O-linked glycosylation in Drosophila melanogaster

Abstract

Glycosylation, or the addition of sugars to proteins, is a highly conserved protein modification defined by both the monosaccharide initially attached to the protein as well as the amino acid to which it is attached. O-linked glycosylation represents a diverse group of protein modifications occurring on the hydroxyl groups of serine and/or threonine residues. O-glycosylation can have wide-ranging effects on protein stability and function, which translate into crucial consequences at the organismal level. This review will summarize structural and biological insights into the major O-glycans formed within the secretory apparatus (O-GalNAc, O-Man, O-Fuc, O-Glc and extracellular O-GlcNAc) from studies in the fruit fly Drosophila melanogaster. Drosophila has many advantages for investigating these complex modifications, boasting reduced functional redundancy within gene families, reduced length/complexity of glycan chains and sophisticated genetic tools. Gaining an understanding of the normal cellular and developmental roles of these conserved modifications in Drosophila will provide insight into how changes in O-glycans are involved in human disease and disease susceptibilities.

Major O-linked glycans in Drosophila

While the major types of O-glycans and the enzymes that generate them are conserved between Drosophila and mammals, the length and complexity (glycan composition and branching) tend to be reduced in Drosophila. A seminal study by the Tiemeyer lab characterized the major O-glycans (excluding glycosaminoglycans) present during Drosophila embryogenesis and in a developing organ [1]. Detailed mass spectrometric analysis characterized the predominant O-glycans to be the core 1 disaccharide (Galβ1,3GalNAc); the core 1 disaccharide modified with glucuronic acid (GlcA); a HexNAc monosaccharide (GalNAc or GlcNAc); and a fucose-based trisaccharide, all of which together constituted as much as 96% of the total O-glycans (Fig. 1). This study highlighted both structural similarities between Drosophila and mammalian O-glycans, as well as unique differences, including reduced glycan chain length and the use of GlcA in flies in place of sialic acid (found abundantly in mammals). Moreover, some glycan structures found suggested unique biological functions based on when and where they were present.

Figure 1.

Major O-glycans present in Drosophila melanogaster. The major O-glycans present in Drosophila melanogaster and the enzymes that catalyze their formation are shown. Symbols denoting each saccharide are shown in the legend. * denotes enzymes that are currently unknown/unconfirmed in this pathway. S, serine; T, threonine.

O-GalNAc

O-GalNAc (mucin-type O-glycosylation) is the major type of glycosylation found in the Drosophila embryo and one of the more abundant types of O-glycosylation within both mammals and Drosophila. (Other major O-glycans, such as O-linked xylose-derived glycans, which form the basis of diverse glycosaminoglycans, and cytoplasmic/nuclear O-GlcNAc will be covered in separate sections). While O-GalNAc is well-documented on mucins (conferring their unique structural and rheological properties), this protein modification has also been found on diverse secreted and membrane-bound proteins in Drosophila (and mammals) [2], indicating its potentially broad and complex influence over many cellular processes.

The initial addition of GalNAc through an α O-glycosidic linkage to the hydroxyl group of serine or threonine is catalyzed by a large family of evolutionarily-conserved enzymes known as the UDP-GalNAc:polypeptide N-acetylgalactosaminyltransferases (ppGalNAc-Ts or GalNAc-Ts in mammals; PGANTs in Drosophila; EC 2.4.1.41) (Fig. 1). There are 20 family members in humans, 19 in mice and 10 in Drosophila, all of which are type II transmembrane proteins that reside in the Golgi apparatus. Many family members display unique spatial and temporal expression during Drosophila (and mammalian) development [3,4]. Moreover, certain family members exhibit unique substrate preferences that are dictated by 2 separate domains within these enzymes; a catalytic domain (consisting of a glycosyltransferase domain in which the active site lies) and a ricin-like lectin domain. Structural details of how these 2 domains coordinate the recognition and binding of substrates as well as the hierarchical addition of glycans, are discussed in another section (Polypeptide GalNAc-Ts: from redundancy to specificity).

The biological roles of O-GalNAc glycans have been challenging to decipher given the degree of redundancy built into the first committed step of biosynthesis as well as the hundreds of potential substrates that could be simultaneously affected. However, studies from Drosophila have demonstrated that at least 5 family members of the pgant family are essential for viability or influence viability [5–9]. Genetic ablation of individual pgants has demonstrated roles in the production, packaging and secretion of extracellular matrix components, resulting in defects in epithelial cell polarity [6,8]; loss of cell adhesion [10,11]; and changes in the progenitor cell niche, leading to aberrant cell proliferation [12]. Mechanistically, one family member (PGANT4) has been shown to influence secretion/secretory granule formation by glycosylating a conserved cargo receptor (Tango1) and protecting it from proteolysis [13]. This protective role of O-GalNAc is similar to that seen in mammals, where GALNT3 glycosylates the phosphaturic hormone FGF23, protecting it from proteolysis [14]. More recently, evidence has emerged that O-glycosylation of secretory proteins (cargo) is required for proper secretory granule morphology [15]. This study also demonstrates evidence for differential splicing of functional domains within these glycosyltransferases to alter substrate specificity [15], suggesting an even greater repertoire of enzymes catalyzing the initial transfer of O-linked GalNAc than the gene number would suggest. Given the complexity of the many enzymes involved in O-GalNAc addition, combined with the enormous number of potential substrates, studies in this tractable model system will be crucial to gain a fundamental understanding of the roles of this protein modification in vivo.

After the initial addition of GalNAc, the predominant modification involves the additional of Gal in a β1,3 linkage by another family of enzymes known as the core 1 galactosyltransferases (C1GalTs) (Fig. 1). While only 1 C1GalT exists in mammals (C1GalT1 or T-synthase), as many as 9 potential members exist in Drosophila [16], suggesting additional functional redundancy exists within this family as well. Genetic studies ablating 1 family member (C1GalT1) resulted in neurological phenotypes, including elongated ventral nerve cords and distorted brain hemispheres [17,18] (Table I). While substrates were not identified, there was evidence of disruptions in the extracellular matrix normally present in the nervous system.

Table I.

Developmental phenotypes associated with mutation or knockdown of genes encoding the enzymes involved in O-glycan biosynthesis in Drosophila

Gene	Protein/Enzyme	Glycan formed	Mutant phenotypes	Reference
pgant3	PGANT3	GalNAcα-O-S/T	-decreased secretion of ECM components -loss of integrin-mediated cell adhesion resulting in wing blisters	Zhang et al., 2008 Zhang et al., 2010
pgant4	PGANT4	GalNAcα-O-S/T	-lethal during development -loss of secretory vesicles and secretion defects in digestive system	Tran et al., 2012 Zhang et al., 2014
pgant5	PGANT5	GalNAcα-O-S/T	-lethal during development -loss of gut acidification	Tran et al., 2012
pgant7	PGANT7	GalNAcα-O-S/T	-lethal during development	Tran et al., 2012
pgant35A	PGANT35A	GalNAcα-O-S/T	-lethal during development -irregular formation of embryonic respiratory system -loss of cell polarity and diffusion barrier in respiratory system	Ten Hagen and Tran, 2002 Schwientek et al., 2002 Tian and Ten Hagen, 2007
pgant9	PGANT9	GalNAcα-O-S/T	-semi-lethal during development -loss of one splice variant (PGANT9B) causes abnormal morphology of secretory granules	Tran et al., 2012 Ji and Samara et al., 2018
C1GalTA	C1GalTA	Galß1,3GalNAcα-O-S/T	-elongated ventral nerve cord -distorted brain hemispheres	Lin et al., 2008 Yoshida et al., 2008
rotated abdomen (rt)	dPOMT1	Manα-O-S/T	-semi-lethal during development -clockwise rotation of the abdominal segments -muscle developmental defects, abnormal synaptic transmission -decreased flying and climbing abilities -abnormal axonal connections of sensory neurons leading to abnormal muscle contractions and embryo torsion	Martin-Blanco and Garcia-Bellido, 1996 Ichimiya et al., 2004 Lyalin et al., 2006 Haines et al.,2007 Wairkar et al., 2008 Ueyama et al., 2010 Baker et al., 2018
twisted (tw)	dPOMT2	Manα-O-S/T	-semi-lethal during development -clockwise rotation of the abdominal segments -muscle developmental defects -decreased flying and climbing abilities -abnormal axonal connections of sensory neurons leading to abnormal muscle contractions and embryo torsion	Martin-Blanco and Garcia-Bellido, 1996 Ichimiya et al., 2004 Lyalin et al., 2006 Haines et al.,2007 Ueyama et al., 2010 Baker et al., 2018
Eogt	EOGT	GlcNAcß-O-S/T on EGF repeats	-lethal during larval development -cuticle defects and irregular tracheal morphology -wing blisters and thorax vortex	Sakaidani et al., 2011 Muller et al., 2013
Ofut1	Ofut1	Fucα-O-S/T on EGF repeats	-lethal during development -reduction in Notch signaling leading to loss of wing tissue, thickened wing veins, rough eyes, additional notal macrochaetes, leg segment fusions	Okajima and Irvine, 2002 Okajima et al., 2005 Okajima et al., 2008
fng	Fringe	GlcNAcß1,3Fucα-O-S/T on EGF repeats	-alterations in Notch signaling by regulating Notch-ligand interactions -defects in wing formation -defects in eye development	Irvine and Wieschaus, 1994 Cho and choi, 1998 Correia et al., 2003 LeBon et al., 2014
Ofut2	Ofut2	Fucα-O-S/T on TSR regions	-decreased TSR specific O-fucosyltransferase activity in S2 cells	Luo et al., 2006
rumi	Rumi	Glcß-O-S on EGF repeats	-defects in Notch folding and signaling leading to loss of bristles, defects in wing, eye and leg development -highly penetrant rhabdomere attachment phenotype	Acar et al., 2008 Haltom et al., 2014
shams	Shams	Xylα1,3Glcß-O-S on EGF repeats	-increase in Delta-mediated Notch signaling leading to abnormal wing vein formation and head bristle development	Lee et al., 2013 Lee et al., 2017
Xxylt	Xxylt	Xylα1,3Xylα1,3Glcß-O-S on EGF repeats	-changes in Notch signaling only in sensitized genetic backgrounds	Lee et al., 2013 Haltom and Jafar-Nejad, 2015 Pandey et al., 2018

O-Man

The addition of an O-linked mannose to the hydroxyl group of serines or threonines in Drosophila is catalyzed by 2 protein O-mannosyltransferases known as dPOMT1 and dPOMT2, encoded by the genes rotated abdomen (rt) and twisted (tw) respectively [19–21] (Fig. 1). These enzymes are transmembrane proteins that reside in the ER and are highly homologous to the mammalian orthologs responsible for O-mannosylation [19]. Both genes are required for O-mannosylation in vitro and in vivo, suggesting that they function in a heteromeric complex within the ER [19]. Interestingly, while mammalian O-mannose is usually extended by the addition of other sugars, Drosophila O-mannose exists on proteins as a monosaccharide [1].

O-mannosylation is essential for viability in Drosophila (Table I) as loss of both rt and tw simultaneously is lethal [19]. Loss of either gene results in a decrease in viability and surviving adults display a 90° clockwise rotation of their abdominal segments [19–22]. More detailed characterization shows defects in muscle attachment and architecture [23,24], as well as alterations in the neuromuscular junction [25]. Unlike O-GalNAc glycosylation, only a small number of targets have been identified for O-mannose glycosylation, the primary one being the transmembrane protein dystroglycan (Dg) [23], which is part of the dystrophin-glycoprotein complex (DGC) that has been shown to be required for the proper linkage of the extracellular matrix to the inner cytoskeleton of the cell. Defects in components of the DGC and the enzymes that modify it (including human POMT1 and POMT2) are known to be responsible for human muscular dystrophies [26], which are also characterized by muscle and neurological phenotypes similar to those seen in Drosophila. Additional recent work in Drosophila has suggested a role for rt and tw within the sensory neurons during early development to establish proper muscle contractions and body posture [27]. Interestingly, these phenotypes are not mimicked by loss of Dg, suggesting that other biologically important targets for O-mannosylation that coordinate muscle contraction and nervous system feedback remain to be discovered [27].

O-Fuc

Notch signaling is an essential, conserved pathway that controls cell-cell communication and cell fate decisions during animal development and a number of types of glycosylation have been found to regulate it, including O-linked fucose [28]. The Notch receptor, a transmembrane protein on the surface of cells, binds to ligands (Delta or Serrate) and then undergoes a coordinated series of proteolytic cleavages leading to the release of its intracellular domain, which can then travel to the nucleus and regulate the transcriptional activity of many genes [28]. Thus, factors that regulate Notch receptor or ligand synthesis, stability, trafficking, interactions or the downstream cleavage events can have profound effects on tissue homeostasis and development. Interestingly, both Notch signaling and the glycosyltransferases that modulate it were first discovered in Drosophila.

The addition of O-linked fucose to serine or threonine within the EGF repeats of Notch or its ligands occurs through the action of the O-fucosyltransferase Ofut1 (Pofut1 in mammals) [29,30](Fig. 1). Ofut1 (encoded by the Ofut1 gene) is a soluble ER-localized enzyme that recognizes the consensus sequence C²-X-X-X-X-T/S-C³. The addition of fucose by Ofut1 is thought to regulate Notch signaling by affecting ligand binding [31]. Loss of Ofut1 resulted in lethality during development and phenotypes indicative of loss of Notch signaling (Table I). Additionally, there is also evidence in Drosophila suggesting a role for Ofut1 as a chaperone independent of its enzymatic activity [32,33]. However, a non-enzymatic role for the mammalian Pofut1 has not been found [28]. The O-fucose is further elaborated by the addition of GlcNAc by the Golgi-localized β1,3 N-acetylglucosaminyltransferase known as Fringe. The fng gene was the first example of a genetically well-characterized modulator of Notch signaling that was found to encode a glycosyltransferase [34,35]. Fringe-mediated elongation of O-fucose influences Notch signaling by enhancing the binding of one ligand (Delta) while inhibiting the binding of another (Serrate) [31,36–39].

Another O-fucosyltransferase, known as Ofut2, exists in Drosophila and adds fucose to thrombospondin type 1 repeats, which are found in many transmembrane and secreted proteins [40] (Fig. 1). Its function in vivo in Drosophila is currently unknown.

O-Glc

Work in Drosophila identified another glycosyltransferase Rumi (a member of the GT90 family) that also modifies the extracellular region of Notch to regulate signaling. The gene rumi encodes an ER-localized, soluble O-glucosyltransferase (Poglut1 in mammals) that adds O-linked glucose (Glc) to serine residues within the consensus sequence C¹-X-S-X-P/A-C² of the EGF repeats (Fig. 1). rumi was identified in a screen for regulators of sensory organ development. Mutations in rumi resulted in decreased glucosylation of Notch and a reduction in Notch signaling [41]. Upon loss of rumi, Notch accumulated intracellularly and also failed to undergo proper cleavage at the cell membrane after ligand binding [41]. Studies in Drosophila have suggested that the loss of Rumi influences Notch folding/conformation, thus affecting its ability to undergo regulated cleavage after ligand binding. Additional work in Drosophila suggests that O-glucosylation by Rumi is also required for proper folding and stability of a secreted protein (Eyes shut) that is involved in proper photoreceptor organization in the eye [42]. Structural and biochemical studies of Poglut1 and mammalian EGF repeats further support a role for O-Glc glycans in the stabilization of EGF repeats to allow proper folding [43].

O-Glc can be further modified by the addition of xylose in an α linkage, through the action of the glucoside xylosyltransferase Shams (GXYLT1/2 in mammals) (Fig. 1). shams mutants have phenotypes indicative of Notch gain-of-function phenotypes, suggesting that xylose attached to glucose on Notch negatively regulates Notch signaling. Experimental evidence suggests that xylose may alter the cell surface expression of Notch in certain tissues [44]. However, in other instances Shams appears to affect the ability of Notch to bind Delta expressed on a neighboring cell (trans-Delta ligands) while not affecting its binding to ligands expressed from the Notch-expressing cell (cis-ligands). These studies suggest that the addition of xylose to the O-Glc on Notch regulates the balance of Notch activation occurring by trans-ligands relative to its inhibition by cis-ligands [45]. The mechanism whereby xylose on Notch would differentially affect binding to Delta in trans versus cis is currently unknown.

Finally, a second xylose can be added to Xylα1,3Glc present on EGF repeats through the action of a xyloside xylosyltransferase encoded by the Xxylt gene (CG11388) in Drosophila (XXYLT1 in mammals) [46] (Fig. 1). Loss of this gene results in phenotypes only when combined with other genetic modifiers of Notch signaling, suggesting that it functions primarily in fine-tuning this signaling pathway.

O-GlcNAc (extracellular)

While the presence of O-linked GlcNAc is well-established on nuclear, cytoplasmic and mitochondrial proteins, the addition of an O-linked GlcNAc to secreted and membrane proteins is a more recent discovery. EOGT (encoded by the gene Eogt) is a soluble ER-resident glycosyltransferase that transfers GlcNAc to serine or threonine within the consensus sequence C₅-X-X-G-X-T/S-G-X-X-C₆ in EGF repeating domains of a number of proteins [47–49]) (Fig. 1). Loss of Eogt throughout the animal results in larval lethality, while tissue-specific loss within the wing causes wing blisters [49]. One of the major proteins glycosylated by EOGT is Dumpy (Dp), a large apically localized membrane-anchored protein that is involved in the maintenance of cell-extracellular matrix interactions (Table I). Additionally, O-GlcNAc has been found in the EGF repeats of Notch and the Notch ligands, Delta and Serrate [48]. Genetic interaction studies revealed that the loss of Eogt could be partially rescued by loss of one allele of Notch or its ligands [48], suggesting that Eogt may be involved in downregulating Notch signaling. Additionally, genetic interactions were also noted between Eogt and genes involved in pyrimidine metabolism. The details of how levels of EOGT and Notch may be connected to uracil production remain to be elucidated but suggest that this enzyme may serve to modulate levels of nucleotide sugars within the ER.

Future outlook

Alterations in O-linked glycosylation are associated with many human diseases and syndromes, highlighting their importance in biomedicine. However, a fundamental understanding of how these conserved protein modifications affect protein structure, stability and/or function, combined with insight into how these alterations in protein function translate into cellular and organismal phenotypes is essential for developing well-informed, effective strategies for disease diagnosis and treatment. To that end, the sophisticated genetic and molecular tools unique to Drosophila have both identified previously unknown O-glycans and accelerated our understanding of their complex and diverse functions. Future work in this model organism will continue to provide essential mechanistic insights into the biological roles of these conserved protein modifications.

Abbreviations

The abbreviations used are:

ppGalNAc-T or GalNAc-T or PGANT

UDP-GalNAc:polypeptide N-acetylgalactosaminyltransferase

Man

mannose

Fuc

fucose

Glc

glucose

GalNAc

N-acetylgalactosamine

Gal

galactose

GlcNAc

N-acetylglucosamine

GlcA

glucuronic acid