jumble (Salpingoeca rosetta; UniProt F2TWH0; PTSG_00436; GenBank EGD72416) - curation notes
Note: automated deep research (e.g. just deep-research ... --provider perplexity) was
unavailable in this environment (no provider API keys). There is therefore no
jumble-deep-research-*.md file. These notes were compiled manually from the cached
publications in publications/ and from direct database queries (UniProt, InterPro API).
Identity
- UniProtKB F2TWH0 (TrEMBL, unreviewed, PE 4 "Predicted"), 467 aa, ORF name PTSG_00436,
described in UniProt only as "Uncharacterized protein".
- UniProt features: one predicted N-terminal helical transmembrane segment (residues 12-30,
Phobius), two disordered regions (73-125, 368-389). No domain, family or GO annotation in
UniProt; GOA has zero annotations.
- InterPro API (queried 2026-09-30):
https://www.ebi.ac.uk/interpro/api/protein/uniprot/F2TWH0
returns "entries":0, "domain_architectures":0; the entry/.../protein/uniprot/F2TWH0
endpoints return HTTP 204 (no content). So no InterPro/Pfam/PANTHER member-database
signature is called on this protein. This agrees with the primary paper:
PMID:30556809
- Topology: a single N-terminal TM segment followed by a large C-terminal region is the
classic type II membrane topology of Golgi glycosyltransferases, but that is my inference
from the UniProt/Phobius feature, not something tested.
Primary paper: Wetzel et al. 2018 eLife (PMID:30556809)
Mutant phenotype (demonstrated)
- Jumble is an EMS-induced mutant from the Levin et al. 2014 screen
PMID:30556809
(listed as "Mutant class E Jumble" in Table 1 of PMID:25299189).
- Class C phenotype: amorphous clumps, no rosettes, with and without RIFs
PMID:30556809
- Clumps re-form rapidly after shear; clumping not dependent on cell division; mutant cells
adhere to wild-type cells
PMID:30556809
PMID:30556809
Mapping (demonstrated)
Rescue (demonstrated, partial penetrance)
Predicted enzymatic family (predicted only)
- Remote structural homology only (HHpred/Phyre2) to polypeptide GalNAc-transferases:
PMID:30556809
- No biochemical assay of any glycosyltransferase activity; no donor or acceptor substrate
identified PMID:30556809.
- Homologs in other choanoflagellates and fungi, none in animals
PMID:30556809
Localization (demonstrated for tagged, overexpressed protein; organelle identity inferred)
- Jumble-mWasabi under the efl promoter in wild-type cells
PMID:30556809
PMID:30556809 ...
"rosetta and other choanoflagellates, Jumble-mWasabi localization corresponds to the location of the Golgi apparatus, for which there is not yet a fluorescent marker in S."
- Caveat: no Golgi marker exists in S. rosetta; assignment is by position relative to EM
ultrastructure. Construct is overexpressed and tagged. Still, the authors state the
conclusion directly PMID:30556809
- The L305P mutant protein is retained in the ER
PMID:30556809
Glycosylation phenotype (demonstrated; mechanism ambiguous)
- 21/22 lectins unchanged; jacalin basal staining lost
PMID:30556809
- Restored by complementation
PMID:30556809
- BUT jacalin Western blots of whole-cell lysates showed no clear difference in banding
between wild type and Jumble ("No clear differences in banding pattern were observed among the S. rosetta strains").
The authors themselves leave open whether the defect is in glycosylation or in trafficking
PMID:30556809
Rosetteless in the mutant (demonstrated)
Models (speculative, from Discussion)
- Regulatory glycosylation of adhesion molecules (e.g. cadherins), or a protective sugar
layer masking sticky surfaces; possible glycosylation of Rosetteless or its partners.
Other literature
- PMID:31896587 (Larson et al. 2020, biophysics of rosettes) summarizes jumble as an ECM
regulator: "a C-type lectin called rosetteless (18) and 2 predicted glycosyltransferases called jumble and couscous (19)".
- PMID:41037400 (Combredet, Ansel & Brunet 2025, Cell Rep; abstract only in cache): CRISPR
knockout of jumble (with rosetteless and couscous) as a positive control
["We inactivated three known S. rosetta multicellular developmental regulators (rosetteless, couscous, and jumble)"].
The KO phenotype details are not in the abstract; I have not seen the full text.
- PMID:25299189 (Levin et al. 2014): original isolation of the Jumble mutant (Table 1).
Annotation decisions (reasoning)
- CC Golgi apparatus (GO:0005794), IDA, PMID:30556809 - propose NEW. Direct imaging of
the fusion protein, organelle assigned by position; authors explicitly conclude Golgi.
Caveat noted.
- MF glycosyltransferase activity (GO:0016757) - NOT proposed as NEW. Evidence is remote
structural similarity only (HHpred/Phyre2), low-confidence CDD, no InterPro signature, no
biochemistry. Captured in core_functions as a hedged prediction instead. A specific
activity (e.g. polypeptide GalNAc-transferase) is certainly not justified.
- BP glycosylation - NOT proposed. GO:0070085 glycosylation and GO:0006486 protein
glycosylation are obsolete (verified QuickGO; 0006486 replaced_by GO:0009101 glycoprotein
biosynthetic process). An IMP to GO:0009101 was considered, but (i) the glycosylated
molecule(s) are unknown (not shown to be glycoprotein), (ii) jacalin Western blots of
lysates showed no difference, so the defect may be spatial (trafficking/secretion) rather
than loss of a glycan, and (iii) the authors themselves leave both possibilities open.
Without the MF, participation cannot be established; left as a question.
- BP rosette development / negative regulation of cell-cell adhesion (GO:0022408) - NOT
proposed. Evidence is necessity (mutant/rescue), with mechanism unknown; the protein is
in the Golgi, not at the adhesion interface. Fails the participation test on current data.
There is also no GO term for choanoflagellate rosette (colony) development.