Genetic Interactions Between Drosophila sialyltransferase and β1,4-N-acetylgalactosaminyltransferase-A Genes Indicate Their Involvement in the Same Pathway

Sialylated glycans play a prominent role in the Drosophila nervous system where they are involved in the regulation of neural transmission. However, the functional pathway of sialylation in invertebrates, including Drosophila, remains largely unknown. Here we used a combination of genetic and behavioral approaches to shed light on the Drosophila sialylation pathway. We examined genetic interactions between Drosophila sialyltransferase (DSiaT) and β1,4-N-acetylgalactosaminyltransferase (β4GalNAcT) genes. Our results indicated that β4GalNAcTA and DSiaT cooperate within the same functional pathway that regulates neural transmission. We found that β4GalNAcTA is epistatic to DSiaT. Our data suggest an intriguing possibility that β4GalNAcTA may participate in the biosynthesis of sialylated glycans.

Sialylation is a common type of protein glycosylation in vertebrate organisms (Schauer 2009;Varki and Schauer 2009). In mammals, sialylated glycans affect a plethora of protein interactions in the extracellular milieu, play a variety of important biological roles in development, and influence the physiology of many tissues and organs (Varki 2007(Varki , 2008. Sialylation is prominently enriched in the nervous system of vertebrates and is involved in crucial regulatory processes (Kleene and Schachner 2004;Varki 2008). At the same time, the role and biosynthesis of sialylated glycans in invertebrates is not well understood. Although glycoprotein sialylation is ubiquitous and abundant in mammalian organisms, it accounts for less than 0.1% of the total content of N-glycans in fruit flies . Despite their exceedingly low amount, sialylated glycans have an important function in the Drosophila central nervous system (CNS). Recent studies of Drosophila sialyltransferase (DSiaT), the enzyme mediating the last step in the sialylation pathway, indicated that sialylation regulates neural transmission and development, while representing a tightly controlled process limited to a subset of CNS neurons (Koles et al. 2004;Repnikova et al. 2010). However, the low level of sialylation makes its biochemical investigation in Drosophila a challenging task Koles et al. 2007). Here we used a genetic strategy, combined with the knowledge of glycan structures identified on fly glycoproteins, to shed light on the sialylation pathway in Drosophila.
The structure of Drosophila N-linked glycans indicates that galactose residues (Gal) of LacNAc termini (Galb1,4GlcNAc) serve as acceptors for sialylation Koles et al. 2007). Therefore, a galactosyltransferase attaching b1,4-linked Gal to N-glycans should be required for sialylation, and this enzyme is expected to cooperate with DSiaT in the regulation of neural transmission. However, so far no b1,4 galactosyltransferase (b4GalT) of this type has been identified in invertebrates. In mammalian cells, the corresponding Gal residues are added by one of the six b4GalTs (b4GalT1-6), the enzymes that function with apparent redundancy in modifying N-glycans (Hennet 2002). In Drosophila, the family of most closely related homologs of these b4GalTs consists of two enzymes, b1,4-N-acetylgalactosaminyltransferases A and B (b4GalNAcTA and b4GalNAcTB) (Haines and Irvine and Irvine 2005;Ramakrishnan and Qasba 2007). Despite the fact that b4GalNAcTA and b4GalNAcTB have similar in vitro activities, they have non-redundant functions in vivo (Chen et al. 2007;Haines and Irvine 2005;Haines and Stewart 2007;Stolz et al. 2008). The b4GalNAcTB enzyme modifies glycosphingolipids, and its function affects EGFR signaling during oogenesis (Chen et al. 2007;Stolz et al. 2008). Because b4GalNAcTA is capable of elongating bGlcNAc-termini of glycosphingolipids by adding b1,4-linked GalNAc in vitro, this glycosyltransferase may also have some role in glycosphingolipid biosynthesis in vivo (Chen et al. 2007;Johswich et al. 2009). However, this role is likely to be minor because b4GalNAcTA mutants have no discernable defects of glycosphingolipids, and the endogenous targets of b4GalNAcTA remain largely elusive (Chen et al. 2007;Johswich et al. 2009;Stolz et al. 2008). Mutations in b4GalNAcTA result in behavioral phenotypes, ultrastructural defects of muscles, and neuromuscular junction abnormalities (Chen et al. 2007;Haines and Irvine 2005;Haines and Stewart 2007).
Considering the close evolutionary relationship between b4GalNAcTA/B and vertebrate b4GalTs, we reasoned that these Drosophila enzymes might participate in the biosynthesis of N-linked glycans in vivo. This scenario entails a possibility that b4GalNAcTA/B is involved in the generation of glycan acceptors for DSiaT, and therefore, the mutations in these genes would affect DSiaT-mediated processes.
Here we test this hypothesis using genetic and behavioral approaches.

MATERIALS AND METHODS
Drosophila rearing and genetic stocks Flies were reared in a temperature-controlled (25°) and humidity-controlled (37%) environment at diurnal light conditions. We used the following mutant alleles for DSiaT and b4GalNAcTA/B genes: DSiaT s23 , b4GalNAcTA 4.1 , and b4GalNAcTB GT , designated here as DSiaT -, b4Gal-NAcTA -, and b4GalNAcTB -, respectively. These mutants represent lossof-function alleles and were previously described (Haines and Irvine 2005;Repnikova et al. 2010). Double mutants DSiaT -b4GalNAcTAwere generated by recombination. The DSiaTand b4GalNAcTAsingle and double mutants were confirmed by genomic PCR and sequencing for the presence of corresponding mutations: the DSiaT s23 allele includes two stop codons within the DSiaT coding region that truncate the encoded DSiaT protein sequence at positions Cys18 and Leu377 (Repnikova et al. 2010); the b4GalNAcTA 4.1 allele includes a 609 bp deletion that removes 113 bp upstream of the start codon along with the downstream region encoding the first 143 amino acids of b4GalNAcTA (Haines and Irvine 2005). The following PCR primers were used for genomic PCR reactions: for DSiaT s23 , St-gen-up (59-TTAAGTGCGAGCTAAAGGTCAATGC-39) and Sia-spe (59-CAACTAGTAATCGCGCTCCTCTTCAGTAG-39); for b4GalNAcTA 4.1 , TA-P2 (59-TGCCGCTGCTGTCAGGAT-39) and TA-P3 (59-AACGAAGCGATGAACTGTTTGAAT-39). The b4GalNAcTB GT mutation was confirmed by genomic PCR reactions with two sets of primers that amplify the genomic region of b4GalNAcTB disrupted by gene targeting, as described in Haines and Irvine (2005). The presence of b4GalNAcTBwas also corroborated by scoring the dorsal appendage fusion phenotype in homozygous mutants (Chen et al. 2007). The ectopic expression of b4GalNAcTA was induced using the UAS-GAL4 system (Brand et al. 1994) specifically in neurons (b4GalNAcTA Neuro ) or in muscles (b4GalNAcTA Muscle ) with C155-Gal4 and Dmef2-Gal4 drivers, respectively (Lin and Goodman 1994;Ranganayakulu et al. 1996). We used w 1118 Canton-S as a wild-type control in our experiments.

Heat-induced paralysis assays
We assayed five-day-old males for heat-induced paralysis using the previously described protocol (Repnikova et al. 2010). At least 20 flies were assayed for each data point. Unless indicated otherwise, the heatshock assays were performed with individual flies at 38°.

Statistical analysis
We used Student unpaired t-test with two-tailed distribution to assess the statistically significant differences between groups of related data.

RESULTS AND DISCUSSION
To test the possibility that b4GalNAcTA/B glycosyltransferases could be involved in the functional pathway mediated by sialylation, we examined genetic interactions between DSiaT and b4GalNAcTA/B genes. DSiaT mutations cause a characteristic temperature-sensitive paralysis phenotype (TS paralysis) (Repnikova et al. 2010). We used the TS paralysis assay to characterize genetic interactions between DSiaT and b4GalNAcTA/B. Whereas DSiaT mutants were consistently paralyzed within 7-10 min, neither b4GalNAcTA nor b4Gal-NAcTB mutants showed TS paralysis ( Figure 1). Strikingly, the analysis of double mutants revealed that the b4GalNAcTA mutation suppressed the paralysis phenotype of DSiaT mutants. At the same time, no significant interaction was observed between DSiaT and b4GalNAcTB (Figure 1). To characterize the interaction between DSiaT and b4GalNAcTA in more detail, we assayed the "kinetics" of paralysis by counting the number of paralyzed flies in a 3-min interval after transferring them to the restrictive temperature (38°). We found that b4GalNAcTA mutation semi-dominantly suppressed the phenotype of DSiaT mutants, which indicates that the DSiaT phenotype is very sensitive to the level of b4GalNAcTA activity (Figure 2). It was previously shown that b4GalNAcTA plays separate roles in neural and muscle cells (Haines and Stewart 2007). Thus, we investigated whether the neural or muscle-specific function of b4Gal-NAcTA is responsible for the interaction with DSiaT. We used a rescue strategy with the UAS-GAL4 ectopic expression system (Brand et al. 1994) to induce the expression of b4GalNAcTA specifically in neurons or muscle cells of DSiaT-b4GalNAcTA double mutants. These experiments revealed that the neuronal expression of b4GalNAcTA could suppress the effect of b4GalNAcTA mutation on the paralysis of DSiaT mutants, whereas the expression in muscles did not influence the phenotype of the double mutants ( Figure 3). Therefore, we concluded that it is the neuron-specific function of b4GalNAcTA that affects the paralysis of DSiaT mutants. Moreover, we found that the ectopic expression of b4GalNAcTA in the neurons of DSiaT mutants could further enhance the phenotype ( Figure 4A), which again highlighted that the TS paralysis of DSiaT mutants depends on the neural activity of b4GalNAcTA. The involvement of neural activity of b4GalNAcTA in the DSiaT-mediated pathway is consistent with the fact that DSiaT function is restricted to neurons at all developmental stages (Koles et al. 2004;Repnikova et al. 2010). Collectively, our results indicate that b4GalNAcTA and DSiaT cooperate within the same functional pathway that regulates neural excitability and that b4GalNAcTA is epistatic to DSiaT.
The b4GalNAcTA protein is the closest Drosophila homolog of vertebrate b4GalT1-6 (Haines and Irvine 2005). These b4GalTs are thought to originate from invertebrate b4GalNAcTs during evolution (Haines and Irvine 2005;Ramakrishnan and Qasba 2007). Interestingly, the donor substrate specificity of b4GalT and b4GalNAcT enzymes, including b4GalNAcTA, can be changed between Gal and GalNAc just by a single amino acid substitution in the active site Qasba 2002, 2007). Moreover, the donor and acceptor specificities of mammalian b4GalT1 can be modified through the binding of a protein cofactor, a-lactalbumin (Do et al. 1995;Hennet 2002). Such "flexibility" of the b4GalT/b4GalNAcT catalytic pocket capable of adjusting to different substrates suggests that the b4GalNAcTA specificity may be modified in vivo by a co-factor to synthesize LacNAc termini of Nlinked glycans. This possibility is further supported by the fact that the ability to bind a co-factor is an evolutionarily ancient feature of b4GalT/GalNAcT enzymes, and that this feature is also preserved for Drosophila b4GalNAcTA (Neeleman and van de Eijnden 1996; Ramakrishnan and Qasba 2007). The scenario that b4GalNAcTA may synthesize LacNAc termini, the potential acceptors for sialylation, is also consistent with the epistatic interaction between b4GalNAcTA and DSiaT that was revealed in our experiments. However, this scenario does not rule out that b4GalNAcTA has other functions that are not limited to its role in the DSiaT-mediated pathway. This is supported by the fact that b4GalNAcTA mutants exhibit some phenotypes apparently unrelated to the function of DSiaT, such as muscle abnormalities and the defects of neuromuscular junctions at muscle 6 during the larval stage (Haines and Stewart 2007;Repnikova et al. 2010).
The hypothesis that b4GalNAcTA may be involved in the biosynthesis of DSiaT acceptors predicts that the overexpression of b4GalNAcTA in the nervous system of wild-type flies, in the presence of endogenous DSiaT activity, may result in increased resistance to heat. This possibility is based on the fact that a limiting factor of the insect biosynthesis of complex N-glycans, including sialylated structures, is the high activity of the GlcNAcase fused lobes that could compete with galactosylation by removing GlcNAc from N-linked antennae prior to their elongation with Gal (Leonard et al. 2006;Watanabe et al. 2002). The upregulation of b4GalNAcTA might outcompete GlcNAcase, while protecting glycan termini from trimming by converting them to LacNAc, the substrate for further sialylation. Thus, in the presence of DSiaT, the overexpression of b4GalNAcTA may result in a more efficient biosynthesis of sialylated glycans, which in turn would increase the stability of neural transmission at elevated temperatures. Indeed, we observed that wild-type flies become more resistant to heat when b4GalNAcTA was ectopically overexpressed using a neuronal driver ( Figure 4B).
Taken together our results demonstrated that b4GalNAcTA genetically interacts with DSiaT, indicating that these genes cooperate in the Figure 3 The neuronal function of b4GalNAcTA is required for the paralysis phenotype of DSiaT mutants. Transgenic expression of b4GalNAcTA in neurons of b4GalNAcTA -DSiaTdouble mutants could restore the paralysis phenotype, whereas the expression in muscles had no effect on the phenotype of the double mutants. Mutant alleles used in these experiments were DSiaT s23 and b4GalNAcTA 4.1 . The ectopic expression of b4GalNAcTA was induced using UAS-GAL4 system specifically in neurons (b4GalNAcTA Neuro ) or in muscles (b4GalNAcTA Muscle ) with C155-Gal4 and Dmef2-Gal4 drivers, respectively. At least 20 males were assayed for each genotype. ÃÃ Significant difference with t-test (P , 0.01). Error bars represent SEM.

Figure 2
The kinetics of paralysis indicates that the GalNAcTA mutation semi-dominantly suppresses the phenotype of DSiaT mutants. The kinetics of paralysis was assayed using groups of 10 males as previously described (Repnikova et al. 2010). Each data point represents the average of three independent experiments. WT, wild-type control. Mutant alleles used in these experiments were DSiaT s23 and b4GalNAcTA 4.1 . Error bars represent SD. same functional pathway affecting neural transmission. Our data also suggest an intriguing possibility that b4GalNAcTA may participate in vivo in the biosynthesis of LacNAc termini of N-glycans, including sialylated glycans. In the light of the fact that the loss of b4GalNAcTA activity suppresses the mutant phenotype of DSiaT, it is tempting to speculate that sialic acids may play a masking role, capping LacNAc structures and thus regulating their interactions in the nervous system. However, other scenarios are also possible, and the mechanism of the interplay between b4GalNAcTA and sialylation pathway awaits further investigation using biochemical and in vivo approaches.

ACKNOWLEDGMENTS
We thank Nicola Haines and Ken Irvine for mutant and transgenic b4GalNAcTA/B strains, Michael Tiemeyer and Pradman Qasba for useful discussions, and Naosuke Nakamura for help with the experiments. We also thank Daria Panina and Courtney Caster for comments on the manuscript. This work was supported by NIH grants GM-069952 and NS-075534 (to V.M.P.)