couscous (PTSG_07368) encodes an 866-residue predicted glycosyltransferase of the choanoflagellate Salpingoeca rosetta. The protein has an N-terminal hydrophobic signal sequence/anchor, a PAN/Apple domain, and a C-terminal alpha-mannosyltransferase domain (Pfam Mannosyl_trans3) belonging to the fungal MNN2/MNN5 alpha-1,2-mannosyltransferase family, with which it shares 28-35% identity and a conserved DXD motif. Its enzymatic activity, sugar donor and acceptor substrates have not been determined. A frameshift loss-of-function allele (couscous lw1, originally isolated as the "Branched" mutant) abolishes development of clonal multicellular rosettes and instead causes cells to aggregate promiscuously into shear-sensitive clumps, including with wild-type cells; both defects are rescued by transgenic wild-type couscous. Mutant cells lose a basal patch of jacalin-reactive glycans that normally marks the basal extracellular matrix and fills the rosette interior, and they fail to upregulate and secrete the C-type lectin Rosetteless upon rosette induction, suggesting that Couscous-dependent glycosylation contributes to maturation of the basal extracellular matrix. A tagged, overexpressed Couscous fusion localized to cytoplasmic puncta, the collar and the cell membrane rather than to the Golgi apparatus, so its native compartment is unresolved. couscous expression is regulated by the Hippo pathway components Warts and Yorkie.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0000026 alpha-1,2-mannosyltransferase activity | IEA GO_REF:0000118 | ACCEPT | Summary: TreeGrafter transfer from the PANTHER PTHR31646 (alpha-1,2-mannosyltransferase MNN2) family node. Couscous carries a Mannosyl_trans3 domain that aligns with the alpha-mannosyltransferase domain of yeast and Candida MNN2 at 28-35% identity, including the conserved DXD motif required for nucleotide-sugar/metal binding in GT-A fold glycosyltransferases [PMID:30556809]. No enzymatic assay has been performed, so the activity, linkage and acceptor remain predictions. Reason: The catalytic domain places Couscous within a family whose characterized members (S. cerevisiae Mnn2, Mnn5) are alpha-1,2-mannosyltransferases, and the DXD motif is retained, so there is no sequence evidence of catalytic loss. Loss-of-function mutants show a specific, rescuable glycan (jacalin) defect consistent with a glycosyltransferase [PMID:30556809]. This is the best-supported description of the molecular activity, accepted as a homology-based prediction; the alpha-1,2 linkage specificity in particular has not been tested in a choanoflagellate and the acceptor is unknown. Supporting Evidence: PMID:30556809 The predicted mannosyltransferase domain shares 28% and 35% amino acid sequence identity to alpha 1–2 mannosyltransferase (MNN2) proteins in Saccharomyces cerevisiae and Candida albicans, respectively, including the conserved DXD motif found in many families of glycosyltransferases PMID:30556809 While we have not uncovered the target(s) of the predicted glycotransferases or the exact nature of the interplay between the two phenotypes |
| GO:0000139 Golgi membrane | IEA GO_REF:0000044 | UNDECIDED | Summary: UniProt subcellular-location mapping (ARBA rule) assigning the family-typical Golgi type II membrane protein topology. The only experimental localization of Couscous (C-terminal mWasabi fusion overexpressed from the efl promoter) found puncta in the cytosol, collar and cell membrane, and the authors state it was clearly not at the Golgi, while cautioning that tagging or overexpression may have mislocalized it [PMID:30556809]. Reason: The family default (Golgi, as for yeast Mnn2) conflicts with the one direct observation, which placed tagged Couscous outside the Golgi (possibly ER, untested). The authors themselves conclude localization is uncertain, and the N-terminal hydrophobic segment is ambiguous between a cleaved signal peptide and a type II signal anchor. Neither accepting nor removing the Golgi assignment is justified until native localization is determined. Supporting Evidence: PMID:30556809 While Couscous-mWasabi was clearly not localized to the Golgi, the puncta may co-localize with the ER, where glycosyltransferases are also known to function PMID:30556809 Therefore, we are currently uncertain about the subcellular localization of Couscous protein. |
| GO:0005794 Golgi apparatus | IEA GO_REF:0000118 | UNDECIDED | Summary: TreeGrafter transfer of the Golgi location of fungal MNN2-family mannosyltransferases. Tagged Couscous did not localize to the Golgi in S. rosetta, in contrast to the other rosette-defect glycosyltransferase Jumble, which did [PMID:30556809]. Reason: Same conflict as for Golgi membrane: a family-level expectation versus a single overexpressed-fusion observation that the authors regard as possibly artefactual. Kept undecided pending native localization (antibody staining or endogenous tagging with a Golgi marker). Supporting Evidence: PMID:30556809 Consistent with its role as a glycosyltransferase, Jumble localized to the Golgi apparatus, but Couscous appeared to localize in cytoplasmic puncta and to the cell membrane. |
| GO:0009101 glycoprotein biosynthetic process | IEA GO_REF:0000120 | ACCEPT | Summary: Combined ARBA/InterPro inference from the alpha-mannosyltransferase domain. MNN2-family enzymes extend the mannose outer chains of N- and O-linked glycans on glycoproteins. Couscous loss-of-function mutants lose a basal patch of jacalin-reactive glycans in the extracellular matrix, rescued by the wild-type gene [PMID:30556809]. Reason: A glycosyltransferase would do the catalytic work of adding sugars to glycoproteins, so this process term fits the predicted activity. It is general enough to cover the uncertainty about which glycan class (N- or O-linked) and which acceptor proteins are modified. The rescuable glycan defect in the mutant supports a role in glycoconjugate synthesis, although the authors do not exclude an indirect trafficking effect. No more specific child term (e.g. protein N-linked glycosylation) is justified without acceptor identification. Supporting Evidence: PMID:30556809 The predicted Couscous amino acid sequence contains a specific type of glycosyltransferase domain, an alpha-mannosyltransferase domain, that transfers activated mannose onto the outer chain of core N-linked polysaccharides and O-linked mannotriose PMID:30556809 The same was true for Couscous cells, in which transformation with couscous-mTFP rescued both rosette development and the wild type glycosylation pattern |
| GO:0016020 membrane | IEA GO_REF:0000044 | ACCEPT | Summary: UniProt keyword mapping from a predicted single-pass (type II) membrane topology. Tagged Couscous was seen in puncta and at the cell membrane and collar [PMID:30556809]. Reason: Uninformative but consistent with both the predicted N-terminal membrane anchor and the observed membrane/punctate distribution of the tagged protein. It makes no compartment-specific claim, so it is not undermined by the unresolved Golgi versus ER question. Supporting Evidence: PMID:30556809 In wild type cells transfected with a couscous-mWasabi transgene under the efl promoter, Couscous was found in puncta scattered throughout the cytosol, collar and cell membrane |
| GO:0016757 glycosyltransferase activity | IEA GO_REF:0000002 | ACCEPT | Summary: InterPro2GO mapping from IPR022751 (alpha-mannosyltransferase). Couscous is a predicted glycosyltransferase with a conserved DXD motif; activity has not been assayed [PMID:30556809]. Reason: Correct at the general level and consistent with the more specific alpha-1,2-mannosyltransferase prediction. Retained as a companion IEA; it is the safest statement of the activity given that sugar and linkage specificity have not been measured. Supporting Evidence: PMID:30556809 Couscous maps to a predicted mannosyltransferase with a PAN/Apple domain. |
| GO:0046354 mannan biosynthetic process | IEA GO_REF:0000118 | REMOVE | Summary: TreeGrafter transfer from the fungal MNN2 family, where Mnn2 builds the branched mannose outer chains of cell-wall mannoproteins. The GO term is defined as synthesis of mannan, "the main hemicellulose of soft (coniferous) wood, made up of D-mannose, D-glucose and D-galactose". S. rosetta has no fungal-type mannan cell wall, and the only glycan change detected in couscous mutants is loss of a jacalin-reactive glycan at the basal pole, not a mannan [PMID:30556809]. Reason: The term describes polysaccharide (hemicellulose/cell-wall mannan) biosynthesis, a fungal/plant-lineage function that the TreeGrafter node carries over from yeast Mnn2. There is no evidence that choanoflagellates synthesize a mannan polysaccharide, and Couscous substrates are unknown. The appropriate process for a glycoprotein-modifying mannosyltransferase is already captured by glycoprotein biosynthetic process. This is an over-propagated electronic inference that can be argued against on biological grounds. Propagation Review Root cause: PROPAGATION BAD Failure modes: LINEAGE OR TAXON MISMATCH ROLE CONFLATION Sources checked: PANTHER:PTN001270341 · MNN2-family alpha-1,2-mannosyltransferase node (PTHR31646) SUPPORTS SOURCE BUT NOT TARGET The GO:0046354 IBD on this node is seeded only by Candida albicans MNN2-family genes (all 56 IBA rows from the node list the same six CGD seeds; QuickGO 2026-10-01), yet it reaches oomycetes, green algae and choanoflagellates. A fungal cell-wall mannan process does not transfer to a choanoflagellate enzyme acting on the rosette basal matrix. Supporting Evidence: PMID:30556809 MNN2 proteins catalyze the addition of the first branch of mannose-containing oligosaccharides found on cell wall proteins PMID:30556809 In contrast with wild type cells, the basal patch of jacalin staining was absent or significantly diminished in Couscous and Jumble mutants |
Loading supporting content…
Download this section (compressed HTML)Q: Does Couscous have mannosyltransferase activity in vitro, and what are its donor (GDP-mannose?), linkage and acceptor specificities?
Q: Where does native Couscous reside - Golgi, ER, or a post-Golgi/secretory compartment - and is its N-terminal hydrophobic segment a cleaved signal peptide or a type II signal anchor?
Q: Which glycoproteins carry the Couscous-dependent basal jacalin epitope, and is Rosetteless (with its mucin-like Ser/Thr repeats) or a cadherin among them?
Q: Is the loss of basal glycans in couscous mutants caused by missing glycosylation or by defective trafficking of glycoconjugates to the basal pole?
Q: What does the PAN/Apple domain contribute - substrate recognition, dimerization or binding to extracellular partners?
Q: How does Hippo pathway (Warts/Yorkie) regulation of couscous expression relate to rosette size control?
Experiment: Express the soluble luminal domain (with and without the PAN domain, and a DXD-motif mutant) and assay transfer from GDP-mannose to mannoside acceptors (e.g. alpha-1,6-mannobiose, Man-alpha-O-Ser/Thr peptides), with linkage determined by NMR or exoglycosidase digestion.
Hypothesis: Couscous is an active alpha-1,2-mannosyltransferase.
Type: biochemical enzyme assay
Experiment: Endogenously tag couscous at its locus by CRISPR knock-in (avoiding overexpression) and co-image with ER (HDEL) and Golgi markers; complement with immunostaining using an anti-Couscous antibody.
Hypothesis: Native Couscous resides in an early secretory compartment.
Type: localization / imaging
Experiment: Jacalin affinity enrichment and comparative glycoproteomics of wild-type versus couscous knockout cells, with site-level glycan mapping on Rosetteless and cadherins.
Hypothesis: Couscous glycosylates specific ECM or adhesion glycoproteins, including Rosetteless.
Type: glycoproteomics
Experiment: Complement couscous knockout cells with wild-type couscous versus a DXD-motif catalytic mutant and score rosette formation, clumping and basal jacalin staining.
Hypothesis: Couscous catalytic activity, not only protein presence, is required for rosette development.
Type: genetic complementation
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)