Functional Roles for β1,4-N-Acetlygalactosaminyltransferase-A in Drosophila Larval Neurons and Muscles
Nicola Haines, Bryan A. Stewart

Abstract

Adult Drosophila mutant for the glycosyltransferase β1,4-N-acetlygalactosaminyltransferase-A (β4GalNAcTA) display an abnormal locomotion phenotype, indicating a role for this enzyme, and the glycan structures that it generates, in the neuromuscular system. To investigate the functional role of this enzyme in more detail, we turned to the accessible larval neuromuscular system and report here that larvae mutant for β4GalNAcTA display distinct nerve and muscle phenotypes. Mutant larvae exhibit abnormal backward crawling, reductions in nerve terminal bouton number, decreased spontaneous transmitter-release frequency, and short, wide muscles. This muscle shape change appears to result from hypercontraction since the individual sarcomeres are shorter in mutant muscles. Analysis of muscle calcium signals showed altered calcium handling in the mutant, suggesting a mechanism by which hypercontraction could occur. All of these phenotypes can be rescued by a transgene carrying the β4GalNAcTA genomic region. Tissue-specific expression, using the Gal4-UAS system, reveals that neural expression rescues the mutant crawling phenotype, while muscle expression rescues the muscle defect. Tissue-specific expression did not appear to rescue the decrease in neuromuscular junction bouton number, suggesting that this defect arises from cooperation between nerve and muscle. Altogether, these results suggest that β4GalNAcTA has at least three distinct functional roles.

MANY glycosyltransferases are biochemically well characterized, but identifying the functional role of the glycan structures that they generate is more challenging. Recently, genetic studies in the model system Drosophila melanogaster have resulted in significant contributions to our understanding of glycan function (Nybakken and Perrimon 2002; Schwientek et al. 2002; Haines and Irvine 2003; Wandall et al. 2003; Haltiwanger and Lowe 2004). Many Drosophila glycosyltransferases show significant homology to vertebrae enzymes but typically the Drosophila genome encodes fewer homologous family members, making genetic studies in this organism more informative.

The vertebrate β1,4-galactosyltransferase (β4GalT) family has six family members; in Drosophila two enzymes, β4GalNAcTA and β4GalNAcTB, show strong sequence homology to this family (Haines and Irvine 2005). The biochemical activity and structure of the vertebrate enzymes have been well studied (Lo et al. 1998; Amado et al. 1999; Furukawa and Sato 1999; Gastinel et al. 1999; Ramakrishnan et al. 2002) and all are able to transfer galactose (Gal) to N-acetylglucosamine (GlcNAc), generating LacNAc (Galβ1,4-GlcNAc). This structure is a common feature of mammalian glycoproteins and glycolipids (Amado et al. 1999; Varki et al. 1999), and it is also a major substrate for sialyltransferases (Varki et al. 1999), thus making it essential for the generation of a diverse array of sialylated glycans. A number of studies indicate functional roles for LacNAc in vertebrates (Axford 1999; Chen et al. 2001; Tozawa et al. 2001); however, studying the role of this structure genetically is complicated by redundancy between the enzymes. Mice with targeted deletions in β4GalT-1 develop almost normally until birth, but exhibit neonatal lethality (Asano et al. 1997; Lu et al. 1997). It seems likely that the expression of the remaining β4GalT family members may mask a more severe phenotypic affect (Chen et al. 2006).

The Drosophila β4GalNAcT's, despite strong sequence homology to the vertebrate β4GalT's, utilize a different donor sugar. In vitro assays using the Drosophila enzymes, or other invertebrate homologs of the β4GalT, demonstrate that these enzymes transfer GalNAc to a terminal GlcNAc, generating LacdiNAc (GalNAcβ1,4-GlcNAc) rather than LacNAc (Kawar et al. 2002; Vadaie and Jarvis 2004; Haines and Irvine 2005). While these assays demonstrate that the Drosophila enzymes are able to generate LacdiNAc, comparisons of wild-type and mutant glycan structures will be required to determine their in vivo biochemical activity. LacdiNAc has been found on N-glycans from Drosophila-cultured cells and on Drosophila glycolipids (Seppo et al. 2000; Koles et al. 2004). However, characterization of the major N- and O-linked glycans from Drosophila embryos failed to detect LacdiNAc (North et al. 2006), perhaps suggesting that these structures represent only a small fraction of total glycan structures present in these embryos. Two glycosyltransferases that are predicted to act prior to the synthesis of a LacdiNAc linkage in Drosophila glycolipids (Seppo et al. 2000) are encoded by egghead (egh) and brainiac (brn) (Schwientek et al. 2002; Wandall et al. 2003, 2005). Mutation of these genes results in maternal-effect neurogenic phenotypes and embryonic lethality (Goode et al. 1996a,b).

The homology between the characterized invertebrate β4GalNAcT's and vertebrate β4GalT's raises the possibility that invertebrate LacdiNAc structures carry out biological roles fulfilled by LacNAc in vertebrates. This possibility is appealing, given the characterization of a Drosophila sialyltransferase that demonstrates acceptor specificity toward LacdiNAc at the nonreducing termini of oligosaccharides and glycoproteins but not glycolipid acceptors (Koles et al. 2004). This sialyltransferase activity predicts the existence of Siaα2,6-GalNAcβ1,4-GlcNAc structures on Drosophila N-glycans, and indeed this glycan structure has been identified on the N-glycans of the Drosophila sialyltransferase itself (Koles et al. 2004). Vertebrate LacNAc structures are major substrates for sialyltransferases (Varki et al. 1999). This Drosophila structure has obvious structural similarities to sialylated LacNAc. Investigating the phenotypic consequences of mutants in Drosophila β4GalNAcTs will lead to an understanding of the functional role of these glycan structures and allow functional comparisons with vertebrate glycans to be made.

Previously, null alleles were generated in β4GalNAcTA and β4GalNAcTB (Haines and Irvine 2005). Mutants in β4GalNAcTB have no obvious phenotype but mutants in β4GalNAcTA show an adult behavioral/locomotion phenotype. The flies are uncoordinated, have difficulty righting themselves after a fall, and have a low climbing index, suggesting that β4GalNAcTA plays a role in the adult neuromuscular system (Haines and Irvine 2005). As β4GalNAcTA is widely expressed throughout Drosophila development (Haines and Irvine 2005), we further investigated the functional role of this enzyme at the larval neuromuscular system. We find that β4GalNAcTA has at least three distinct, tissue-specific roles in neurons and muscles. These findings indicate that substrates are present in both neural and muscle cells that require glycosylation by β4GalNAcTA for normal function.

MATERIALS AND METHODS

Drosophila stocks:

The Drosophila stocks were β4GalNAcTA4.1, Df(2R)β4GalNAcTA[20.1], and p[β4GalNAcTA+] as described previously (Haines and Irvine 2005). Upstream activating sequence (UAS)-β4GalNAcTA (N. Haines and K. D. Irvine, unpublished results), a KpnI–XbaI fragment containing the entire coding region of β4GalNAcTA, was excised from the expression construct pMTWB-β4GalNAcTA:V5:His (Haines and Irvine 2005) and ligated into pUAST. The resulting construct pUAS-β4GalNAcTA was transformed into Drosophila using standard techniques. A third chromosome insertion was used in this study.

UAS-G-CaMP (Wang et al. 2004), 24B-Gal4, B185-Gal4 (Davis et al. 1997), and elav-Gal4 (Brand and Perrimon 1993) have been previously described. All crosses were carried out on standard Drosophila media at 25°. Crosses were set up and adult flies were transferred each day to new vials to avoid crowding.

Larval crawling assays:

Third instar larvae were separated from media with 25% sucrose solution, rinsed with distilled water, and then placed individually onto 10-cm agar plates. After 10 min with no handling, the plates were placed under a dissecting microscope equipped with a Nikon Coolpix 995 digital movie camera. The microscope light was not used; illumination was from room ceiling lights. Images were collected at a rate of 10/sec with an ATI all-in-wonder 7500 video card and ATI multimedia software. Thirty seconds of movie was recorded for each larvae, after which time wild-type larvae generally moved from the field of view. Movies were later viewed and the number of forward and backward body-wall contractions were counted.

Immunostaining:

Third instar larvae were dissected along the dorsal midline in zero calcium HL3 saline (Stewart et al. 1994). Upon removal of digestive tract, fat bodies, and the main trachea, the preparations were fixed in 4% formaldehyde in phosphate buffered saline (PBS) for 10 min The tissue was then washed in PBS, 0.1% triton, and 0.5% BSA and stained. Tetramethyl rhodamine isothiocyanate (TRITC)–phalloidin (Sigma, St. Louis) was used at 1 μg/ml, rabbit antisynaptotagmin antibody 1:2000 (gift from N. Reist), and Texas red AffiniPure goat antihorseradish peroxidase (anti-HRP; Jackson ImmunoResearch, West Grove, PA) at 1:100. Secondary antibody was goat anti-rabbit Alexafluor488 (Molecular Probes, Eugene, OR) at 1:500. Stained preparations were mounted in 90% glycerol in PBS.

Neuromuscular junctions analysis:

Larval preparations stained with anti-HRP and antisynaptotagmin were imaged with ×40 magnification on a Zeiss LSM 510 confocal microscope. Images of the neuromuscular junctions (NMJ) at muscle 6 and 7 from segments 3 or 4 were exported to imageJ (National Institutes of Health). Total length of the NMJ was obtained by measuring all the branches from the HRP-stained image, and the number of branches off the main branch were counted to obtain the branch number. Images were viewed at ×2 zoom and the number of boutons were counted by assessing synaptotagmin staining.

Muscle measurements:

TRITC–phalloidin-stained and -mounted larval preparations were viewed with a ×10 lens on a Nikon E600FN microscope equipped with a Hamamatsu (Bridgewater, NJ) Orca ER CCD camera. Images showing the hemi-segment musculature from segments 3 or 4 were obtained from the camera with simplePCI (Compix, Mars, PA) software and exported to imageJ, which was used to make measurements directly from the images. For sarcomere measurements, preparations were viewed with ×63 magnification on a Zeiss LSM 510 confocal microscope. Muscle 6 from segments 3 or 4 was examined to ensure that there was no obvious damage to the muscle and a Z-stack of 8–12 sections 0.6 μm apart through the central part of muscle 6 was collected. Stacks were exported to imageJ, and, as described in the text and in Figure 4, regions of the muscle showing clear, aligned phalloidin banding through the Z-stack were selected. ImageJ was used to obtain gray-scale intensity across these sarcomeres. Peak-to-peak distances were obtained for an average of eight consecutive sarcomers/muscle. No more than two muscles from any one larva were used.

Physiology:

Physiology assays were carried out in HL3 saline with 1 mm calcium, as previously described (Stewart and McLean 2004). Excitatory junctional potentials (EJPs) were analyzed using the cursor options of Clampfit. Miniature excitatory junctional potentials (mEJPs) were analyzed with mini-analysis (Synaptosoft).

Calcium imaging:

Third instar larvae were dissected along the dorsal midline in HL3 saline with 1 mm Ca. Digestive tract, fat bodies, main trachea, and the central nervous system were removed. Preparations were rinsed with HL3 and 1 mm Ca, and 600 μl of this solution was placed in the dissection chamber. Images of muscle 6 from abdominal segment 3 or 4 were taken using a ×60 water lens on a Nikon microscope equipped with a Hamamatsu Orca ER CCD camera. Images were taken with a 0.01-sec exposure every 2 sec and collected with simple PCI (Compix) software, which also controlled the camera shutter. Directly following the third image capture (6 sec) 10 μl 3 m KCl was added to the chamber (final concentration 50 mm K), G-CaMP signal slowly increased in response to the potassium addition and little muscle contraction occurred, allowing the muscle to stay in focus throughout the experiment. In total, 35 images were collected. Image series were exported as avi files to imageJ. A 50 × 50 pixel region of the center of muscle 6 was selected from each data set that avoided nonmuscle regions, such as areas where the trachea was visible, and the intensity of this region in each frame of the series was measured. These intensities were normalized to the average intensity of the image before addition of potassium and are expressed as the change in fluorescence/initial fluorescence (ΔF/F). However, the mean intensity of the original prepotassium images was similar for each genotype; the mean initial gray-scale intensity for each genotype was the following: control, 110.4 and β4GalNAcTA4.1, 110.8. Genotypes were UAS-G-CaMP/+; β4GalNAcTA4.1/+; 24Bgal4/+ control; UAS-G-CaMP/+; β4GalNAcTA4.14GalNAcTA4.1; and 24B-Gal4/+.

Statistics:

Statistical comparisons were made using ANOVA or unpaired t-tests as indicated. All error bars represent SEM.

RESULTS

β4GalNAcTA mutant larvae show frequent backward crawling:

Wild-type Drosophila larvae show progressive sustained forward movement and rarely move backward. To initiate forward crawling, the posterior body-wall muscles contract and this contraction propagates in a wave toward the anterior, terminating on extension of the mouth hooks (see supplemental movie A at http://www.genetics.org/supplemental/). When backward crawling is observed, muscle contraction begins at the anterior and progresses toward the posterior of the larvae. We examined the crawling behavior of third instar larvae mutant for β4GalNAcTA. Short movies were made of larvae crawling along an agar plate, and the number of forward and backward body-wall contractions (BWC) were counted and BWC/minute were obtained and averaged for each genotype.

Larvae mutant for β4GalNAcTA display a reduced number of forward body-wall contractions and show frequent backward crawling (Figure 1 and supplemental movie B at http://www.genetics.org/supplemental/). Forward crawling in mutants is frequently interrupted with pauses, backward contractions, head swinging, and changes in direction. These abnormal crawling characteristics were observed in larvae homozygous for β4GalNAcTA4.1. This mutant allele has a small deletion within the β4GalNAcTA reading frame that is predicated to delete the first 143 amino acids of the protein product (Haines and Irvine 2005). The crawling phenotype is also seen in larvae carrying the β4GalNAcTA4.1 allele in trans to a larger deletion that removes the entire β4GalNAcTA reading frame as well as a neighboring gene, Df(2R)β4GalNAcTA[20.1]. To further verify that this phenotype is due to loss of β4GalNAcTA, we examined the crawling pattern of β4GalNAcTA4.1 mutants carrying a genomic rescue construct for β4GalNAcTA. These larvae display a wild-type crawling pattern (Figure 1 and supplemental movie C at http://www.genetics.org/supplemental/), confirming that the phenotype is due to loss of β4GalNAcTA.

Figure 1.—

β4GalNAcTA mutant larvae display a crawling phenotype. The graph shows the average number of forward (+) and backward (−) body-wall contractions/minute recorded for each genotype. Mutants show decreased forward and increased backward movement compared to yw controls. This phenotype is rescued by a genomic transgene containing a wild-type copy of β4GalNAcTA.

Size, branching, and bouton number at the NMJ is reduced:

Third instar larval NMJs of β4GalNAcTA mutants were examined for morphological defects by staining with antibodies. Anti-HRP recognizes an epitope expressed throughout the Drosophila nervous system and allows visualization of the NMJ. A second antibody, against the synaptic vesicle protein synaptotagmin, was used to facilitate the quantification of synaptic boutons. To reduce variability in this experiment, the muscle 6 and 7 NMJ was examined from larval segments 3 or 4. These two muscles receive innervation from the same motor neurons, with the motor neurons forming synaptic boutons on both muscles. These experiments revealed that the nerve terminal branches of β4GalNAcTA mutant larvae are significantly shorter than those of wild type (Figure 2, A–G). The mutant NMJs also have fewer branches and the number of synaptic boutons, as observed with synaptotagamin staining, is reduced (Figure 2, H and I). These differences are not due to a change in muscle size in the mutants. The NMJ expands to keep up with the rapid increase in volume of the muscle during larval growth. There is a strong correlation between NMJ length and muscle volume (Schuster et al. 1996). We measured the combined surface area of muscles 6 and 7 as an approximation of muscle size and found no significant difference using ANOVA statistical comparison with a 95% confidence interval (area ± SEM, n; yw 77,950 ± 2366 μm2, n = 18; β4GalNAcTA4.1 to 77,304 ± 2531 μm2, n = 24; β4GalNAcTA4.1/20.1 76,371 ± 2403 μm2, n = 20; β4GalNAcTA4.1;p[β4GalNAcTA+] 78,720 ± 3914 μm2, n = 17).

Figure 2.—

NMJ size, branching, and bouton number are reduced in β4GalNAcTA mutants. (A–D) Third instar larval NMJs at muscles 6 and 7 or (E and F) at muscles 12 and 13 stained with anti-HRP (red) and antisynaptotagmin (green). (A) yw and (B) β4GalNAcTA4.1: NMJ is smaller than control and is less branched. (C) β4GalNAcTA4.1/20.1 and (D) β4GalNAcTA4.1;p[β4GalNAcTA+]/+. The reduced NMJ size is also apparent at the NMJs at muscles 12 and 13 (E) β4GalNAcTA4.1/20.1 and is also rescued by the genomic transgene (F) β4GalNAcTA4.1;p[β4GalNAcTA+]/+. Quantification of muscles 6 and 7 NMJ length (G), number of boutons (H), and number of branches (I). ANOVA revealed a significant variation among the genotypes shown for NMJ length, bouton number, and branch number (P < 0.0001), and post-tests showed that β4GalNAcTA4.1 and β4GalNAcTA4.1/20.1 were different from yw (P < 0.01). Both β4GalNAcTA4.1 and β4GalNAcTA4.1/20.1 were significantly different from β4GalNAcTA4.1;p[β4GalNAcTA] (P < 0.01). No difference was seen between yw and β4GalNAcTA4.1;p[β4GalNAcTA] or the two mutant genotypes (P > 0.05). yw n = 18; β4GalNAcTA4.1, n = 24; β4GalNAcTA4.1/20.1, n = 20; β4GalNAcTA4.1;p[β4GalNAcTA], n = 17.

Physiological analysis of the neuromuscular synapse:

The reduction of the NMJ size could alter the function of the synapse. To detect such changes, we measured spontaneous mEJPs and nerve-stimulated EJPs synaptic release using intracellular recording. No significant changes in resting membrane potential or muscle input resistance were detected (yw −66.8 ± 0.53 mV; 4.6 ± 0.56 mΩ, β4GalNAcTA4.1 −67.3 ± 0.91 mV; 4.8 ± 1.19 mΩ, β4GalNAcTA4.1/20.1 −68.8 ± 1.78 mV; 3.25 ± 0.17 mΩ, β4GalNAcTA4.1;p[β4GalNAcTA+] −68.8 ± 0.8 mV, 5 ± 0.44 mΩ, n = 6 for each genotype). Amplitude of mEJP and evoked EJP were also similar for each genotype (Figure 3, B and C). While these results suggest that the physiology of the synapse is not significantly altered by loss of β4GalNAcTA, we did find a significant reduction in the frequency of mEJP in mutants compared to yw controls and rescued larvae (Figure 3, A and D).

Figure 3.—

Synapse physiology is normal in mutants but mEJP frequency is reduced. (A) mEJP frequency is significantly reduced in β4GalNAcTA mutants compared to yw controls or larvae carrying a genomic rescue transgene. ANOVA comparison of the four genotypes shown revealed a significant variation among them (P < 0.0001), and post-tests showed that β4GalNAcTA4.1 and β4GalNAcTA4.1/20.1 were different from yw (P < 0.01) and β4GalNAcTA4.1;p[β4GalNAcTA] (P < 0.01). No difference was seen between yw and β4GalNAcTA4.1;p[β4GalNAcTA] or the two mutant genotypes (P > 0.05). No significant differences are seen in either mEJP amplitude (B) or EJP amplitude (C). (D) Representative traces of mEJP recording. n = 6 for each genotype.

β4GalNAcTA mutant muscles are hypercontracted:

Muscles 6 and 7 lie side by side, resulting in a long rectangular outline in each abdominal hemi-segment (Figure 4B). Although there is no significant difference in the area of the muscles between mutants and controls (see above), we did note that in mutants these muscles had a more square than rectangular appearance. To investigate this further, the length along the muscle 6/7 boundary and the combined width of these two muscles were measured (Figure 4B). In mutants, the length was shorter and the width wider than that of controls (length ± SEM, width ± SEM, n; yw 508 ± 8.2 μm, 143 ± 4.2 μm, n = 22; β4GalNAcTA4.1 482 ± 7.6 μm, 159 ± 3.8 μm, n = 22; β4GalNAcTA4.1/20.1 483 ± 11.8 μm, 158 ± 4.2 μm, n = 22; β4GalNAcTA4.1;p[β4GalNAcTA] 510 ± 9.6 μm, 149 ± 6.9 μm, n = 17). This change in muscle width and length resulted in a highly significant difference in the width/length ratio for mutants compared to control and rescued larvae (Figure 4A).

Figure 4.—

β4GalNAcTA muscles are hypercontracted. (A) β4GalNAcTA mutants have a significantly increased ratio of muscle width/length. ANOVA comparison of the width/length ratios of the four genotypes shown revealed a significant variation among them (P < 0.0001), and post-tests showed that β4GalNAcTA4.1 and β4GalNAcTA4.1/20.1 were different from yw (P < 0.001). β4GalNAcTA4.1 was significantly different from β4GalNAcTA4.1;p[β4GalNAcTA] (P < 0.001) and β4GalNAcTA4.1/20.1 was different from β4GalNAcTA4.1;p[β4GalNAcTA] (P < 0.05). No difference was seen between yw and β4GalNAcTA4.1;p[β4GalNAcTA] or the two mutant genotypes (P > 0.05). yw, n = 22; β4GalNAcTA4.1, n = 22; β4GalNAcTA4.1/20.1, n = 22; β4GalNAcTA4.1;p[β4GalNAcTA], n = 17. Muscle measurements were from the width and length of muscles 6 and 7 combined, as shown B. (C) Sarcomere size is reduced in mutants. ANOVA revealed a significant variation among the genotypes shown (P < 0.0001), and post-tests showed that β4GalNAcTA4.1 and β4GalNAcTA4.1/20.1 were different from yw (P < 0.001). Both β4GalNAcTA4.1 and β4GalNAcTA4.1/20.1 were significantly different from β4GalNAcTA4.1;p[β4GalNAcTA] (P < 0.001). No difference was seen between yw and β4GalNAcTA4.1;p[β4GalNAcTA] or the two mutant genotypes (P > 0.05). yw, n = 108; β4GalNAcTA4.1, n = 102; β4GalNAcTA4.1/20.1, n = 60; β4GalNAcTA4.1;p[β4GalNAcTA], n = 84). (D) How sarcomeres were measured. Graph represents gray-scale intensity across the sarcomeres shown below. Peak-to-peak distance was taken as measurement of sarcomere unit size. First peaks are aligned but by 28 μm the β4GalNAcTA4.1 plot is out of phase with yw and rescued plot.

Short, wide muscles could result from hypercontraction of the resting muscle. In hypercontracted muscles, due to the increased overlap of the actin and myosin filaments, the Z-bands are closer together and the length of the individual sarcomere units (distance from one Z-band to the next) is therefore decreased. Using phalloidin to stain F-actin, the banding pattern of the contractile apparatus in the larval muscles can be clearly observed (Figure 4, B and D). Phalloidin stripes are broad, but intensity is strongest at the center of each band, the location of the Z-band. We collected high magnification (×63) confocal images of phalloidin-stained muscles, and regions of the muscle where the banding pattern of the contractile apparatus was square on the image were selected. The Z-section through the muscle was then examined to ensure that sarcomeres remained aligned and square through at least a 1.8-μm section of the muscle. A line was drawn on the image at a right angle to the sarcomeres and the gray-scale intensity along the line was obtained and plotted (Figure 4D). Peak-to-peak distance was taken as a measure of sarcomere length. Sarcomeres from mutants were significantly smaller than those of control or rescued larvae. We did not observe a change in the width of the phalloidin band in mutant sarcomeres, i.e., the half-width of the curve. These results confirm that muscles in β4GalNAcTA mutants are hypercontracted.

Analysis of muscle calcium:

Calcium homeostasis is critical to many aspects of muscle function, and aberrant calcium regulation in muscles could result in hypercontraction. We investigated calcium dynamics in the β4GalNAcTA mutant muscles to determine if perturbations in calcium regulation that could potentially underlie the muscle hypercontraction phenotype are present. A UAS-coupled, genetically encoded calcium indicator, G-CaMP (Nakai et al. 2001; Wang et al. 2004), was expressed in β4GalNAcTA mutants using a mesoderm Gal4 driver, 24B-Gal4. Although it was not possible to measure resting calcium levels, we were able to measure calcium dynamics in response to potassium-induced depolarization. We observed a slow (>10–20 sec) increase in fluorescence throughout the muscle following potassium-induced depolarization (Figure 5, A and B). Calcium transients with a similar time course have been identified in rodent skeletal myoblasts (Jaimovich et al. 2000; Estrada et al. 2001), and although these are unrelated to the very fast (in milliseconds) release of calcium that is responsible for muscle contraction, we used these slow calcium waves as an indicator of calcium regulation in muscle. In control and β4GalNAcTA mutants, the size of the change in fluorescent intensity was similar, but mutants took several seconds longer to reach peak intensity (Figure 5). This result suggests that the calcium dynamics that generate these slow waves in Drosophila muscles are altered in β4GalNAcTA mutants.

Figure 5.—

Calcium dynamics in response to potassium-induced depolarization are altered in β4GalNAcTA muscles. (A) Muscles expressing G-CaMP calcium indicator as they respond to potassium-induced depolarization. Fluorescent intensity increases in the seconds following potassium addition (t = 0), and then decrease. Change in ΔF/F is shown at bottom left of each image. At t = 8, control is close to peak intensity and, at t = 16 ΔF/F, has decreased from the maximum. β4GalNAcTA4.1 mutants are slower to respond to the depolarization; ΔF/F at t = 16 is greater than t = 8. (B) Graph shows examples of ΔF/F over the time course of the experiment. Control, black triangles; red ovals, β4GalNAcTA4.1. (C) Graph shows time at peak fluorescent intensity for different genotypes; controls, n = 15; β4GalNAcTA4.1, n = 10. *P = 0.0087 t-test.

Distinct neural and muscle requirements for β4GalNAcTA:

The larval crawling defect, NMJ, and muscle phenotypes all point to functional roles for β4GalNAcTA in the neuromuscular system. To determine the relative contributions of nerve and muscle gene expression to these phenotypes, we used the Gal4/UAS system (Brand and Perrimon 1993) to selectively express β4GalNAcTA. A UAS-coupled β4GalNAcTA construct was expressed in the homozygous β4GalNAcTA4.1 mutant background using the pan-neural driver elav-Gal4, the mesoderm-specific driver 24B-Gal4, or B185-Gal4, which drives expression in neurons and muscle.

Driving expression of the enzyme with either elav-Gal4 or B185-Gal4 resulted in clear rescue of the crawling phenotype (Figure 6A). These larvae displayed wild-type levels for both forward and backward BWC (Figure 6A). In contrast, no rescue of the crawling phenotype was seen with the mesoderm driver 24B-Gal4 (Figure 6A). These results demonstrate that the β4GalNAcTA is required in neurons for normal larval crawling behavior.

Figure 6.—

Tissue-specific expression of β4GalNAcTA rescues the mutant phenotypes. (A) Larval crawling phenotype is rescued with elav or B185-Gal4-driven expression of UAS-β4GalNAcTA, 24B-Gal4 results in no rescue of this phenotype. (B) Larval muscle hypercontraction was assessed by the width-to-length ratio of muscles 6 and 7 combined, as in Figure 4. 24B-Gal4 rescues the muscle hypercontraction phenotype, while B185-Gal4 results in some rescue. No rescue of this phenotype is seen with elav-Gal4. (C) Number of boutons at muscle 6/7 NMJ as assessed by antisynaptotagmin staining. None of the Gal4 drives rescue the reduced bouton number phenotype. Genotypes are β4GalNAcTA4.1;UAS-β4GalNAcTA/+ (control), elav-Gal4/+; β4GalNAcTA4.1;UAS-β4GalNAcTA/+ (elavGal4), β4GalNAcTA4.1;UAS-β4GalNAcTA/24B-Gal4 (24BGal4), and β4GalNAcTA4.1 B185-Gal4/β4GalNAcTA4.1;UAS-β4GalNAcTA/+ (B185Gal4).

The muscle hypercontraction phenotype, as assessed by the muscle width/length ratio, was clearly rescued with 24B-Gal4-driven expression of the enzyme (Figure 6B). Some rescue was obtained with B185Gal4. elav-Gal4 resulted in no rescue of this phenotype (Figure 6B). This suggests that the muscle phenotype is caused by loss of β4GalNAcTA activity in muscles and not neurons and implies that at least one substrate exists in muscle that is regulated by β4GalNAcTA.

The reduced size, branching, and bouton number at the NMJ of β4GalNAcTA mutants was not rescued with any of the Gal4 drivers tested (Figure 6C), although it was rescued with the genomic rescue transgene (Figure 2). This phenotype could result from interaction between neurons and muscles and these Gal4 drives do not provide adequate temporal or spatial expression of the enzyme to rescue this function.

DISCUSSION

β4GalNAcTA, like its homolog β4GalNAcTB, is expressed broadly throughout Drosophila development. However, flies homozygous for null mutations in either enzyme or doubly mutant for both enzymes are viable and fertile with no clear morphological phenotypes. This demonstrates that these enzymes are not essential for viability in Drosophila. The observation that adult β4GalNAcTA mutant flies are uncoordinated provided an indication that β4GalNAcTA plays functional roles in the neuromuscular system (Haines and Irvine 2005). Here we have followed up this finding by analysis of the neuromuscular system of β4GalNAcTA mutant larvae. This approach has led to the identification of neural and muscle phenotypes associated with loss of β4GalNAcTA and has expanded our understanding of the functional significance of this enzyme. Given the glycosyltransferase activity of β4GalNAcTA, it is likely that these phenotypes stem from loss of glycan structures generated by the enzyme. Glycosyltransferases, however, can play other roles, for example, as protein chaperons (Okajima et al. 2005), and such activities remain possible mechanisms for β4GalNAcTA function.

We have demonstrated that loss of β4GalNAcTA is associated with abnormal larval crawling behavior, reductions in the size of the NMJ, and reduced spontaneous release frequency at the neuromuscular synapse and that mutant muscles are hypercontracted. Importantly, our data demonstrate that each of these phenotypes is specifically due to loss of β4GalNAcTA. The lack of a distinguishable difference between β4GalNAcTA4.1 homozygotes and individuals carrying this allele in trans to the Df(2R)β4GalNAcTA[20.1] deficiency that removes the entire β4GalNAcTA region demonstrates that the β4GalNAcTA4.1 allele behaves as a genetic null. Further, we find that a genomic transgene containing the β4GalNAcTA region rescues each phenotype in the β4GalNAcTA4.1 mutant background.

The unusual crawling phenotype observed in β4GalNAcTA mutants is specifically caused by loss of β4GalNAcTA in neurons as this phenotype can be rescued to wild type by neural Gal4-driven expression of the enzyme. Larval locomotion depends on central neural circuits that generate patterned discharge in motor nerves and on sensory input, which plays an important role in modifying the pattern of motor activity (Suster and Bate 2002). β4GalNAcTA larvae show waves of coordinated muscle contraction, suggesting that the circuits responsible for generating coordinated muscle contraction are functioning normally. The reduced forward crawling, increased backward crawling, frequent pauses, head swinging, and high amounts of backward contraction show similarities to the crawling pattern of larvae in which sensory neuron activity has been eliminated due to expression of tetanus toxin (Suster and Bate 2002). Possibly β4GalNAcTA, and the glycans that it generates, play a role in sensory neurons and loss of this function impacts larval crawling pattern. A similar spontaneous backward-crawling phenotype has been reported in a mutant in the calcium sensor calmodulin, and this behavior is due to loss of calmodulin function in neurons (Wang et al. 2003). The underlying link between the calmodulin mutant and spontaneous backward crawling has not been established.

While examining the neuromuscular system of β4GalNAcTA mutants, it became clear that the larval muscles showed a subtle change in shape. This shape change is most obvious when observing muscles 6 and 7, as these are large, longitudinal muscles that run on each side of the midline. In β4GalNAcTA mutants, these muscles appear more square in contrast to their normal, long rectangular appearance. We went on to measure the muscles and found that although muscle surface area was similar to controls, mutant muscles were shorter and wider than controls and had significantly larger width/length ratios. As this correlates with a decrease in size of the sarcomere units, we conclude that the β4GalNAcTA mutant muscles are hypercontracted. This muscle hypercontraction can be rescued by 24B-Gal4-driven expression of the enzyme and, to a lesser extent, by the weaker driver B185-Gal4. Interestingly, this hypercontraction phenotype has little obvious effect on the larvae, as crawling behavior does not seem to be influenced by this muscle phenotype. β4GalNAcTA mutants with elav-Gal4-driven expression of UAS-β4GalNAcTA have normal crawling behavior in our crawling assay, but the muscles of these larvae remain hypercontracted. Few studies have focused on Drosophila larval muscles, and this result suggests that muscle defects may go undetected when common morphological and behavioral assays are employed to examine Drosophila mutants.

Excitatory depolarization results in release of calcium out of the sarcoplasmic reticulum and muscle contraction. We reasoned that a change in calcium regulation that leads to an increase in intracellular calcium in the β4GalNAcTA mutants could underlie the muscle hypercontraction phenotype. Using the genetically encoded fluorescent calcium indicator G-CaMP, we found a delay in slow calcium dynamics in mutant muscles following potassium-induced depolarization. It is not clear whether such a delay could lead to increased intracellular calcium over time; however, this result does indicate that loss of β4GalNAcTA function has some affect on calcium dynamics in the muscle. Many muscle proteins impact calcium regulation and glycoproteins are known to be involved. For example, muscle calcium regulation is disrupted in dystrophic muscles (Blake et al. 2002). Muscular dystrophy can result from mutation of glycosyltransferases that generated glycan structures on dystroglycan and are essential for its function (Michele and Campbell 2003; Barresi and Campbell 2006).

Loss of β4GalNAcTA results in reduced NMJ size, branch number, and bouton number. Our physiology assays demonstrate that the function of the synapse is relatively unaffected, suggesting that the mechanisms that regulate synaptic strength are able to compensate for the reduced NMJ size. We found that mEJP frequency is reduced; however, this could simply result from the reduced overall size of the NMJ. This result nevertheless suggests that β4GalNAcTA plays a role in fine tuning NMJ morphology.

Glycoproteins, such as fasciclin II, a member of the immumgobulin superfamily of cell adhesion molecules are known to influence Drosophila NMJ morphology, and members of this protein family play similar roles in vertebrates (Schuster et al. 1996; Benson et al. 2000; Packard et al. 2003).

The data presented here clearly indicate that β4GalNAcTA plays distinct roles in neurons and muscles in Drosophila. Together, our results suggest that further investigations focused on sensory neurons, and calcium regulators in muscle, will aid progress toward identifying the target molecular pathways and functional substrates of β4GalNAcTA.

Acknowledgments

We thank Ken Irvine for generously providing fly stocks and reagents used in this study and N. Reist for the antisynaptotagmin antibody. Research in B. A. Stewart's Laboratory is supported by the Natural Sciences and Engineering Research Council of Canada, Mizutani Foundation for Glycoscience, the Premiers Research Excellence Award, and the Canada Research Chairs program.

Footnotes

  • Communicating editor: R. S. Hawley

  • Received September 5, 2006.
  • Accepted November 7, 2006.

References

View Abstract