Abstract
N-glycans are molecularly diverse sugars borne by over 70% of proteins transiting the secretory pathway and have been implicated in protein folding, stability, and localization. Mutations in genes important for N-glycosylation result in congenital disorders of glycosylation that are often associated with intellectual disability. Here, we show that structurally distinct N-glycans regulate the activity of an extracellular protein complex involved in patterning of somatosensory dendrites in Caenorhabditis elegans. Specifically, aman-2/Golgi alpha-mannosidase II, a conserved key enzyme in the biosynthesis of specific N-glycans regulates the activity of the Menorin adhesion complex without obviously affecting protein stability and localization of its components. AMAN-2 functions cell-autonomously to ensure decoration of the neuronal transmembrane receptor DMA-1/LRR-TM with high-mannose/hybrid N-glycans. Moreover, distinct types of N-glycans on specific N-glycosylation sites regulate the DMA-1/LRR-TM receptor, which together with three other extracellular proteins forms the Menorin adhesion complex. In summary, specific N-glycan structures regulate dendrite patterning by coordinating the activity of an extracellular adhesion complex suggesting that the molecular diversity of N-glycans can contribute to developmental specificity in the nervous system.
INTRODUCTION
Development of a nervous system in metazoans requires the coordinated interactions of extracellular molecules to ensure correct neuronal morphogenesis, and to establish connectivity (Jan & Jan, 2010; Dong et al, 2015; Lefebvre, 2021). Most of these extracellular proteins are glycoconjugates, i.e. carry different types of glycans attached to the protein backbone. Glycans are molecularly the most diverse molecules in nature, in part because they are not genetically encoded. Yet their structures are not random and are therefore conceptually attractive to broaden the molecular diversity and specificity of extracellular proteins and their interactions during development. For example, glycosaminoglycans, a class of glycans, have been suggested to modulate protein-protein interactions and provide information during development by way of their structural diversity (Holt & Dickson, 2005; Bülow & Hobert, 2006; Poulain & Yost, 2015; Masu, 2016). Whether the structural diversity of other classes of glycans such as N-glycans and O-glycans serve similar functions is unclear.
N-glycans are a structurally diverse group of glycans that fall into four classes: high-mannose, hybrid, complex, and paucimannose-type N-glycans, and are invariantly attached via an asparagine to a protein backbone (Stanley et al, 2015). Importantly, 70% of all proteins transiting the endoplasmic reticulum are post-translationally N-glycosylated (Apweiler et al, 1999). Therefore, the structural diversity of N-glycans could significantly expand the repertoire and specificity of protein interactions in the extracellular space. N-glycans in general have been shown to be important for protein folding, stability, and localization (Stanley et al., 2015). Moreover, mutations in genes involved in N-glycosylation in humans result in Congenital disorders of glycosylation (CDG), which are multi-syndromic and often include neurological symptoms, including intellectual disability (Freeze, 2006; Jaeken & Peanne, 2017; Chang et al, 2018; Ng & Freeze, 2018). Studies in vertebrates and invertebrates have shown that mutants that compromise N-glycan biosynthesis or N-glycan attachment result in defects in cell surface localization of cell adhesion molecules and axon guidance cues (Sekine et al, 2013; Medina-Cano et al, 2018; Mire et al, 2018). The question of whether and how specific classes of N-glycans modulate extracellular pathways or complexes during nervous system development has not been explored.
Here we use PVD somatosensory dendrites, which display complex and stereotyped branching patterns in the nematode Caenorhabditis elegans (Fig.1A) (Oren-Suissa et al, 2010; Smith et al, 2010; Albeg et al, 2011) to investigate the role of different classes of N-glycans during development. We found that aman-2/Golgi alpha-mannosidase II, a conserved enzyme important for the synthesis of complex and paucimannose-type N-glycans is required for PVD dendrite morphogenesis. Specifically, aman-2/Golgi alpha-mannosidase II ensures the correct decoration of the leucine-rich transmembrane receptor DMA-1/LRR-TM in PVD with high-mannose and paucimannose-type N-glycans on specific N-glycosylation sites. Rather than controlling trafficking or surface localization of DMA-1/LRR-TM, we provide evidence that correct N-glycosylation of DMA-1/LRR-TM is essential for the function of DMA-1/LRR-TM as part of the Menorin pathway during PVD patterning. This pathway comprises two conserved cell adhesion molecules SAX-7/L1CAM and MNR-1/Menorin that function from the epidermis, and a secreted chemokine LECT-2/Chondromodulin II from muscle that together with DMA-1/LRR-TM form a high affinity cell adhesion complex (Fig.1B) (Inberg et al, 2019; Sundararajan et al, 2019). Together, our experiments suggest that distinct classes of N-glycans serve specific functions beyond protein folding and localization, and can contribute to developmental specificity during neuronal morphogenesis.
RESULTS
The N-glycosylation enzyme AMAN-2/Golgi alpha-mannosidase II is required for PVD dendrite patterning
To identify additional factors that regulate the Menorin pathway, we performed an unbiased genetic screen for factors that modify a partial loss of function allele of the chemokine lect-2/Chondromodulin II (Diaz-Balzac et al, 2016; Zou et al, 2016). We isolated dz261 as a strong enhancer of the partial loss of function lect-2 mutation in addition to other alleles in known genes of the Menorin pathway (Fig.1A,D, Fig.EV1A-C). Using a combination of mapping, sequencing and transformation rescue we identified dz261 as an allele of aman-2/Golgi alpha-mannosidase II (Fig.1C, Fig.EV1A-C), which encodes a central enzyme in the N-glycosylation biosynthetic pathway not previously implicated in dendrite development. The allele dz261 is likely a complete loss of function mutation as it introduces an early stop codon (Fig.1C, Fig. EV1A-C). Quantifications of PVD branching patterns in aman-2(dz261) null mutants, as well as four additional nonsense/deletion alleles of aman-2, demonstrate that this Golgi alpha-mannosidase II is required for the formation of quaternary branches, but not for the formation of secondary or tertiary branches of PVD dendrites (Fig.1D,E, Fig. EV1D). These observations are reminiscent of hypomorphic alleles of dma-1/LRR-TM (Tang et al, 2019) and suggest that aman-2 may be necessary for full functionality of the Menorin pathway.
AMAN-2/Golgi alpha-mannosidase II positively regulates the Menorin pathway
To directly test the genetic relationship between aman-2/Golgi alpha-mannosidase II and the Menorin pathway, we performed double mutant analyses. We found that the loss of aman-2/Golgi alpha-mannosidase II strongly enhances the severity of PVD branching defects in partial loss-of-function mutants of lect-2/Chondromodulin II (gk864764) and mnr-1/Menorin (dz213, Ramirez-Suarez & Bülow, unpublished) (Fig.1A,E,F,G), both of which are essential components of the conserved Menorin cell adhesion complex, and act as positive regulators of PVD development (Fig.1E-G) (Dong et al, 2013; Salzberg et al, 2013; Diaz-Balzac et al., 2016; Zou et al., 2016). In contrast, loss of aman-2/Golgi alpha-mannosidase II suppressed the self-avoidance defects of tertiary dendrites in partial loss of a function mutation of kpc-1/Furin (dz254) (Fig.1H,I, Fig. EV1C), a known negative regulator of the Menorin pathway (Schroeder et al, 2013; Salzberg et al, 2014; Dong et al, 2016). Therefore, we conclude that aman-2/Golgi alpha-mannosidase II normally functions to positively regulate the Menorin pathway to ensure correct PVD dendrite patterning.
AMAN-2/Golgi alpha-mannosidase II does not serve obvious functions in regulating transport or abundance of the DMA-1/LRR-TM
Mutations in N-glycosylation are often associated with protein folding defects and trafficking blocks due to misfolding (Stanley et al., 2015) and can, for example, result in lower abundance of cell surface proteins such as cell adhesion proteins in the nervous system (Medina-Cano et al., 2018). Because protein misfolding is more likely to occur at elevated temperatures (Gasser et al, 2008; Vabulas et al, 2010), we tested whether PVD branching defects in aman-2/Golgi alpha-mannosidase II mutations get progressively more severe with increasing temperatures. We found no significant increase in dendrite branching defects at 25°C compared to 15°C in aman-2(gk248486) mutant animals, in contrast to hypomorphic lect-2(gk864764) mutant animals (Fig. EV2A, B). Previous work showed that mutations causing a secretory block as a result of a defective unfolded protein response trap a DMA-1::GFP reporter in the cell body of PVD (Wei et al, 2015; Salzberg et al, 2017). We therefore questioned whether the loss of aman-2/Golgi alpha-mannosidase II can lead to defects in protein folding and trafficking, and a possible secretory block. We analyzed the amount and number of puncta of the DMA-1::GFP reporter in the soma, as well as in dendrite branches, and found that DMA-1::GFP fluorescence in both the soma and primary dendrites, and the number of DMA-1::GFP puncta in tertiary dendrites remained unaffected in aman-2(gk248486) mutant animals (Fig.EV2C-E). Moreover, localization or abundance of LECT-2/Chondromodulin II and SAX-7/L1CAM were also not obviously affected by loss of aman-2/Golgi alpha-mannosidase II (data not shown). Taken together, these findings suggest that AMAN-2/Golgi alpha-mannosidase II does not primarily function to ensure protein folding, stability, or transport of factors of the Menorin pathway, but may rather regulate more specific aspects of the Menorin pathway during PVD patterning.
Enzymatic activity of AMAN-2/Golgi alpha-mannosidase II is required cell-autonomously in PVD to form higher order branches
The octasaccharide GnMan5Gn2 in a specific linkage configuration is the unique precursor to hybrid, complex, and paucimannose N-glycans (Fig.2A,B;Fig.3A) (Moremen, 2002; Paschinger et al, 2019). AMAN-2 is a Golgi alpha-mannosidase II, which is conserved from yeast to humans, and cleaves two specific mannose residues from GnMan5Gn2, thereby generating the substrate for formation of complex and paucimannose., N-glycans (Fig.2A,B;Fig.3A) (Moremen, 2002; Paschinger et al., 2019). To determine where AMAN-2/Golgi alpha-mannosidase II functions and whether enzymatic activity is required for its role in PVD dendrite morphogenesis, we investigated transgenic expression of a wildtype AMAN-2 cDNA under the control of heterologous promoters in PVD, muscle or epidermis for their ability to rescue aman-2 mutant defects. Expression in PVD, but not muscle or epidermis rescued the enhanced phenotypes in PVD dendrite branching of lect-2(gk864764); aman-2(gk248486) and mnr-1(dz213) aman-2(gk248486) double mutants (Fig.2C). These observations suggest that AMAN-2/Golgi alpha-mannosidase II functions cell-autonomously to pattern PVD dendritic arbors.
Since AMAN-2/Golgi alpha-mannosidase II canonically functions as an enzyme (Moremen, 2002; Shah et al, 2008), we next asked whether catalytic activity is required for its role in PVD dendrite branching. We approached this both genetically and pharmacologically. Prior studies showed that two highly-conserved aspartates are part of the conserved catalytic site in AMAN-2/Golgi alpha-mannosidases II (D306 and D443) and act sequentially to cleave off two mannose residues (Fig.2D) (Shah et al., 2008). We found that an AMAN-2 cDNA with both aspartates mutated, and hence likely catalytically dead, failed to rescue the defects in lect-2(gk864764); aman-2(gk248486) and mnr-1(dz213) aman-2(gk248486) double mutants (Fig. 2C). To address the possibility that mutating the catalytic residues compromised the stability or structure of AMAN-2, we took advantage of swainsonine, a compound that specifically inhibits Golgi alpha-mannosidase II (Lu et al, 2014). We found that exposing animals to swainsonine resulted in PVD defects that were indistinguishable from the effects of a null mutation in aman-2, either in combination with a partial loss of function allele of lect-2, or in wild type animals (Fig. 2E-G). In other words, the pharmacological inhibition of AMAN-2/Golgi alpha-mannosidase II activity resulted in the same phenotypic consequences as genetically inactivating or removing the enzyme. Collectively, these findings lead us to conclude that the catalytic activity of AMAN-2/Golgi alpha-mannosidase II is essential to support dendrite patterning in PVD. This further implies that N-glycosylation of a molecule expressed in PVD is crucial for normal dendrite arborization.
The presence of abnormal N-glycans in aman-2/Golgi alpha-mannosidase II mutants results in defective PVD arborization
In eukaryotes, N-glycosylation is initiated in the Endoplasmic Reticulum (ER) with the synthesis of a 14-saccharide glycan on the phosphorylated polyisoprenol lipid dolichol-P-P (Stanley et al., 2015). Subsequently, the saccharide is transferred by a multiprotein complex termed oligosaccharyltransferase (OST) from dolichol-P-P to the aspartate within a NXS/T motif in nascent proteins as they are translocated into the ER (Stanley et al., 2015). As N-glycosylated proteins transit the Golgi, the glycans undergo a series of enzymatic modifications that add and remove specific sugar residues to lead to a wide array of possible N-glycan structures (Fig.3A) (Stanley et al., 2015). For example, in one of the earlier steps, the enzyme MGAT1 adds a N-acetylglucosamine residue to Man5Gn2 to form GnMan5Gn2 (Fig.3A). Genetically removing MGAT1 in mice results in complete loss of complex and hybrid N-glycans and early embryonic death, demonstrating that these N-glycans are essential for mammalian development (Ioffe & Stanley, 1994; Metzler et al, 1994). GnMan5Gn2 is the substrate for AMAN-2/MAN2A, which sequentially removes two mannose residues to form GnMan3Gn2 (Fig.3A) (Stanley et al., 2015). These reactions are followed by either removal or addition of additional sugars, or modification by a host of other conserved enzymes that lead to either complex or paucimannose N-glycans (Fig.3A). To determine which specific N-glycans are missing in aman-2 mutants, and are therefore required for branching of PVD dendrites, we systematically tested whether mutations in any of the genes downstream of aman-2 (including hex-2/hexosaminidase, hex-3/hexosaminidase, fut-8/FUT8 Fucosyltransferase, and gly-20/MGAT II) would also enhance the partial lect-2 loss of function allele. We found that removing the genes encoding these enzymes alone, or in combination, did not enhance the partial lect-2 loss of function allele (Fig.EV3A). Mutating MGAT1/N-acetylglucosaminyltransferase-I (in worms encoded by three paralogous genes gly-12, gly-13, gly-14) (Chen et al, 1999; Chen et al, 2003), which acts immediately before AMAN-2/Golgi alpha-mannosidase II, also showed no effects (Fig.EV3B). Collectively, these findings suggest that no lack of specific N-glycans downstream of MGAT1, or of AMAN-2, alone are responsible for the observed defects in PVD dendrites.
Previous structural studies of N-glycans in aman-2(tm1078) null mutant animals (Paschinger et al, 2006) established that loss of aman-2/Golgi alpha-mannosidase II in C. elegans caused (1) a loss of the normal products of AMAN-2, including complex N-glycans (Fig.3A, shaded in green) and (2) a buildup of GnMan5Gn2, the substrate of AMAN-2 (Fig.3A)(Paschinger et al., 2006). This GnMan5Gn2 intermediate was found to serve as substrate for enzymes further downstream (including FUT-8/Fut8 Fucosyltransferase and PCT/Phosphorylcholine-transferase) leading to the appearance of abnormal N-glycans, not normally present in wildtype animals (Fig.3a, shaded in red) (Paschinger et al., 2006). To determine whether the defects in PVD branching were caused by the absence of wild type N-glycans, or the presence of abnormal GnMan5Gn2 N-glycans, we mutated both MGAT1 and AMAN-2 in the partial loss of function backgrounds of lect-2 and mnr-1. The prediction was that, if the enhancement of the partial loss of function alleles lect-2 or mnr-1 by loss of aman-2 is caused by abnormal N-glycans, then removal of MGAT1, the preceding enzyme would suppress that enhancement. Indeed, loss of MGAT1 did suppress the enhanced phenotypes in lect-2; aman-2 and mnr-1; aman-2 double mutants (Fig.3B-D). These data indicate that one or more structurally abnormal N-glycans with a terminal N-acetylglucosamine are responsible for the observed defects in PVD patterning.
DMA-1/LRR-TM N-glycosylation is changed in aman-2/Golgi alpha-mannosidase II mutants
Since we demonstrated that (1) aman-2/Golgi alpha-mannosidase II genetically interacts with the Menorin pathway, and (2) AMAN-2/Golgi alpha-mannosidase II activity is required cell-autonomously in PVD to regulate branching, we hypothesized that AMAN-2/Golgi alpha-mannosidase II directly regulates N-glycans on at least one component of the Menorin complex in PVD. Treatment of whole worm lysates with the bacterial PNGase F glycosidase, which cleaves all N-glycans from Asn, resulted in distinct downward shifts in molecular weight of both DMA-1 and KPC-1, indicating that N-glycans were present on both proteins in vivo and had been removed (Fig.3E,EV4C). Thus, both PVD-expressed proteins, DMA/LRR-TM and KPC-1/Furin, are N-glycosylated, consistent with a previous report for DMA-1/LRR-TM (Feng et al, 2020). Another cell-autonomous factor and possible candidate, HPO-30/Claudin (Smith et al, 2013), contains no predicted N-glycan consensus motifs.
We next determined whether loss of AMAN-2 resulted in altered N-glycosylation. Interestingly, the absence of aman-2/Golgi alpha-mannosidase II resulted in a clear increase in DMA/LRR-TM molecular weight, whereas the size of KPC-1/Furin remained unaffected (Fig.3F,EV4C). The upward shift in DMA-1 size following the loss of aman-2 is consistent with our genetic data establishing that the presence of larger, abnormal GnMan5Gn2 N-glycans gives rise to the PVD mutant phenotype. To determine what types of N-glycans are attached to DMA-1 in wild type and aman-2 mutant backgrounds, we treated lysates with additional endoglycosidases: Endo H, which cleaves hybrid/high-mannose N-glycans, and Endo D, which cleaves only paucimannose N-glycans. Based on the size of the shifts observed, we conclude that in wildtype animals, DMA-1 possesses primarily, but not exclusively, hybrid/high-mannose N-glycans with a smaller amount of paucimannose N-glycans. In contrast, in the absence of AMAN-2, all N-glycans on DMA-1/LRR-TM were converted to hybrid/high-mannose type (likely GnMan5Gn2-derived N-glycans) with no or little detectable paucimannose N-glycans (Fig.3G,EV4A,B). Additionally, proteomic studies identified C. elegans SAX-7/L1CAM and LECT-2/Chondromodulin II as glycoproteins (Kaji et al, 2007), which we confirmed by a shift in molecular weight upon digestion of all N-glycans by PNGase F (Fig.EV4D,E). However, in the absence of aman-2/Golgi alpha-mannosidase II, neither SAX-7/L1CAM and LECT-2/Chondromodulin II displayed obvious changes in molecular weight, suggesting that they did not carry abnormal N-glycans in an aman-2 mutant background. Collectively, our data show that among the N-glycosylated proteins of the Menorin complex, only DMA-1/LRR-TM was significantly affected by the loss of aman-2/Golgi alpha-mannosidase II and carried altered N-glycans in aman-2 null animals.
AMAN-2/Golgi alpha-mannosidase II modulates N-glycans on DMA-1/LRR-TM to regulate PVD morphogenesis
The DMA-1/LRR-TM receptor contains four predicted N-glycosylation motifs, all of which reside in leucine rich repeats (Fig.4A). To establish whether the N-glycosylation of DMA-1/LRR-TM is essential for its role in PVD dendrite branching, we mutated all four sites, alone and in combinations. Using CRISPR/Cas9-based genome editing, we converted the asparagine residues of the four predicted N-glycan attachment sites to glutamine to maintain chemical similarity but eliminate the possibility of N-glycosylation. We found that only when abolishing predicted N-glycan attachment site 4 (N386), alone or in combination with other sites, was PVD quaternary branching compromised (Fig.4B). These results reveal that N-glycosylation of DMA-1/LRR-TM is required for its role in PVD patterning of quaternary branches and highlight the importance of N386 within a membrane proximal LRR repeat. We then assessed whether abolishing N-glycan attachment sites in an aman-2/Golgi alpha-mannosidase II loss of function background had any effects on dendrite patterning. Analysis of different site-specific mutations revealed that some N-glycosylation sites on DMA-1/LLR-TM enhance the severity of PVD dendrite branching defects in an aman-2 mutant background (Fig.4C-D). The results in an aman-2 mutant background suggest that having some type of N-glycan on site 4 of DMA-1/LRR-TM, even if abnormal, is better than having no N-glycan at all (cf. S3 and S123); second having such abnormal N-glycans on sites 1-3 further compromises DMA-1 function during PVD development (cf. S4 in control vs aman-2 mutant background, Fig.4B,C). Lastly, the mutant phenotype resulting from a presumptive loss of all N-glycans on DMA-1 (S1234) is enhanced in an aman-2 mutant background, suggesting that N-glycosylated proteins other than DMA-1 may serve additional functions during PVD morphogenesis or that abnormal N-glycans on cryptic N-glycosylation sites further compromise function (Fig.4C).
DISCUSSION
While prior studies established the importance of N-glycosylation during neuronal development, our studies establish an important role for specific classes of N-glycans in mediating neuronal development, and specifically dendrite patterning. They suggest that the DMA-1/LRR-TM receptor in PVD must be decorated with specific hybrid/high-mannose or paucimannose N-glycans as a result of AMAN-2 activity and that these specific N-glycans are important for functioning of the Menorin complex in PVD dendrite morphogenesis. A possible explanation is that the N-glycans on specific N-glycosylation sites in DMA-1/LRR-TM function permissively to maintain high affinity binding of DMA-1/LRR-TM to other members of the Menorin complex through specific N-glycans (Fig.4E). This interaction could be compromised by the formation of abnormal glycans in the absence of AMAN-2 leading to destabilization of the complex. Therefore, our genetic and biochemical data, together with analytical data of N-glycans in aman-2/Golgi alpha-mannosidase II mutants (Paschinger et al., 2006), underscore the importance of AMAN-2/Golgi alpha-mannosidase II as a linchpin for the creation of complex and paucimannose N-glycans, and avoidance of larger, abnormal hybrid and high-mannose-type N-glycans. Since metabolite availability can influence N-glycan synthesis and flux (reviewed in (Dennis et al, 2009)), these findings also raise the possibility that environmental factors can intersect with intrinsic genetic programs to regulate extracellular adhesion complexes during neural development by modulating N-glycosylation.
Previous studies demonstrated that N-glycosylation is important for folding and surface localization of cell adhesion molecules and axon guidance factors, such as L1CAMs and ephrins, respectively (Sekine et al., 2013; Medina-Cano et al., 2018; Mire et al., 2018). On the other hand, in vitro experiments suggested that N-glycans can regulate protein-protein interactions of cell adhesion molecules (Fogel et al, 2010; Labasque et al, 2014). Our findings demonstrate in vivo, that not only N-glycans per se, but that specific classes of N-glycan structures are important to modulate cell-cell signaling, and possibly, receptor-ligand binding and complex formation. This is reminiscent of the role of O-fucose glycans on the Notch receptor extracellular domain, which affect its signaling and ligand interactions (Moloney et al, 2000). Given that over 70% of proteins transiting the secretory pathway are N-glycosylated (Apweiler et al., 1999), these findings raise the possibility that specific N-glycan structures are important determinants to regulate the interactions of extracellular complexes during nervous system development more broadly and could contribute to the specificity that is required in the nervous system. In this context it is interesting to note that over 70 congenital disorders of glycosylation have been described that affect genes in the N-glycosylation pathway (Freeze, 2006; Ng & Freeze, 2018), of which many are associated with intellectual disability or other neurological symptoms (Jaeken & Peanne, 2017; Chang et al., 2018). While no mutations in Golgi alpha-mannosidase II in humans have been described to date, it is conceivable that such mutations exist, and may result in neurological phenotypes. Regardless, our studies provide the conceptual framework for studies into developmental defects of the nervous system in mutants of genes involved in N-glycosylation and this growing class of congenital disorders.
AUTHOR CONTRIBUTIONS
Conceptualization Ideas: MR, HEB; Validation: MR; Formal Analysis: MR, HEB; Investigation: MR; Resources: NJRS, CADB; Writing – Original Draft: MR; Writing – Review & Editing: MR, NJRS, CADB, HEB; Visualization Preparation: MR; Project Administration: HEB; Funding Acquisition: MR, NJRS, CADB, HEB.
DECLARATION OF INTERESTS
The authors declare no competing interests.
MATERIALS AND METHODS
C. elegans handling
All strains were maintained using standard methods (Brenner, 1974) and experiments were performed at 20°C, except where indicated otherwise. Phenotypic analysis was performed in 1-day-old adults, with no more than 4-5 eggs present. For details and a complete list of strains used and generated in this study, see resources table.
Cloning of mutant alleles
The dz261 allele was obtained from a forward genetic screen for modifiers of the lect-2(gk864764) hypomorphic allele. Using a combination of whole genome sequencing and single nucleotide polymorphism mapping (Minevich et al, 2012), we narrowed down the region to a 5Mb interval (11MB-16Mb on chromosome V (Fig.EV1A). This region contained 7 polymorphisms with predicted functional consequences. We injected seven fosmids in pools and found that only the pool which contained a fosmid covering aman-2 resulted in rescue (Fig.EV1B). In addition, we obtained three nonsense alleles in aman-2(gk248486, gk248477, gk619253) from the Million Mutation Project (Thompson et al, 2013) and one deletion allele (tm1078, kind gift from the Mitani lab).
Details of genetic screen and cloning
The lect-2(gk864764) hypomorphic strain was treated with EMS in accordance with standard chemical mutagenesis protocols (Kutscher & Shaham, 2014) and F1 progeny were scored for enhancement, suppression, or modification of PVD branching phenotype. A SNP-mapping-WGS (Doitsidou et al, 2016) approach was used to map aman-2(dz261) between 10Mb and 16Mb of chromosome V. This region contained seven candidates with nonsense, missense, frameshift or splice site mutations, including one in aman-2. The dz261 mutation was further confirmed by Sanger sequencing of the original isolate, identifying a nonsense mutation W237Opal in the aman-2 locus.
Heterologous rescue of PVD dendrite branching defect
The aman-2 cDNA was cloned under control of heterologous promoters: hypodermal Pdpy-7 (Gilleard et al, 1997), body wall muscle Pmyo-3 (Okkema et al, 1993), and PVD Pser2prom3 (Tsalik et al, 2003). All constructs were injected at 5 ng/μl into wdIs52 II; him-5(ok1896) V together with the Pmyo-2::mCherry marker at 50 ng/μl and pBluescript at 50 ng/μl. Males from transgenic lines were then crossed into lect-2(gk864764) II; wyIs581 IV; aman-2(gk248477)V and wyIs581 IV; mnr-1(dz213) aman-2(gk248486) V.
Cloning constructs and transgenesis
To assemble tissue specific expression constructs used for heterologous rescue experiments, the aman-2 cDNA clone yk11g705 (kind gift of Yoji Kohara) and cloned under control of the following promoters: PVD ser2prom-3 (Tsalik et al., 2003), hypodermal dpy-7p (Gilleard et al., 1997), body wall muscle myo-3p (Okkema et al., 1993). These constructs were injected at 5 ng/μl together with myo-2p::mCherry as an injection marker at 50 ng/μl, and BlueScript as DNA filler. Point mutants in the aman-2 cDNA were introduced by site-specific mutagenesis (NEB Q5 Site-Directed
Mutagenesis). All plasmids contained the unc-54 3’UTR.
Pharmacology
Experiments in which the activity of AMAN-2 was blocked pharmacologically were performed with the compound swainsonine (1 mg swainsonine #16860 vials, Fisher Scientific catalog #NC1670046). Dosage experiments were performed from 50-500μm of swainsonine in dissolved in agar of NGM plates, with 300μm being sufficient to elicit phenotypes in PVD. After drying for 24 hours, 200μL of OP50 E. coli was seeded onto each plate, and 5 young adult worms were left to self-fertilize and lay eggs. The F1 generations were analyzed, imaged, and quantified. DMSO was used a control in plates not treated with swainsonine.
Molecular Biology
Immunoprecipitation and Western blot analyses of glycoproteins were performed using standard SDS-PAGE methods. Whole C. elegans lysates were prepared in RIPA buffer and sonicated in a Biorupter benchtop waterbath sonicator for 15 minutes. Lysates were treated with 1 unit at temperatures as indicated in NEB protocols (linked in Key Resources Table) endoglycosidases PNGase F, Endo H, or Endo D where specified. Overnight immunoprecipitation of lysates with anti-GFP antibody prior to SDS-PAGE was performed for proteins with low expression levels (DMA-1::GFP and KPC-1::GFP).
Immunoprecipitation and Western blot Analysis
Five full plates of DMA-1::GFP and KPC-1::GFP tagged worms were washed in RIPA buffer pH7.0 and lysed for 15 minutes in a Biorupter water bath used previously published methods of whole worm protein extraction (Li & Zinovyeva, 2020). Post lysis, 20 uL of Protein A/G Plus Agarose beads (Santa Cruz sc-2003) and 1 uL of anti-GFP antibody (Roche 11814460001) were used to pull down DMA-1::GFP and KPC-1::GFP overnight at 4C. Ten gravid adult SAX-7::GFP::FLAG animals and twenty LECT-2::mNG::FLAG animals were sufficient to see robust expression post Western Blot. These samples were boiled and loaded directly into the gels. Gradient gels (4-12% GenScript) were used in all experiments. For all anti-FLAG blots, a concentration of 1:800 anti-Flag (Sigma F1804) and 1:5000 anti-mouse HRP (Millipore AP308P) were used. For all anti-GFP blots, a concentration of 1:500 anti-GFP (Roche 11814460001) and 1:5000 anti-mouse HRP (Millipore AP308P) were used.
CRISPR/Cas9 mediated gene editing
CRISPR-Cas9 constructs were designed and dpy-10 co-crispr protocol followed as previously described (Dickinson & Goldstein, 2016) to make single point mutations of predicted N-glycan attachment Asparagine residues. Predicted sites were identified using NetNGlyc 1.0 Server (Gupta & Brunak, 2002). A battery of guideRNAs were designed to direct Cas9 cuts near sites of interest, and homologous repair template oligomers were designed to mutate Asparagine residues to Glutamine. All plasmids were delivered via microinjection in the gonads of animals. Strains with combinations of mutated sites were generated by sequential injections and/or multiple simultaneous edits. Note, that all edits were made in strains with the C-terminus of DMA-1 already tagged with a 2XFLAG immunotag before the PDZ binding domain (Dong et al., 2016). Strains EB4219 through EB4238 in the Key Resources Table were obtained using these methods.
Imaging
Fluorescent images were captured in live C. elegans using a Plan-Apochromat 40×/1.4 or 63x/1.4 objective on a Zeiss Axioimager Z1 Apotome. Worms were immobilized using 1 mM Levamisole and Z stacks were collected. Maximum intensity projections were used for further analysis and tracing of dendrites. For quantification of branching, 1-day-old adults were mounted onto slides and immobilized with 1mM Levamisole. In both cases of capturing images and counting, and counting live on the microscope, the number of “Ts” (secondary and tertiary branches), “Os,” (self-avoidance defects), and/or quaternary branches within 100 μm of the primary branch anterior to the cell body were quantified.
Quantification and statistical analysis
Statistical comparisons were conducted on Prism 8 GraphPad Software using Mann-Whitney, Kruskal-Wallis, Z-, or two-sided ANOVA tests as appropriate. Statistical significance is indicated as ns, not significant; *p≤0.05; **p≤0.01; ***p≤0.001 and ****p≤0.0001. This study includes no data deposited in external repositories.
EXPANDED VIEW FIGURES
ACKNOWLEDGEMENTS
We thank Yehuda Salzberg, Pamela Stanley, Robert Townley, Peri Kurshan and members of the Bülow laboratory for comments on the manuscript and helpful discussions during the course of this work. We thank Kang Shen, Iain Wilson, Shohei Mitani for reagents, and Yuji Kohara for the yk11g705 cDNA clone. We are grateful to Meera Trivedi for sharing the dzIs117 strain prior to publication. Some strains were provided by the Caenorhabditis Genome Center (funded by the NIH Office of Research Infrastructure Programs P40 OD010440). This work was supported by grants from the National Institute of Health (NIH): R01NS096672 and R21NS111145 to HEB; F31NS100370 to MR; T32GM007288 and F31HD066967 to CADB; P30HD071593 to Albert Einstein College of Medicine. NJRS was the recipient of a Colciencias-Fulbright Fellowship and HEB of an Irma T. Hirschl/Monique Weill-Caulier research fellowship.