Previous studies indicated that CSAS is expressed in the nervous system and functions together with DSiaT in a pathway that affects neural transmission (Islam et al., 2013). However, the expression of CSAS has not been characterized in detail. To determine the expression of CSAS in different cells, we created a LexA reporter construct based on a genomic BAC clone that included the CSAS gene along with a large surrounding genomic region (Venken et al., 2009, see Materials and methods). We modified the BAC-CSAS by replacing part of the CSAS coding region with the sequence encoding LexA::p65 transcription activator (Pfeiffer et al., 2010) using recombineering (Venken et al., 2009) to generate a CSAS-LexA driver. This strategy has been useful to generate reporters with expression patterns that correspond to endogenous genes (Venken et al., 2009). We combined CSAS-LexA with LexAop2-mCD8-GFP and LexAop-GFP.nls reporters to label cell surfaces and nuclei of CSAS-expressing cells, respectively, and analyzed the expression pattern of CSAS at different developmental stages. Double-labeling experiments using Repo as a glial marker revealed that CSAS is expressed in many glial cells in the CNS throughout development and in adult flies (Figure 2A–B); in contrast, no expression was detected in neurons (Figure 2C). This is a surprising result, considering that DSiaT, the enzyme that functions downstream of CSAS in the sialylation pathway, is expressed only in neurons but not in glial cells (Repnikova et al., 2010). To confirm that CSAS-LexA expression recapitulates the endogenous expression of CSAS, we carried two sets of control experiments. First, we introduced the original BAC-CSAS clone as a transgene in flies and combined it with CSAS knockout, which resulted in full rescue of the temperature-sensitive (TS) paralysis phenotype of CSAS mutants (Figure 2—figure supplement 1). This supported the notion that the genomic clone includes all important regulatory elements to induce CSAS in endogenous manner. Second, we used CSAS-LexA to drive the transgenic expression of CSAS cDNA in CSAS mutants. This also fully rescued the TS paralysis, supporting the notion that CSAS-LexA recapitulates the CSAS endogenous expression (Figure 2—figure supplement 2). To further confirm that the expression of CSAS is indeed confined to the cells without DSiaT expression, we carried out double-labeling experiments to visualize CSAS and DSiaT-expressing cells simultaneously. To label DSiaT-expressing cells, we used a transgenic BAC-DSiaT-HA construct carrying a large genomic locus including the DSiaT gene modified with a 3xHA tag sequence to allow immunodetection (see Materials and methods). In agreement with previous studies (Repnikova et al., 2010), the expression of DSiaT-HA was detected only in differentiated neurons labeled by Elav, but not in neural progenitors expressing a neuroblast marker Deadpan (Figure 2—figure supplement 3). Importantly, we observed no overlap between the expression patterns of CSAS and DSiaT (Figure 2D–F). Taken together, these results show that CSAS and DSiaT are expressed in distinct cell populations within the CNS, glial cells, and neurons, respectively.

To investigate the cell-specific requirement of CSAS in the nervous system, we carried out rescue experiments using UAS-GAL4 ectopic expression system (Brand et al., 1994). CSAS function was shown to be required for normal neural transmission, while CSAS mutations cause locomotor defects and TS paralysis phenotype (Islam et al., 2013). Using cell-specific GAL4 drivers, we induced the transgenic expression of UAS-CSAS in CSAS homozygous mutants and assayed them for TS paralysis. Glial-specific expression of transgenic CSAS using drivers expressed in all glial cells (Repo-Gal4), ensheathing (Mz709-Gal4), astrocyte-like (dEAAT1-Gal4), neuropile ensheathing glia (R56F03-Gal4), or subperineurial glia (Gli-Gal4) could fully rescue the phenotype of CSAS mutants (Auld et al., 1995; Ito et al., 1995; Sepp et al., 2001; Rival et al., 2004; Doherty et al., 2009; Kremer et al., 2017), while the expression in neurons using a pan-neuronal driver (C155-GAL4) or other neuronal drivers broadly expressed in the nervous system (Mj85b-Gal4, 1407-Gal4) (Lin and Goodman, 1994; Dubnau et al., 2001; Kraft et al., 2016) did not result in rescue (Figure 3A, Figure 3—figure supplements 1–2). Interestingly, perineurial driver (R85G01-Gal4, Kremer et al., 2017) could partially rescue the phenotype, even though perineurial glia is separated from the brain by a tightly sealed layer of subperineurial cells maintaining the blood-brain barrier, and thus perineurial cells are not well poised to provide CMP-Sia for brain functions. This partial rescue is potentially explained by the fact that R85G01-Gal4 was found to be also expressed in a small number of cortex and astrocyte-like glial cells (Weiss et al., 2022). Taken together, our results demonstrated that CSAS expression in glial cells, but not in neurons, is sufficient to restore neural function in CSAS mutants.

Sialylation mutants have locomotion defects, such as an inability to promptly right themselves after falling upside down (Repnikova et al., 2010; Islam et al., 2013). Expression of UAS-CSAS in glial cells using Repo-Gal4 rescued this locomotion phenotype of CSAS mutants, while expression in neurons using C155-Gal4 did not result in rescue (Figure 3B). To assess the requirement of CSAS in synaptic transmission, we examined the function of motor neurons using electrophysiological assays at the neuromuscular junctions (NMJs). In sialylation mutants, larval motoneurons exhibit defects in excitability associated with a pronounced decrease of excitatory junction potentials (EJP) at NMJs (Repnikova et al., 2010; Islam et al., 2013). We analyzed EJPs in CSAS mutant third instar larvae and assessed the cell-specific requirement of CSAS by transgenic rescue. We used Repo-Gal4 and C164-Gal4 (Choi et al., 2004) drivers to induce the expression of UAS-CSAS in glial cells and motoneurons of CSAS mutants, respectively. CSAS expression in glial cells was sufficient to restore normal EJPs, however, CSAS expressed in motoneurons did not rescue neurotransmission defects (Figure 3C). These results provide compelling evidence that CSAS normally functions in glial cells but not in neurons, consistent with the results of behavioral assays.

We also examined the cell-specific requirement of CSAS by downregulating its function in different cells. To this end, we knocked down CSAS specifically in glial cells or neurons by expressing UAS-CSAS-RNAi using Repo-Gal4 or C155-Gal4, respectively. To potentiate the RNAi-mediated knockdown, we co-expressed UAS-CSAS-RNAi with UAS-Dcr-2 (Dietzl et al., 2007) and used a genetic background that was heterozygous for a CSAS deletion allele (Islam et al., 2013). The knockdown of CSAS in glial cells resulted in the TS paralysis phenotype similar to that of CSAS null mutants. No paralysis was induced by knocking down CSAS in neurons (Figure 3D). Taken together, our data show that CSAS is necessary and sufficient in glial cells to support normal neural functions.

Previous studies using an endogenously expressed tagged version of DSiaT demonstrated that DSiaT could be detected in neurons but not in glial cells (Repnikova et al., 2010). However, whether DSiaT is required specifically in neurons was not examined. Although this question can be in principle clarified by a rescue strategy using UAS-Gal4 system, this approach has been hampered by the ‘leaking’ expression of UAS-DSiaT that was able to rescue the DSiaT mutant phenotypes without the presence of a Gal4 driver. As an alternative approach, we investigated the cell-specific requirement of DSiaT by RNAi-mediated knockdown. To increase the efficiency of knockdown, we carried out the knockdown in heterozygotes for a DSiaT null allele, DSiaT^S23/+, which did not show the TS paralysis phenotype themselves. When DSiaT was downregulated by the expression of UAS-DSiaT-RNAi in neurons using C155-GAL4, the flies became paralytic at elevated temperature, showing the TS paralysis, a phenotype that recapitulated that of DSiaT null mutants. In contrast, DSiaT knockdown in glial cells did not cause the mutant phenotype (Figure 4). These results show that DSiaT function is required in neurons, consistent with the expression pattern of DSiaT.

Genetic and phenotypic analyses previously demonstrated that CSAS and DSiaT genes work in the same functional pathway affecting neural transmission (Repnikova et al., 2010; Islam et al., 2013). Although the biochemical activities of their protein products were characterized in vitro (Koles et al., 2004; Mertsalov et al., 2016), the roles of these genes in sialylation were not examined in vivo. Considering the unusual separation of CSAS and DSiaT expression patterns at the cellular level, we decided to test their requirements for the biosynthesis of sialylated glycans in vivo. First, we analyzed the production of CMP-Sia in wild-type flies and CSAS mutants by a liquid chromatography-mass spectrometry approaches (see Materials and methods). A prominent peak corresponding to CMP-Sia was detected in wild-type flies as shown before (van Scherpenzeel et al., 2021), while no CMP-Sia was found in CSAS mutants. Transgenic rescue using UAS-GAL4 system resulted in the restoration of CMP-Sia biosynthesis in the mutants (Figure 5A–B). These results revealed that the production of CMP-Sia in Drosophila specifically requires CSAS activity. Second, we examined N-glycans in CSAS and DSiaT mutants by mass spectrometry. Sialylated glycans are present in Drosophila at extremely low levels (Aoki et al., 2007; Koles et al., 2007). We decided to focus our analyses on third instar larval brains because CSAS and DSiaT show prominent expression during late larval stages (Figure 1B–D and Repnikova et al., 2010; Islam et al., 2013). We found N-glycan profiles were dominated by high- and pauci-mannose glycans in all genotypes, with hybrid and complex structures representing a small portion of the total N-glycome (Figure 5C–D), consistent with previous studies that analyzed N-glycans produced in embryos and adult heads (Aoki et al., 2007; Koles et al., 2007). Sialylated structures were detected in wild-type larval brains, but were not detected in CSAS or DSiaT mutants (Figure 5C–D). These results demonstrated that CSAS and DSiaT are essential for the biosynthesis of sialylated N-glycans in vivo, and that each of these genes plays a non-redundant role in this pathway.

The ectopic expression of CSAS induced by C155-Gal4 is predicted to generate CMP-Sia, the sugar donor required for DSiaT, in neurons. However, even though DSiaT is endogenously expressed and functions in neurons, the neuronal expression of UAS-CSAS could not rescue the phenotype of CSAS mutants (Figure 3A–B). This unexpected result may indicate that the sialylation pathway is blocked in neurons upstream of the CSAS-mediated step. For instance, neurons may have a limited capacity to synthesize Sia, a CSAS substrate, due to low activity of NANS (Figure 1), the evolutionarily conserved Sia synthase previously shown to be required for sialylation in Drosophila cultured cells (Granell et al., 2011). To test this hypothesis, we co-expressed UAS-NANS and UAS-CSAS transgenic constructs in neurons of CSAS mutants. Unlike the neuronal expression of UAS-CSAS alone, this co-expression could significantly rescue the phenotype of CSAS mutants (Figure 6). Taken together, these results indicated that the biosynthesis of Sia is indeed blocked in neurons due to a low level of endogenous NANS activity.

The facts that CSAS expression is restricted to glial cells while the biosynthesis of Sia is downregulated in neurons indicate that sialylation is tightly controlled in the nervous system, and that CSAS can potentially play a key regulatory role in the sialylation pathway. To shed light on this possibility, we decided to test how sensitive the neural functions are to heat stress conditions at different levels of CSAS activity. To this end, we assayed TS paralysis phenotype of flies with varied levels of CSAS, including CSAS null mutants, heterozygous mutants, wild-type, and CSAS overexpression genotypes. We found that the paralysis phenotype is highly sensitive to the level of CSAS activity. CSAS homozygous mutants exhibited the strongest phenotype, while the paralysis phenotype of CSAS heterozygotes was intermediate between that of homozygous mutants and wild-type flies with matching genetic backgrounds (Figure 7A). Remarkably, flies with CSAS transgenic overexpression were more tolerant to heat shock than wild-type counterparts (Figure 7B). This additional protection from heat-induced stress by upregulation of CSAS suggests that the endogenous level of CSAS may be a limiting factor in the sialylation. Together, these data support the hypothesis that sialylation can play a regulatory role in modulating neuronal transmission and promoting the stability of neural signaling during stress conditions.

Oxidative stress is known to affect the nervous system in various ways, with neuronal excitability abnormalities being among the immediate consequences of ROS overproduction (Wang et al., 2011). Considering that sialylation is involved in the regulation of neuronal excitability, we decided to test if the sialylation pathway plays a role in ameliorating the effect of oxidative stress. Using a paraquat-induced oxidative stress paradigm (see Materials and methods), we tested the viability of flies with different levels of CSAS activity, including CSAS mutant, rescue, wild-type, and overexpression genotypes. CSAS mutants were highly sensitive to oxidative stress as compared to matching ‘wild-type’ control. The glial-specific rescue resulted in significantly decreased mortality of mutants. Moreover, the overexpression of CSAS in glial cells of wild-type flies provided additional protection from oxidative stress, further increasing survivorship (Figure 8A–B). These results demonstrated that CSAS is required for protection from oxidative stress and suggested that the modulation of CSAS can underlie an endogenous mechanism that helps maintain neural functions during stress conditions.

Previous studies found that DSiaT and CSAS are required for normal neuronal excitability and revealed strong synergistic interactions of these genes with para, the Drosophila gene encoding voltage-gated sodium channel (Repnikova et al., 2010; Islam et al., 2013). These results suggested that the level of Para is potentially affected by sialylation. To test this hypothesis, we employed para-GFP, a Para allele that endogenously expresses a functional GFP-tagged version of the channel (Ravenscroft et al., 2020). The expression of Para-GFP was analyzed in DSiaT mutants using western blots. We analyzed flies on day 7 after eclosion, at the age when the TS paralysis phenotype of sialylation mutants becomes prominent (Repnikova et al., 2010; Islam et al., 2013). The level of Para-GFP was decreased in DSiaT mutants, as compared to a matching ‘wild-type’ control genotype, while the transgenic expression of DSiaT could restore the level of Para-GFP in the mutants (Figure 9). To test if this effect on Para could be due to changes in gene expression, we examined para mRNA by qRT-PCR. The level of para mRNA was not affected in DSiaT mutants, which indicated that the effect is posttranscriptional (Figure 9—figure supplement 1). As a control for conceivable non-specific effects on membrane proteins in DSiaT mutants, we also analyzed the expression of an irrelevant GFP-tagged membrane protein expressed in neurons (mCD8-GFP). The expression of the control protein was not affected in DSiaT mutants (Figure 9—figure supplement 1), which supported the hypothesis that DSiaT is specifically required to maintain the normal level of voltage-gated sodium channels in neurons.

The following strains used in the study were obtained from the Bloomington Stock Center, Indiana University: C155-Gal4 (#458), repo-Gal4 (#7415), 1407-Gal4 (#8751), Act-Gal4 (#3954), C164-Gal4 (#33807), UAS-DSiaT-RNAi (#44528), LexAop2-mCD8-GFP (#32203), LexAop-GFP.nls (#29954), LexAop-FLP (#55820). UAS-CSAS-RNAi (#101396) was obtained from the Vienna Drosophila Resource Center. Gli and MZ709 Gal4 lines were obtained from Michael Stern (Rice University), Mj85b-Gal4 was from Josh Dubnau (Stony Brook University). CSAS and DSiaT mutant alleles and UAS-DSiaT and UAS-CSAS transgenic constructs were previously described (Repnikova et al., 2010; Islam et al., 2013). All mutant alleles were multiply outcrossed to w¹¹¹⁸ Canton S which was used as a ‘wild-type’ genetic background. Unless indicated otherwise, Drosophila strains were reared in a controlled environment at 25°C in 60% humidity with 12 hr day/night light cycles. See Supplementary file 1 for detailed information on genotypes used in experiments. Genetic strains created in this study are available per request.

CSAS-LexA driver line was generated using a BAC clone (CH322-158A02, CHORI P[acman] library [Venken et al., 2009], BACPAC Genomics, CA, USA) with ~22.1 kb genomic fragment including the CSAS ORF (0.97 kb) approximately in the middle. We used recombineering approach (Venken et al., 2008) to replace the CSAS coding region of the last exon with the sequence encoding LexA::p65 transcription activator (Pfeiffer et al., 2010). The resulting BAC construct included intact upstream, downstream, and intron sequences of the CSAS ORF region and thus predicted to express LexA::p65 in the endogenous CSAS pattern when introduced in vivo. BAC-DSiaT-HA was generated using a similar strategy. The BAC clone CH322-7B13 carrying ~20.1 kb genomic fragment with DSiaT ORF (2.97 kb) approximately in the middle was used in recombineering to introduce in-frame a 3xHA tag coding sequence at the 3’ end of the DSiaT coding region. A short linker encoding 16 amino acids was included with the 3xHA tag sequence due to the presence of a LoxP site. The linker was previously shown not to significantly affect protein expression and localization (Venken et al., 2008). Outside of the tag insertion, BAC-DSiaT-HA contained unchanged genomic sequences and is expected to express a 3xHA-tagged DSiaT protein in the endogenous pattern in vivo. Transgenic Drosophila strains carrying CSAS-LexA and DSiaT-HA constructs were obtained by phiC31 integrase-mediated insertion (Bischof et al., 2007). We generated UAS-NANS construct by inserting NANS cDNA into the pUASTattB vector using standard molecular cloning techniques. A full-length NANS cDNA clone (IP20889) was obtained from the Drosophila Genomics Resource Center (Indiana University, Bloomington, IN, USA).

Behavioral assays were performed essentially as described previously (Repnikova et al., 2010; Islam et al., 2013). Briefly, flies were collected on the day of eclosure and aged for 5 days, during which they were transferred once on day 3 to a fresh-food vial. For TS paralysis assays, unless indicated otherwise, individual flies were transferred to empty vials and the vials were submerged in a 38°C temperature-controlled water bath. We defined paralysis as a condition when a fly is down and unable to stand and walk for at least 1 min. At least 20 flies were assayed for each genotype. For righting assays, individual flies were placed in a vial and allowed to acclimate for 10 min. The vials were tapped on a soft foam pad five times twice, the time they spent on their back was recorded. Two trials for each fly with a 10 min time between trials were performed. Same-sex animals (females, unless indicated otherwise) were compared in any particular behavioral experiment. To decrease the effect of differences in genetic background, we used genotypes that were outcrossed to the same reference background (w¹¹¹⁸ Canton S) and/or siblings from the same parents. At least 20 flies were assayed for every genotype, unless indicated otherwise.

Brains were dissected in ice-cold Ringer’s solution, washed, and fixed in fresh fixative solution (4% paraformaldehyde, 50 mM NaCl, 0.1 M Pipes, pH 7.2) for 20 min at room temperature with gentle agitation. Fixed tissues were analyzed by immunostaining and microscopy. Immunostaining was performed using fluorescent secondary antibodies essentially as described earlier (Lyalin et al., 2006). The following primary antibodies and corresponding dilutions were used for immunostaining: mouse anti-GFP 8H11 (1:100), anti-Repo 8D12 (1:10), rat anti-Elav 7E8A10 (1:10), mouse anti-Brp nc82 (1:10), all from Developmental Studies Hybridoma Bank; rabbit anti-GFP from Invitrogen (1:800), rat anti-HA from Roche (1:1000), rat anti-Dpn from Abcam (1:500). The following secondary antibodies were used: goat anti-rabbit and anti-mouse Alexa Fluor 546 and 488 (1:250), all from Invitrogen; donkey anti-mouse and anti-rabbit Cy3 (1:250) and FITC (1:150), from Jackson Laboratories. Stained samples were mounted on slides in Vectashield (Vector Laboratories) and imaged using Zeiss Axio Imager M2 fluorescence microscope with ApoTome module for optical sectioning or Zeiss 510 META Confocal microscope. Images were processed using Zeiss Zen and ImageJ software.

Intracellular recordings were performed from NMJs of dissected third instar larvae essentially as previously described (Islam et al., 2013). Briefly, free-moving larvae were dissected in ice-cold Ca²⁺-free HL3.1 buffer and EJP recordings were performed at 0.5 mM Ca²⁺ at room temperature. EJPs were evoked by directly stimulating the segmental nerve innervating a hemisegment A3 Muscle 6/7 using a glass capillary electrode at 0.2 Hz with stimulus pulses of 0.3 ms duration. We recorded from a single muscle per animal, while collecting and averaging 20 EJPs from each larva. There were no differences in input resistance, time constant τ, and resting membrane potential among different genotypes tested in the experiments shown, and the EJP amplitudes were corrected for nonlinear summation (Martin, 1955). Data were processed with Mini Analysis Program by Synaptosoft, Clampfit, and Excel.

To assess sensitivity to oxidative stress, we used a paraquat-induced stress paradigm as previously described (Zou et al., 2000) with some modifications. Briefly, flies were collected on the day of eclosure and aged in groups of 10 for 6–7 days on regular medium, while been transferred once to new vials on day 3. Before exposure to paraquat, flies were starved in empty vials for 6 hr, and then transferred into new vials containing 23 mm filter paper discs (Whatman G3) soaked with 40 mM paraquat in 5% sucrose, or 5% sucrose as a control. Flies were kept at a chamber with 100% humidity during starvation and paraquat exposure. They were transferred into new vials with fresh paraquat/sucrose every 48 hr. Mortality was assessed at defined time points as indicated. For statistical analyses, >10 independent experiments were performed for every genotype.

Sample preparation and LC-MS analyses were carried out essentially as previously described with minor modifications (Willems et al., 2019). Briefly, lyophilized flies were homogenized using a glass potter with 400 µL of ice-cold 75 mM ammonium carbonate (Honeywell, Fluka) buffer adjusted to a pH of 7.4 with acetic acid. The homogenates were transferred to 1.5 mL Eppendorf tubes and snap-frozen in liquid nitrogen to be further stored in –80°C. Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) was performed to each homogenate and three technical replicates were made using an equivalent of 270 µg protein content. Prior to metabolite extraction, 10 µM of ¹³C3-N-acetylneuraminic acid (Merck) was added as an internal standard to all replicates. Each sample was incubated with extraction solvent (2:2 acetonitrile/methanol [vol/vol]) for 5 min at –20°C. Samples were then centrifuged at 16,000 × g for 3 min followed by drying of the supernatant in a vacuum centrifuge at room temperature. The pellet was reconstituted in 100 μl of MilliQ.

Samples were analyzed using reverse-phase ion pair chromatography (Agilent Technologies 1290 Infinity) coupled to a triple-quadrupole mass spectrometer operating in negative ion mode (Agilent Technologies 6490 Triple Quad LC/MS). Chromatography was performed on an Acquity UPLC column (Waters, HSS T3 1.8 μm, 2.1×100 mm) using a gradient of mobile phase A (10 mM tributylamine, 12 mM acetic acid, 2 mM acetyl acetone, 3% MeOH in MilliQ) and mobile phase B (10 mM tributylamine, 12 mM acetic acid, 2 mM acetyl acetone, 3% MeOH in 50% acetonitrile and 50% MilliQ). The flow rate was 0.4 mL/min with a column temperature of 40°C and the injection volume was 2 µL per sample. CMP-Neu5Ac and several other metabolites were analyzed based on the indicated MRM transitions: CMP-Neu5Ac (613.14->78.9 m/z), UDP-HexNAc (606.07->273 m/z), Neu5Ac (308.1->87 m/z), glucosamine-6 phosphate (258->79 m/z), and KDN (3-deoxy-D-glycero-D-galacto-2-nonulosonic acid) (267.08->87 m/z). Peaks were annotated after comparison with peaks from commercial standards. For statistical analyses, data were obtained from three biological replicates, each including 100 adult flies (50 males plus 50 females) and analyzed using three technical repeats (two technical outliers were removed using the ROUT method; Motulsky and Brown, 2006). Data were plotted as a relative response to the internal standard (¹³C3-N-acetylneuraminic acid). Data analysis was performed using Agilent MassHunter Quantitative Analysis software for peak integration and GraphPad Prism software.

N-glycan isolation and analyses were carried out essentially as previously described (Aoki et al., 2007; Koles et al., 2007). Briefly, third instar larvae were rinsed several times in ice-cold PBS and brains were manually dissected on ice in PBS. After dissection, the brains were flash frozen in heptane on dry ice. Approximately 200 µl total volume of brains were collected for each genotype. Brains were homogenized and the resulting extracts were delipidated by organic solvent as previously described (Aoki et al., 2007). The resulting protein preparations were trypsinized and subjected to PNGaseF digestion to release N-glycans. Released N-glycans were analyzed as their permethylated derivatives by nanospray ionization mass spectrometry in positive ion mode. The permethylated N-glycans were dissolved in 50 µl of 1 mM sodium hydroxide in methanol/water (1:1) for infusion into an orbital ion trap mass spectrometer (Orbi-LTQ; Thermo Fisher Scientific) using a nanospray source at a syringe flow rate of 0.60 µl/min and capillary temperature set to 210°C. For fragmentation by collision-induced dissociation in MS/MS, a normalized collision energy of 35–40% was applied. Detection and relative quantification of the prevalence of individual N-glycans was accomplished using the total ion mapping (TIM) functionality of the Xcalibur software package version 2.0 (Thermo Fisher Scientific) as previously described (Aoki et al., 2007). For TIM, the m/z range from 600 to 2000 was automatically scanned in successive 2.8 mass unit windows with a window-to-window overlap of 0.8 mass units, which allowed the naturally occurring isotopes of each N-glycan species to be summed into a single response, thereby increasing detection sensitivity. Most N-glycan components were identified as singly, doubly, or triply charged, sodiated species (M+Na) in positive mode. Peaks for all charge states were summed for quantification. Graphic representations of N-glycan monosaccharide residues are consistent with the Symbol Nomenclature for Glycans (SNFG) as adopted by the glycobiology communities (Neelamegham et al., 2019). All raw mass spectrometric data were deposited at GlycoPost (Watanabe et al., 2021), accession # GPST000260.

Samples for western blot analysis were prepared essentially as previously described (Ravenscroft et al., 2020). Briefly, 10–15 flies were homogenized in 1× Laemmli gel loading buffer (30 µL per fly), supplemented with 5 mM EDTA, 1 mM PMSF, and 1× protease inhibitor cocktail (Sigma). Insoluble material was removed by centrifugation 15 min at 18,000 × g, +4°C. Proteins were separated using 4–20% SDS-PAGE and transferred onto nitrocellulose membrane (Bio-Rad). Membrane was blocked in 5% non-fat dry milk in 1× TBST, pH 8.0 and developed using rabbit anti-GFP (1:4000, Thermo G10362) and goat anti-rabbit-HRP (1:9000, Jackson ImmumoResearch 111-035-0030) primary and secondary antibodies, respectively. For protein loading control, membranes were stained with Ponceau S prior immunostaining. SuperSignal Pico PLUS Chemiluminescent substrate (Thermo 34577) was used to develop the western blots. Chemiluminescent signal was recorded on GE/Amersham I600 imager and quantified using ImageJ.

Total cellular RNA was isolated from 7-day-old adult flies using TRIzol reagent (Thermo Cat. #15596026) according to the manufacturer’s protocol. Ten to fifteen flies were used in each experiment. Quantity and quality of isolated RNA was evaluated spectrophotometrically and with agarose gel-electrophoresis. cDNA was synthesized using 2 µg of total RNA by Maxima First Strand cDNA Synthesis Kit (Thermo Cat. #K1671). qRT-PCR was performed using PowerUp SYBR Green Master Mix (Thermo Cat. #A25741) on Bio-Rad CFX96 Real-Time PCR instrument. Ct values were determined with Bio-Rad’s CFX Manager software. Relative expression of para was assessed by 2^-ΔΔCt method using α-tubulin as a control (Livak and Schmittgen, 2001; Ponton et al., 2011).

The following primers were used in these experiments:

All experiments were performed at least three times (biological replicates), unless indicated otherwise in text. Whenever it was possible, each experiment included at least three technical repeats. Unless indicated otherwise, data points shown in all figures represent different biological replicates. Statistical analyses in experiments with multiple groups of data were performed by one-way ANOVA with Tukey post hoc comparisons. Survival curves were compared using log-range tests. In all figures, 1, 2, and 3 asterisks represent p values of <0.05, <0.01, and <0.001, respectively; NS indicates that no significant differences were found (p>0.05). Details on statistical analysis are included in figure legends, text, and supplementary materials. The sample size required for reliable statistical analyses was determined empirically, based on our previous experience and knowledge of the system and the assays. No power analysis was used to predetermine the sample size. GraphPad Prism software was used for statistical analyses.