteractions between them. However, it remains to be determined whether other mechanisms exist to regulate this balance.
Sugar modifications of the Notch receptors are known to affect Notch pathway activation at various steps, including folding, trafficking, ligand binding and potentially cleavage [20][21][22]. For instance, addition of N-acetylglucosamine (GlcNAc) to O-fucose residues on Notch EGF repeats by Fringe proteins differentially regulates the response of Notch to its ligands by regulating Notch-ligand interactions [23][24][25][26]. It is noteworthy that Fringe proteins promote the binding of Notch receptors to Delta ligands both in cis and in trans configurations [11] and as such, are not likely to be involved in regulating the balance between these opposing activities of Delta ligands. Another type of Notch sugar modification is the addition of O-glucose onto Notch EGF repeats by the protein O-glucosyltransferase 1 (Poglut1). In Drosophila, Poglut1 is encoded by rumi [27] and promotes Notch activation [27][28][29][30]. O-glucosylated EGF repeats also serve as a substrate for the addition of xylose by glucoside xylosyltransferase (GXYLT) enzymes [31,32]. Functional studies of the fly GXYLT, Shams, indicate that addition of xylose onto a specific subset of Notch EGF repeats (EGF16-20) negatively regulates Notch signaling in specific contexts, i.e. wing veins and head bristles [32]. Loss of xylose from Notch results in increased cell surface expression of Notch in the pupal wing but not in third instar larval wing discs [32], suggesting that additional mechanisms underlie the context-specificity of the shams loss-of-function phenotype.
Here, we provide evidence that Notch xylosylation by Shams decreases Delta-mediated trans-activation of Notch by reducing the binding of Notch to trans-Delta without affecting the binding of Notch to cis-ligands. The effect of loss of Shams on trans-activation of Notch by overexpressed Delta can be mimicked by decreasing the level of endogenous cis-ligands, suggesting that upon loss of Notch xylosylation, Delta trans-activation overcomes Notch cisinhibition by ligands. Altogether, our observations indicate that Shams regulates the balance between trans-activation and cis-inhibition of Notch by Delta to ensure optimal Notch activation in several contexts during Drosophila development.

Results
Increased gene dosage of Delta enhances the wing vein loss upon lack of Notch xylosylation To assess the role of Delta in the wing vein loss phenotype observed in shams mutants, we performed gene-dosage experiments using Delta genomic rescue transgenes [11]. Providing two additional genomic copies (4X) of Delta in a wild-type background does not generate any adult wing phenotypes at 30˚C (Fig 1A and 1B) or at room temperature [11]. The absence of phenotype is likely due to a simultaneous increase in the level of cis-inhibition and trans-activation of Notch upon increasing ligand levels. As previously reported, loss of shams results in a temperature-sensitive loss of distal part of adult wing veins L4 and L5 and a partial loss of the posterior cross-vein ( Fig 1D) [32]. In a shams Δ34/Df null background, providing one additional copy of Delta results in a fully penetrant, partial loss of wing vein L2 in addition to L4, L5 and posterior cross vein (Fig 1E), suggesting that shams mutants are sensitive to increased Delta levels compared to control animals. We performed similar genetic interaction experiments in flies harboring wild-type Notch gt-wt or xylosylation-deficient Notch gt-16_20 genomic transgenes [28]. Increasing the gene dosage of Delta does not result in wing vein loss in N +/+ ; N gt-wt /+ animals, which have three copies of the wild-type Notch (Fig 1G and 1H). However, providing an additional copy of Delta in N +/+ ; N gt-16_20 /+ animals results in a partially penetrant loss of the distal wing vein L5 (Fig 1J and 1K), which resembles the shams mutant phenotype at 25˚C [32]. Together, these data indicate that Notch signaling in shams mutants is sensitive to Delta levels and support the hypothesis that lack of Notch xylosylation affects Delta-mediated signaling. We also examined the effects of a Serrate transgene in similar experiments. Providing two additional copies of Serrate does not generate any wing vein loss in a wild-type background ( Fig  1C) [11]. Moreover, increasing Serrate gene dosage does not enhance the wing vein loss phenotype in a shams Δ34/Df null background ( Fig 1F). Finally, N +/+ ; N gt-wt /+ and N +/+ ; N gt-16_20 /+ animals do not exhibit wing vein loss upon addition of an extra copy of Serrate (Fig 1I and 1L). These results indicate that in the context of wing vein formation, lack of xylosylation does not render Notch sensitive to Serrate levels.
Removing one copy of Delta suppresses the loss of wing vein and head bristles in shams mutants Genetic interaction experiments were performed to examine the effect of decreasing Delta levels on the shams mutant phenotypes. Loss of one copy of Delta in Delta 9P/+ (Dl 9P/+ ) animals Providing an additional genomic copy of Notch (N gt-wt ) induces extra wing vein material at the L2 (red asterisk) and occasionally at the L5 wing vein indicative of Confluens phenotype. (H) N gt-wt/+ ; Dl gt-wt/+ animals exhibit no wing vein defects (compare to G). (I) N gt-wt/+ ; Ser gt-wt/+ animals do not show extra vein around L2 but occasionally show extra wing vein near L5. (J) Providing an additional genomic copy of Notch lacking the functional xylosylation sites (N gt-16_20 ) induces extra wing vein material near the L5 wing vein. (K) N gt-16_20/+ ; Dl gt-wt/+ animals exhibit a partial loss of L5 (71% penetrant, n = 17). (L) N gt-16_20/+ ; Ser gt-wt/+ wings are indistinguishable from N gt-16_20 wings (n = 15) (compare to J). https://doi.org/10.1371/journal.pgen.1006723.g001 Xylosylation modulates Delta-mediated Notch signaling results is extra wing vein material (Fig 2A) [33]. When one copy of Delta is removed in a shams Δ34/Δ34 background, the shams mutant wing vein loss is completely suppressed, and the extra wing vein phenotype of Dl 9P/+ is partially suppressed (Fig 2B and 2C). We have previously reported that loss of shams also results in the loss of post-vertical (PV) and ocellar (OC) bristles in the adult head [32]. Genetic interaction studies indicate that removing one copy of Delta in shams mutants rescues the loss of head bristles (Fig 2F) similar to the wing vein loss phenotype. Together, these observations support the notion that the shams loss-of-function phenotypes are due to increased Delta-mediated signaling. We also examined the effect of decreasing Serrate levels on the above-mentioned phenotypes (loss of wing vein and head bristles). Removing one copy of Serrate does not affect the loss of wing vein and head bristles in shams mutants (Fig 2D-2F). These observations indicate that altered Serrate-mediated signaling is not likely to contribute to shams loss-of-function phenotypes. Surprisingly, removing one copy of Serrate in shams Δ34/Δ34 mutant animals results in wing margin loss in some animals (S1A Fig; 21% penetrant, n = 73), which resembles the loss-of-function mutants of Notch and Serrate. Adding one copy of a Serrate genomic rescue transgene [11] fully rescues the wing margin phenotype without affecting the wing vein loss caused by the loss of shams (S1B Fig), indicating that the observed wing margin loss is indeed due to decreased Serrate levels in a shams mutant background. This indicates that in the wing margin, shams might play a redundant role which is revealed when one copy of Serrate is removed. Thus, a role for Shams in Serrate-mediated Notch signaling remains a distinct possibility in the context of wing margin formation.

Loss of shams enhances Delta-mediated Notch activation in Drosophila wing tissue
To examine the effects of loss of shams on the Notch activity in the wing imaginal disc, we performed clonal analysis in the 3rd instar larval wing discs by using the MARCM system [34]. We expressed Flippase with an Ultrabithorax enhancer (Ubx-FLP), which induces FRT-mediated recombination in the wing imaginal discs with a high efficiency [27,35]. We first examined the effect of loss of shams on Notch activity by staining for Notch targets Cut and Wingless (Wg) in shams Δ34 mutants clones. These mutant clones do not exhibit clear abnormalities in the expression of Cut and Wg when compared to adjacent wild-type or heterozygous cells (Fig 3A-3B'). However, slight broadening observed in Cut and Wg expression in shams Δ34 mutants clones supports a mild increase in Notch activation (Fig 3A-3B'). To further assess the effect of loss of shams on ligand-mediated Notch activity, we examined Cut expression in single-mutant clones for ligands (Ser rev6-1 and Dl RevF10 ) and double-mutant clones for ligands and shams (shams Δ34 Ser rev6-1 and Dl RevF10 shams Δ34 ). To analyze the effects of loss of shams on each ligand without interference from the other ligand, we focused on mutant clones which cross the dorsal-ventral (DV) boundary. Ser rev6-1 clones show a highly penetrant loss of Cut, except for the cells that directly abut the wild-type tissue (Fig 3C, 3C' and 3G; n = 40). This observation indicates that when cells around the DV boundary lack Serrate, they fail to activate the Notch target Cut despite the presence of Delta in the clones. In contrast, Cut-positive cells non-adjacent to the clone boundary are present in 42% of shams Δ34 Ser rev6-1 doublemutant clones which cross the DV boundary ( Fig 3D, 3D' and 3G; n = 31). Since Delta is the only ligand remaining in these clones, these observations are in agreement with the conclusion that loss of Shams promotes Delta-mediated Notch activation. Next, we performed a similar analysis on Delta single-mutant and Delta shams double-mutant clones, where Serrate is the only remaining ligand. Dl RevF10 single-mutant clones crossing the boundary showed a~65% penetrant loss of Cut expression (Fig 3E, 3E' and 3G; n = 29). Similarly, Dl RevF10 shams Δ34 double-mutant clones which cross the DV boundary exhibited a~64% penetrant loss of Cut expression ( Fig 3F, 3F' and 3G; n = 34). The comparable loss of Cut expression in Dl RevF10 and Dl RevF10 shams Δ34 clones suggests that loss of shams does not alter the activation of Notch by Serrate, which is the only remaining ligand inside these clones.
In addition to wing disc stainings, we examined the effects of shams Δ34 , Ser rev6-1 and Dl RevF10 single-mutant and shams Δ34 Ser rev6-1 and Dl RevF10 shams Δ34 double-mutant clones on the adult wing size. Although the clones in these adult wings are not marked, inspection of 3rd instar wing imaginal discs indicates that Ubx-FLP induces sizeable clones in 100% of the discs. Adult wings harboring shams Δ34 clones do not exhibit any wing margin phenotypes, although some animals show wing vein loss ( Fig 3I). Adult wings harboring clones of the Ser rev6-1 allele show severe wing scalloping (Fig 3J), in agreement with previous reports on this allele and other Serrate null alleles [36][37][38]. However, the degree of scalloping in adult wings harboring shams Δ34 Ser rev6-1 double-mutant clones is significantly less than that of the wings harboring Ser rev6-1 clones ( Fig 3K). Indeed, the average area of the wings harboring Serrate clones was significantly less than that of the wings harboring Serrate shams double-mutant clones (Fig 3N; P<0.05; n = 13 for each genotype). In contrast, the average area of wings harboring Dl RevF10 clones was not significantly different from the wings harboring Dl RevF10 shams Δ34 double mutant clones (Fig 3L-3N; P>0.05; n = 13 for each genotype). These results are in agreement with the data obtained from clonal analysis in larval wing discs and further support the notion that loss of shams partially suppresses the phenotypes caused by the loss of Serrate, likely by enhancing the Delta-mediated Notch activation in the developing wing.

Loss of Shams augments trans-activation of Notch by ectopic Delta
In some cell types, the decision whether Notch signaling is activated or inhibited is determined based on the relative levels of Notch trans-activation and cis-inhibition by its ligands [6][7][8]. Given the complex interplay between trans and cis functions of ligands and potential feedback mechanisms, we used the GAL4-UAS system [39] to overexpress Delta or Serrate along the anterior-posterior boundary of the developing 3rd instar wing imaginal discs and assessed the effects of loss of shams on the cis and trans effects of each ligand on Notch. When raised at 25˚C, animals overexpressing Delta within the Dpp expression domain (dpp>Dl) exhibit two phenotypes in wing imaginal discs: loss of the Cut-positive cells and disruption of the endogenous wing margin within the Dpp expression domain due to cis-inhibition ( Fig 4A and 4A'; arrowhead), and ectopic Cut-positive cells that flank the Dpp expression domain in the dorsalposterior quadrant due to trans-activation ( Fig 4A and 4A'; arrows). In a shams Δ34/Δ34 mutant background, there is no obvious effect on cis-inhibition within the Delta-expressing stripe ( Fig  4B and 4B'). However, we observe broad ectopic activation of Cut in a significant portion of the dorsal-anterior quadrant (Fig 4B and 4B'). These observations support the notion that loss of Shams promotes Delta-mediated trans-activation of Notch without affecting cis-inhibition. Of note, in late 2nd instar wing discs, the dpp-GAL4 driver shows a much broader anterior expression domain compared to 3rd instar wing discs (Fig 4C), providing a likely explanation for the wide Cut expression domain observed in the dorsal-anterior quadrant of dpp>Dl shams Δ34/Δ34 3rd instar discs.
To determine the effects of loss of Notch xylosylation on Serrate-mediated trans-activation and cis-inhibition of Notch, we performed similar overexpression experiments using a Tomato-tagged Serrate transgene (UAS-Ser Tom ) [40]. As expected, overexpression of Ser Tom results in cis-inhibition of Notch at the endogenous wing margin region and trans-activation of Notch outside of the Dpp expression domain in the ventral compartment ( Fig 4D and 4D'). In contrast to dpp>Dl, no changes in Cut activation in or outside of the Dpp expression domain was observed between dpp>Ser Tom and dpp>Ser Tom shams Δ34/Δ34 animals (Fig 4D-4E'). These data indicate that in the developing wing disc, loss of Shams does not affect the trans-activation and cis-inhibition of Notch mediated by overexpressed Serrate.

Notch trans-activation by dpp>Dl overcomes its cis-inhibition by endogenous ligands in shams mutants
Given the dramatic increase in the ability of dpp>Dl to trans-activate Notch upon loss of Shams, we decided to further explore the mechanism that prevents dpp>Dl from activating Notch in a shams +/+ background. In the sensory bristle lineage, decreasing the endogenous levels of Delta enhances the trans-activation of Notch by overexpressed Delta [9]. Accordingly, we examined the effects of removing one copy of endogenous Delta and/or Serrate on the expression of Cut in dpp>Dl larvae. In a Dl RevF10 or Ser rev2-11 heterozygous background,  (Fig 5C and 5C'). In addition, Delta-mediated cis-inhibition is significantly decreased, because many cells in the Dpp expression domain express Cut and the endogenous wing margin is restored (Fig 5C and 5C'; arrow). These observations indicate that cis-inhibition of Notch by endogenous ligands potently limits the ability of overexpressed Delta to trans-activate Notch outside of the Dpp domain and cooperates with overexpressed Delta to cis-inhibit Notch inside of the expression domain, in agreement with the previous report by Jacobsen et al [9]. Therefore, loss of shams and decreasing endogenous ligands both result in a similar increase in the ability of dpp>Dl to trans-activate Notch in 3rd instar wing discs, suggesting that in shams mutants, Delta mediated trans-activation of Notch is able to overcome the cis-inhibitory effect of endogenous ligands.

Cell-based aggregation and binding assays indicate increased binding between trans-Delta and Notch upon shams knockdown
To determine whether the effects of Shams on Delta-mediated signaling can be explained at the level of Notch-ligand binding, we performed S2 co-culture assays and assessed the effects of shams knock-down (KD) on the rate of aggregation formation and size of the aggregates formed between S2-Notch (S2-N) and S2-Delta (S2-Dl) or S2-Serrate Tom (S2-Ser Tom ) cells, which are indications for Notch-ligand binding [40,41]. For control experiments, we used plain S2 cells, which do not express Notch and Delta [41], and EGFP dsRNA. When co-cultured with plain S2 cells treated with shams dsRNA, S2-Dl cells form small aggregates which do not increase with time ( Fig 6A). Co-culture of S2-Dl and control S2-N cells for one minute resulted in the formation of small aggregates, which increased in number and size as time elapsed (Fig 6A). Co-culture of S2-Dl with Shams KD S2-N cells formed much larger cell aggregates at 5 and 15 minutes (Fig 6A).Quantification of cell aggregates containing more than 6 cells at 5 minutes of co-culture showed that the number of aggregates was significantly higher when S2-N cells were incubated with shams dsRNA (Fig 6B; P<0.01). qRT-PCR experiments indicate that shams dsRNA decreases shams mRNA levels greater than 80% in S2 cells (Fig 6C). These observations suggest that Shams decreases the binding between Notch and Delta in trans. Of note, co-culture experiments between S2-Ser Tom and S2-N cells showed that the size and the number of the S2-Ser Tom /S2-N aggregates did not change when shams levels were decreased (Fig 6A and 6B). These observations suggest that Shams KD does not affect the binding between Notch and trans-Serrate.
To more directly assess the effects of Shams KD on Notch-ligand binding, we used a quantitative receptor-ligand binding assay [23,42]. We incubated control and Shams KD S2-N cells with various concentrations of alkaline phosphatase (AP)-tagged ligand extracellular domains (Delta-AP or Serrate-AP) and asked whether Shams KD alters the binding of each ligand to S2-N cells in terms of the AP activity. Binding of each ligand-AP to plain S2 cells, which do not express Notch [41], was used as control. In agreement with the aggregation assays, treating S2-N cells with shams dsRNA significantly increased the binding of Delta-AP to these cells compared to EGFP dsRNA treated S2-N cells (Fig 6D). However, the amount of Serrate-AP bound to Shams KD and control S2-N cells was comparable at all four Serrate-AP concentrations tested (Fig 6D). These observations indicate that Shams negatively regulates Notch-Delta interaction but does not affect Notch-Serrate interaction.
Binding of Notch with trans-ligands can be affected by cell surface levels of Notch. Recent work has shown that combined loss of O-fucose and xylose residues from Notch alters its trafficking in 3rd instar wing discs, but loss of xylose by itself does not [43]. We have previously reported that loss of Shams or mutations that prevent the addition of xylose-glucose-O glycans to Notch EGF16-20 did not affect cell-surface expression of Notch in the 3rd instar wing discs, but enhanced the cell surface levels of Notch in the developing pupal wing disc [32]. Accordingly, we examined whether shams KD in S2 cells affects the cell surface expression of Notch.  Xylosylation modulates Delta-mediated Notch signaling observation indicates that shams KD does not affect the surface level of Delta in S2 cells. We also examined whether loss of shams affects surface expression of Delta in vivo. Detergent-free staining of 3rd instar wing imaginal discs did not show any significant difference in surface expression of Delta between shams Δ34 mutant clones and the neighboring wild-type and heterozygous cells (S3E Fig; n = 12 clones), indicating that Shams does not regulate the surface levels of Delta in 3rd instar wing imaginal discs.

Shams knockdown does not alter the inhibitory effect of cis-ligands on cell aggregation
To determine whether Shams modulates binding between Notch and cis-ligands, we asked whether shams KD affects the ability of ligands co-expressed with Notch to decrease the aggregation between the Notch-expressing cells and S2-Dl cells. To this end, we performed aggregation assays between S2-Dl cells and S2 cells transiently transfected with a Notch expression vector as control (S2-N transient ) or equal amounts of Notch and cis-ligand expression vectors (S2-N&Dl transient or S2-N&Ser transient ). The relative aggregation between S2-Dl and S2-N transient cells in the presence and absence of cis-ligands was used as an indication for the degree of cisinhibitory effect of each ligand. For example, if the number of aggregates between S2-Dl and S2-N&Ser transient cells is 25% of the number of aggregates between S2-Dl and S2-N transient cells (i.e. relative aggregation 25%), we would conclude that cis-Serrate was able to block the interaction between Notch and trans-Delta by 75%. Co-culture of S2-Dl and control S2-N transient cells formed small aggregates, which grew in size and number with time, similar to the co-culture between S2-Dl and stable S2-N cells (Figs 7A and 8A, compare to Fig 6A). The aggregation was dramatically decreased when S2-Dl cells were co-cultured with S2-N&Dl transient or S2-N&Ser transient cells (Figs 7A, 7B, 8A and 8B). This is most likely due to cis-inhibition of Notch by ligands expressed in the same cell, as shown previously in Drosophila and mammalian cellculture assays [11,40]. Co-culture of S2-Dl cells and shams KD S2-N transient cells formed aggregates more quickly and resulted in larger aggregates, recapitulating the results seen with stable S2-N cells (Figs 7A and 8A). Addition of cis-ligands to these cells (shams KD S2-N&Dl transient or S2-N&Ser transient ) also decreased their aggregation with S2-Dl cells (Figs 7A, 7B, 8A and 8B). Importantly, quantification of aggregates after five minutes of co-culture showed that the magnitude of this cis-inhibition was comparable to the cis-inhibition observed in co-culture between S2-Dl and control S2-N&Dl transient or S2-N&Ser transient cells, which were incubated with EGFP dsRNA instead of shams dsRNA (Figs 7C and 8C). These observations suggest that shams KD does not diminish the ability of cis-ligands to oppose the binding between Notch and trans-Delta.
One caveat of this experiment is that the relative level of cis-ligands expressed in these cells might be too high and therefore be able to inhibit trans-Delta/Notch binding irrespective of shams levels. To address this concern, we performed three additional sets of aggregation assays for each cis-ligand and successively decreased the ratio of cis-ligand expression plasmid to the Notch expression plasmid used in each set (0.5:1, 0.2:1, 0.1:1). To ensure similar baseline levels of trans-Dl/Notch binding, we used the same amount of Notch expression plasmid in all experiments and only changed the amount of cis-ligand expression plasmids. Aggregation between S2-Dl and S2-N transient cells was once again inhibited by expression of cis-ligands (both cis-Delta and cis-Serrate) even at lowest levels. Quantitation of the number of cell aggregates greater than six cells at five minutes of co-culture showed that the magnitude of inhibition by cis-ligands was concentration-dependent: the lower the ratio of cis-ligand to Notch expression plasmid, the lower the ability of cis-ligands to decrease the number of aggregates between S2-Dl and S2-N transient cells (Figs 7C and 8C). These data indicate that the level of cis-ligands used in our assays are not saturating. Under each cis-ligand/Notch ratio, the cis-inhibitory effect of each ligand on S2-Dl/S2-N transient aggregation was almost identical in cells treated with EGFP dsRNA and shams dsRNA (Figs 7C and 8C). These observations indicate that shams KD does not decrease the ability of cis-ligands to block the interaction between trans-Delta and Notch, and suggest that Shams does not affect the binding between Notch and cis-ligands. Taken together, these cell culture data support the conclusion that loss of Shams specifically enhances the binding of Notch with trans-Delta without affecting its binding with cis-ligands.

Discussion
Modifications of the extracellular domain of Notch receptor by glycosylation influence its activity in different contexts [20][21][22]. Xylosylation of the Notch receptor by the glycosyltransferase enzyme Shams has been reported to negatively regulate Notch signaling in Drosophila [32]. Loss of Notch xylosylation is associated with increased Notch expression at the cell surface in pupal wing disc but not in the larval wing imaginal discs [32]. However, the mechanisms underlying the tissue-specific phenotypes of shams and the effects of xylosylation on other steps of Notch signaling such as ligand binding and cis-inhibition are not known. In the present study, we provide several lines of evidence suggesting that the wing vein loss phenotype observed upon loss of Notch xylosylation results from a specific increase in Delta-mediated Notch activation. First, adding one copy of Delta enhances wing vein loss phenotype in shams mutant animals. Second, a Notch transgene with mutations in functional xylosylation sites results in wing vein loss when combined with an extra copy of Delta. Third, removing one copy of Delta suppresses the wing vein phenotype in shams mutant animals. Last, Shams KD in S2-N cells enhances their binding with AP-tagged Delta and their aggregation with S2-Dl cells without affecting cell surface levels of Notch. Since functional xylose residues reside in EGF16-20 of the Notch receptor [32], our work identifies the glycosylation of this domain as a novel mechanism for the modulation of Delta-mediated Notch activation in Drosophila. Although Notch EGF11-12 are required for binding of Notch to both ligands [2], a mutation in Notch EGF8 affects Notch-ligand binding and signaling in a ligand-selective manner [42]. Together with this report [42], our data suggest that distinct EGF repeats other than EGF11-12 are involved in preferential or exclusive modulation of the response of Notch to its ligands. Moreover, the differential effects of decreasing Notch xylosylation on the binding and response of Notch to trans-Delta versus cis-ligands suggest different domains or structural conformations of Notch might be involved in binding to trans-versus cis-ligands.
The increased Delta-mediated signaling upon loss of shams can theoretically be due to increased trans-activation, decreased cis-inhibition, or both. If the primary mechanism for activation of Notch signaling and loss of wing vein in shams mutants were decreased cis-inhibition by Delta, removing one copy of Delta should have enhanced the wing vein loss phenotype in shams mutants. On the contrary, the shams wing vein loss was suppressed upon Delta heterozygosity. Moreover, loss of shams dramatically enhances Notch trans-activation but does not affect Notch cis-inhibition by Delta in dpp>Dl animals, and shams KD promotes S2 cell aggregation mediated by Notch and trans-Delta without decreasing the inhibitory effect of cis-Delta on Notch in these assays. Therefore, although it is still possible that Shams plays a minor role in the interaction of Notch with cis-ligands in certain contexts, our data strongly suggest that the primary mechanism for the wing vein loss in shams mutants is an increase in Notch trans-activation by Delta, not a decrease in Notch cis-inhibition by Delta.
In line with a previous report in the bristle lineage [9], we find that removing one copy of Delta and/or Serrate in dpp>Dl animals results in ectopic activation of Notch in the dorsalanterior quadrant, indicating that cis-inhibition by endogenous ligands normally opposes the trans-activation of Notch by ectopic Delta in this region. These observations are in agreement with quantitative analyses indicating that the balance between the activity of trans-and cisligands determines whether a given cell assumes a signal-receiving state or not [11,12]. A similar increase in Delta-mediated trans-activation of Notch is seen in dpp>Dl animals upon loss of Shams despite the presence of endogenous ligands. Accordingly, we propose that Shams functions to regulate the balance between trans-activation of Notch by Delta and cis-inhibition of Notch by ligands, and that in the absence of shams, trans-activation of Notch by Delta overcomes the cis-inhibitory effects of ligands. Our cell aggregation assays suggest that Shams mediates this role by impeding the ability of Notch to bind Delta in trans, without altering the binding of Notch to cis-ligands. As shown in Fig 1, increasing the gene dosage of Delta by itself or combined with an additional copy of wild-type Notch does not result in wing vein loss, likely because the balance between the Notch and Delta levels and also the balance between the trans and cis activities of Delta are preserved. Based on our model, loss of xylose residues on Notch due to loss of Shams or mutations in biologically-relevant Shams target sites on Notch tips the balance between trans-and cis-activities of Delta in favor of trans-Delta, as evidenced by the net gain of Notch signaling and loss of wing vein in animals with three copies of Delta and simultaneous loss of Notch xylosylation.
Both Shams and Fringe regulate Notch signaling by adding carbohydrate residues to Olinked monosaccharides on Notch EGF repeats and generating disaccharides: xylose-glucose-O in the case of Shams, and GlcNAc-fucose-O in the case of Fringe [23,25,32,45]. Moreover, as shown here for Shams and previously for Fringe [23,46], both proteins regulate Notch-ligand interactions. However, the effects of these enzymes on the binding and response of Notch to Delta versus Serrate and to trans-Delta versus cis-Delta seem to be distinct from each other. Fringe promotes Delta-mediated trans-activation and simultaneously decreases Serrate-mediated trans-activation of Notch [26]. Moreover, it has recently been shown that Fringe proteins affect the trans and cis interaction of Notch with each ligand in the same direction, i.e., they promote Notch-Delta interactions both in cis and in trans, and inhibit Notch-Serrate interaction both in cis and in trans [11]. In contrast, our data indicate that Shams decreases the binding of Notch to and its activation by trans-Delta without affecting its interactions with cis-ligands. Further, our aggregation assays and most of our in vivo observations indicate that binding of Notch to and its activation by trans-Serrate is not significantly affected by Shams, although the appearance of a low-penetrance wing margin loss phenotype in shams -/-Serrate +/animals suggests that Shams might play a redundant role in Serrate-induced Notch signaling in some contexts. The functional differences between Shams and Fringe likely explain their distinct mutant phenotypes in the wing, i.e. loss of wing vein in the case of shams and wing vein thickening and loss of wing margin in the case of fringe mutants [47]. The distinct roles of Fringe and Shams in regulating Notch signaling along with the differentially distributed EGF repeats with Shams (xylose) versus Fringe (GlcNAc) elongation across the Notch extracellular domain [32,48] suggest that the combined function of the sugar modifications mediated by these enzymes ensures optimal level of Notch pathway activity in several contexts during fly development.
Ligand-receptor binding assays were performed as described previously [42] with minor modifications. In brief, S2 cells (2 x 10 6 cells/well) were transiently transfected with the constructs expressing the extracellular domain of Delta or Serrate fused to alkaline phosphatase (6.0 μg of either pMT-Delta-AP or pMT-Serrate-AP per well). After 24 hours, the transfected cells were induced by adding 0.7mM CuSO 4 to the media. After three days of induction, conditioned media were collected and the amount of AP was determined by quantifying AP activity in each medium by following the manufacturer's instructions (Phospha-Light System, Applied Biosystems) and using FLUOstar OPTIMA (BMG Labtech). In parallel, plain S2 or S2-N cells (2 x 10 6 cells/well) were induced by 0.7mM CuSO 4 after 24 hours of treatment with EGFP dsRNA or shams dsRNA. After three days of induction, cells (S2 or S2-N cells) were collected and incubated with conditioned media (containing defined concentrations of Delta-AP or Serrate-AP) for 90 minutes on rotary shaker. After three times washing and cell lysis, the endogenous AP activity was heat inactivated (60˚C for 10 min). The amount of AP-tagged ligand bound to S2-N cells was assayed as per the manufacturer's instructions (Phospha-Light System, Applied Biosystems) and using FLUOstar OPTIMA (BMG Labtech). Binding of each ligand (Delta-AP or Serrate-AP) with plain S2 cells (which lack endogenous Notch protein [41]) was used as control. Experiments were performed in triplicate and repeated three times.
Cells were incubated with CuSO 4 for 1-2 days and then used in aggregation assays. For each aggregation assay, 2.5 x 10 5 of the dsRNA-treated cells (S2, S2-N, S2-Dl) were mixed with 5 x 10 5 S2-Dl, S2-Ser Tom or S2-N cells (induced with 0.7 mM CuSO 4 for 3 hours prior to coculture) in a total volume of 200 μl medium in a 24-well plate.
For co-culture assays expressing cis-ligands, after 24 hours of treatment with either EGFP or shams dsRNA, S2 cells were transiently transfected with 2 μg total of either pBluescript (control), pBluescript and pMT-Notch (S2-N transient ), pMT-Notch and pMT-Delta (S2-N&Dl transient ), or pMT-Notch and pMT-Serrate (S2-N&Ser transient ) using 3 μL of FuGENE HD (Promega). One μg of pMT-Notch was used in all assays. The remaining 1 μg was either pBluescript alone, pMT-Delta alone, pMT-Serrate alone, or a mixture of 0.5 μg, 0.2 μg or 0.1 μg of pMT-Delta or pMT-Serrate and pBluescript (Figs 7C and 8C). For aggregation assays, 2.5 x 10 5 dsRNA treated cells were mixed with 5 x 10 5 S2-Dl cells (induced for 3 hours prior to co-culture with CuSO 4 ) in a total volume of 200μL in a 24-well plate. Co-cultured cells were gently shaken at 150 rpm to allow aggregation. Images of aggregate formation were taken at reported time points. The number of cell aggregates was quantified using a hemocytometer after 5 minutes of co-culture. Each assay was repeated at least three times. P-values were determined either by Student's ttest or by One-Way ANOVA with Tukey's multiple comparisons test.
Generation of dsRNA for S2 aggregation assays EGFP dsRNA primers: Forward primer-GAAATTAATACGACTCACTATAGGGGGTGAGCAAGGGCGAGGAG Reverse primer-GAAATTAATACGACTCACTATAGGGGGTCTTTGCTCAGGGCGG shams dsRNA Primers: Forward primer-TTAATACGACTCACTATAGGGGAGATGCTGTATGTGGACACGGAT Reverse Primer-TTAATACGACTCACTATAGGGGAGATCCGTGGATAACCTTAACGA Template DNAs were generated by PCR amplification from pAct-EGFP plasmid DNA or y w genomic DNA. Purified PCR products were used as template for in vitro transcription reactions using the T7 MEGAscript kit (Ambion). Double-stranded RNA was purified with RNeasy Mini kit (Qiagen).

qRT-PCR assays
shams and Rp49 (RpL32) mRNA expression in S2 cells treated with either EGFP or shams dsRNA were assayed by qRT-PCR using TaqMan One-Step RT-PCR Master Mix and TaqMan primers/probe sets (Life Technologies Dm02144576_g1 and Dm02151827_g). Relative shams mRNA levels were compared using the 2 -ΔΔCT method. P-values were determined by Student's t-test.