rgn (M9PFV8): evidence and ProtNLM claim review

Regeneration is a secretory-pathway C-type-lectin-domain protein implicated in imaginal-disc regeneration and intestinal epithelial repair. Genetic perturbation affects blastema formation, intestinal progenitor signaling and regenerative growth, while its ligand specificity and direct subcellular localization remain unresolved.

Exact input: M9PFV8, 808 residues. The accession was fetched explicitly with the gene-directory alias; no canonical-sequence substitution is made.

Raw emitted predictions: rgn-predictions-source.json. Source features: rgn-uniprot.txt, with an exact extraction in rgn-sequence-evidence.json.

Sequence and domain evidence

These are sequence/domain observations or explicitly named feature predictions, not measurements of biological function. ARBA assertions and ProtNLM-derived UniProt names are not counted as validation.

DR   InterPro; IPR001304; C-type_lectin-like.
DR   InterPro; IPR016186; C-type_lectin-like/link_sf.
DR   InterPro; IPR016187; CTDL_fold.
FT   SIGNAL          1..23
FT                   /evidence="ECO:0000256|SAM:SignalP"
FT   DOMAIN          70..199
FT                   /note="C-type lectin"
FT                   /evidence="ECO:0000259|PROSITE:PS50041"

ProtNLM claims

The snapshot emits names and location/keyword statements, with no GO or EC prediction for this target. Each actual statement is assessed below; no GO term has been substituted for it. Categories follow the function-prediction rubric, with nonspecific “uncharacterized” names marked UNC because they contain no testable function. CNN records an independently supported existing annotation; it does not assert a particular training-set composition.

Kind	Verbatim emitted statement	Assessment	Evidence and limitation
Name	C-type lectin domain-containing protein	CNN	PROSITE identifies a C-type lectin-like domain at residues 70-199; PMID:18485344 independently describes the signal sequence and CTLD in Rgn. The domain claim is supported without assigning carbohydrate specificity.
Location	Secreted (SL-0243)	UNC	The signal peptide and lectin domain make secretion plausible, and PMID:18485344 independently discusses this architecture. However, the exact long isoform has curated trans-Golgi/vesicle assertions and no direct secretion assay; secretory-pathway entry does not establish release as a soluble extracellular protein.

Literature evidence

PMID:18485344(https://pubmed.ncbi.nlm.nih.gov/18485344/): “We identified three genes, regeneration (rgn), augmenter of liver regeneration (alr) and Matrix metalloproteinase-1 (Mmp1) expressed specifically in blastema cells during disc regeneration.”
PMID:41935796(https://pubmed.ncbi.nlm.nih.gov/41935796/): 2026 abstract-only evidence that Rgn deficiency in gut progenitors disrupts Notch signaling, insulin/TOR-linked redox control and intestinal stem-cell maintenance during epithelial repair.
PMID:42486707(https://pubmed.ncbi.nlm.nih.gov/42486707/): bibliographic cached corrigendum to PMID:41935796; the cache does not expose correction details.

Annotation decisions

PAINT provenance: just fetch-panther-paint PTHR23211 returned exactly one annotated node, PTN000575073, with two IBD rows for mammalian TGN38/TGOLN trans-Golgi localization.

GO:0005802 trans-Golgi network (IBA): UNDECIDED. PANTHER:PTN000575073 propagates mammalian TGN38/TGOLN trans-Golgi localization within PTHR23211:SF0, which also contains M9PFV8; same-subfamily propagation and unresolved direct Rgn localization argue against a confident REMOVE even though the 808-residue fly CTLD protein has no target evidence for TGN38-like residence.
GO:0030140 trans-Golgi network transport vesicle (IBA): UNDECIDED. The same PANTHER:PTN000575073 node propagates mammalian TGN38/TGOLN transport-vesicle residence, but Rgn has no direct evidence for TGN-derived vesicle localization and the SF0 assignment keeps the transfer unresolved rather than demonstrably wrong.
GO:0030246 carbohydrate binding (IEA): UNDECIDED. Several proteins retain this fold without the canonical carbohydrate-binding activity. The exact target has no mapped ligand-binding assay.
GO:0030246 carbohydrate binding (ISS): UNDECIDED. PMID:16475980 directly assays DL1 and discusses Drosophila lectins. It supplies useful family biology but does not establish carbohydrate binding by the long Rgn isoform; target-specific ligand or conserved-binding-site evidence is required.
GO:0042246 tissue regeneration (IEP): ACCEPT. PMID:18485344 reports rgn expression in regenerating blastema cells, matching the tissue-regeneration process established by perturbation evidence in the same paper.
GO:0042246 tissue regeneration (IMP): ACCEPT. The full text of PMID:18485344 maps rgn alleles to CG6014 and reports altered blastema regeneration after genetic perturbation, and the 2026 abstract in PMID:41935796 reports a progenitor-deficiency phenotype in gut epithelial repair. The gene-level process is well supported although the molecular mechanism and isoform contributions remain unresolved.

Research provenance

Genuine external literature research is requested through the repository Falcon wrapper, with perplexity-lite configured as fallback. Provider output is retained separately as rgn-deep-research-<provider>.md; its source leads are checked against the underlying publications and exact sequence record.

2026-10-21 — Full re-review, independent of the ProtNLM2 predictions

Scope: re-examined all six GOA rows, the description and core functions against PMID:18485344 (full text), PMID:41935796 (abstract), the falcon deep research, the UniProt record and a new sequence analysis (rgn-bioinformatics/rgn_architecture.py, RESULTS.md, rgn_architecture_results.json). ProtNLM2 output was not consulted.

Identity

M9PFV8 (808 aa; FlyBase isoforms B/C) is the longer of the two UniProt products of rgn/CG6014 (FBgn0261258). The other, Q9Y102 (800 aa; isoforms A/D), is identical except that it lacks YVQLISNS (M9PFV8 residues 160-167) inside the CTLD. Both entries are unreviewed (TrEMBL), neither is a fragment, and both share PTHR23211:SF0. The benchmark accession is therefore a full-length gene product, not a fragment. No reviewed canonical entry exists, so the same-gene-canonical rule does not apply.
Architecture: signal peptide 1-23 (SignalP), CTLD 70-199 (PROSITE PS50041), long charged disordered tail (MobiDB-lite 244-808 in segments). No transmembrane segment: UniProt has no TRANSMEM feature, and a Kyte-Doolittle scan found none after residue 30.

Evolutionary placement

PANTHER puts Rgn in PTHR23211 "TGN38" (single subfamily SF0, together with TGOLN2/TGN38). PAINT has a single annotated node, PTN000575073 (taxon Eukaryota), with IBDs for trans-Golgi network and TGN transport vesicle, seeded by the vertebrate TGN38 proteins.
Rgn lacks both determinants of TGN38 localization: the TM anchor and the cytoplasmic YQRL motif. Local alignment of Rgn to TGOLN2 (score 95) and rat TGN38 (79.5) falls within the composition-matched shuffle distributions (34.5% and 12% of shuffles score as high or higher) and maps to Rgn's charged disordered region. Rgn's CTLD aligns to REG1A (161), REG3A (101) and lectin-galC1 (81) well above every shuffle; these are all PTHR22803 proteins.
Conclusion: no residue-level support was detected for the TGN38 grouping beyond compositional bias. This is a family-placement concern for PTHR23211 (see the PANTHER review, interpro/panther/PTHR23211/PTHR23211-review.yaml, which marks coherence UNKNOWN). Note that the CTLD similarity to REG proteins also does not make Rgn a REG ortholog. PANTHER separates them, and the McClure et al. "predicted ortholog of mammalian Reg" claim rests on BLAST of the CTLD alone.
The CTLD keeps its cysteines (74, 91, 174, 190, 198) but lacks the EPN/QPD and WND motifs, and UniProt's PROSITE-ProRule caution says conserved residues are missing.

Decisions (changes from the previous version)

GO:0005802 trans-Golgi network (IBA): UNDECIDED -> REMOVE. Propagation is bad for this target because it lacks the TM and sorting determinants, and pairwise similarity to the seeds is not above shuffles.
GO:0030140 TGN transport vesicle (IBA): UNDECIDED -> REMOVE, for the same reason.
GO:0030246 carbohydrate binding (IEA, ISS): UNDECIDED -> MARK_AS_OVER_ANNOTATED. The fold is present but lacks the sugar/Ca2+ motifs. The ISS donor DL1 is in a different PANTHER family, and no ligand has been measured.
GO:0042246 tissue regeneration (IEP, IMP): ACCEPT, unchanged. Supporting quotes were replaced by quotes that state the genetic requirement. GO has no more specific imaginal-disc regeneration term.
I proposed no NEW terms. Notch-signalling and ISC-maintenance terms from the abstract-only gut paper were not added: the mechanism is unclear, and the paper reads as a necessity phenotype rather than participation.
Description and core function were rewritten to state the architecture, the two isoforms, the absence of a TM segment and the unknown molecular activity.

Relative to review next

By PANTHER subfamily, Rgn's only experimentally characterised relatives are TGOLN2 (human O43493) and the rodent TGN38/Tgoln proteins (PTHR23211:SF0). None has a review in the repo. Given the placement concern above, TGOLN2 is a family member, not a demonstrated ortholog.
Modules: none found under modules/ that include rgn, TGOLN2 or imaginal-disc regeneration.