Secreted glycoprotein of the chymotrypsin-like (S1) peptidase fold, synthesised as a precursor with an N-terminal signal peptide and an extended prodomain and released from cells as a processed, N-glycosylated protease-homology domain. Despite retaining the canonical Ser-His-Asp catalytic triad (His175, Asp246, Ser316), PRSS23 lacks the Ile16-Asp194 zymogen activation switch that orders the activation-domain loops and builds the S1 substrate pocket and oxyanion hole of active S1 proteases; the substitutions that dismantle this switch are conserved across vertebrate orthologs. No serine hydrolase activity is detectable from the secreted protein by activity-based probe labelling or from the recombinant domain across a panel of chromogenic peptide substrates, and cellular phenotypes attributed to PRSS23 persist when the putative catalytic serine is mutated, so the protein is best understood as a serine pseudoprotease whose molecular activity has not been established. It is expressed in the ovary, where it marks atretic follicles and peri-ovulatory stroma and theca, and in the developing cardiovascular system, where loss of the zebrafish ortholog blocks endothelial-to-mesenchymal transition at the atrioventricular canal. High PRSS23 expression is associated with poor survival in ovarian cancer, and its depletion reduces proliferation, increases anoikis sensitivity and impairs peritoneal dissemination.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0004252 serine-type endopeptidase activity | IEA GO_REF:0000002 | REMOVE | Summary: Electronic domain-match assignment that direct assay now contradicts; PRSS23 is a secreted serine pseudoprotease. Reason: This annotation has no experimental support anywhere in GOA. It is IEA only, derived from InterPro signatures IPR001254/IPR018114 via GO_REF:0000002, i.e. from the presence of an S1 protease fold and an intact Ser-His-Asp triad. Akhtar et al. 2026 assayed the protein directly and found no activity by two independent methods: a fluorophosphonate activity-based probe failed to label secreted PRSS23 in conditioned media in which multiple other secreted serine hydrolases labelled robustly, and the recombinant protease domain showed no catalytic-serine-dependent turnover across a panel of chromogenic pNA substrates spanning diverse P1 residues, judged against a matched Ser316Ala control. The structural basis is also identified: PRSS23 retains the triad but has lost every component of the Ile16-Asp194 zymogen activation switch that orders the activation-domain loops and creates the S1 pocket and oxyanion hole, and these substitutions are conserved across vertebrate orthologs. Consistently, the pro-tumorigenic phenotypes survive mutation of Ser316, and independent work in gastric cancer found eIF4E binding and progrowth phenotypes retained in catalytic triad mutants. The authors propose reclassification as a serine pseudoprotease. Removing an electronic inference that the only direct measurement refutes is exactly the case CLAUDE.md sanctions; no curator's experimental judgment is being second-guessed here. Supporting Evidence: PMID:41985786 In complementary biochemical assays, PRSS23 showed no detectable serine hydrolase activity in either activity-based probe labeling of conditioned media or chromogenic peptide substrate assays using the recombinant protease domain. PMID:41985786 Taken together with the disrupted Ile16-Asp194 activation switch inferred from sequence analysis, these findings indicate that secreted PRSS23 lacks detectable serine protease activity. PMID:41985786 Multiple probe-reactive secreted hydrolases are detected in CM; however, no probe-reactive band corresponding to PRSS23 is detected at the expected molecular weight in either CM or anti-HA IP fractions. |
| GO:0005576 extracellular region | IEA GO_REF:0000044 | ACCEPT | Summary: Correct, and now directly demonstrated: endogenous epitope tagging shows a processed, glycosylated PRSS23 fragment in conditioned media. Reason: The annotation derives from the UniProt SubCell mapping SL-0243 (Secreted), and endogenous 3xHA tagging in ovarian carcinoma cells confirms it: the protein enters the secretory pathway as a precursor, is processed at the RRKR motif, and is released as a ~36 kDa N-glycosylated protease-homology domain recovered from conditioned media by immunoprecipitation and LC-MS/MS. GO has since obsoleted GO:0005615 extracellular space, the replacement previously proposed here, and merged it into GO:0005576 (replaced_by), so this term is now the correct location term. Supporting Evidence: PMID:41985786 These data indicate that in ovarian cancer cells PRSS23 enters the secretory pathway as a precursor protein that is processed, likely at the canonical RRKR motif, and secreted as a glycosylated protease domain fragment after removal of the prodomain. |
| GO:0006508 proteolysis | IEA GO_REF:0000002 | REMOVE | Summary: The biological-process counterpart of the refuted catalytic annotation; falls with it. Reason: Same provenance as GO:0004252 - IEA from InterPro IPR001254/IPR018114 via GO_REF:0000002, with no experimental support in GOA. There is no evidence that PRSS23 cleaves any substrate: neither activity-based probe labelling of the secreted protein nor a chromogenic substrate panel applied to the recombinant domain detected catalytic-serine-dependent hydrolysis, and the Ile16-Asp194 activation switch required to assemble a competent S1 active site is absent and conserved absent across vertebrates. The paper draws the annotation moral explicitly for this gene. Supporting Evidence: PMID:41985786 Across this substrate panel, we did not observe reproducible substrate turnover above background that depended on the putative catalytic serine; WT and Ser316Ala preparations exhibited similar rates for all substrates tested PMID:41985786 Thus, sequence-based annotation focused solely on the catalytic triad will tend to overestimate catalytic competence and underestimate the true prevalence of serine pseudoproteases, as appears to be the case for PRSS23. |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | REMOVE | Summary: Bare protein binding from a systematic binary interactome screen; carries no functional information. Reason: No more informative molecular function is supportable from this evidence, so the row is removed rather than rewritten. The supporting datum is a single yeast-two-hybrid pair from the HuRI reference interactome (partner UniProtKB:Q9BY50, SEC11C, the catalytic subunit of the signal peptidase complex), which for a signal-peptide-bearing secretory protein is more plausibly a biogenesis contact than a functional partnership. Notably, the one direct partner reported from a focused study - eIF4E, bound through the trypsin-like domain - is not captured here, and even that interaction is hard to reconcile with a secreted protein. Removal does not imply the reported interaction is false. |
| GO:0005788 endoplasmic reticulum lumen | TAS Reactome:R-HSA-8952289 | KEEP AS NON CORE | Summary: Secretory-pathway transit compartment, asserted because PRSS23 is a FAM20C phosphorylation substrate. Reason: The Reactome reaction R-HSA-8952289 ("FAM20C phosphorylates FAM20C substrates") places PRSS23 among secreted proteins phosphorylated by the secretory-pathway kinase FAM20C; UniProt records the resulting Ser109 phosphosite (ECO:0000269|PubMed:26091039). The localisation is therefore real but reflects the route the protein takes to the outside of the cell rather than where it acts. Retained as non-core. |
| GO:0070062 extracellular exosome | HDA PMID:19199708 Proteomic analysis of human parotid gland exosomes by multid... | KEEP AS NON CORE | Summary: Detected in a parotid-gland exosome proteomics dataset; consistent with secretion but not an informative site of action. Reason: A high-throughput mass-spectrometry identification in purified exosomes. It is compatible with the established secretion of PRSS23 and with its recovery from conditioned media, but exosomal co-purification is a common outcome for secreted proteins and does not establish that exosomes are where PRSS23 functions. |
| GO:0005634 nucleus | IDA GO_REF:0000054 | MARK AS OVER ANNOTATED | Summary: Single fusion-protein localisation screen, hard to reconcile with a signal-peptide protein that endogenous tagging shows being secreted. Reason: GO_REF:0000054 is the LIFEdb resource, which annotates localisation of expressed fluorescent fusion proteins in living cells. PRSS23 has a cleaved signal peptide (UniProt SIGNAL 1..19), two N-glycosylation sequons, a FAM20C phosphosite acquired in the secretory pathway, and endogenous 3xHA tagging demonstrates that the mature protein is processed and secreted; nuclear residence is difficult to reconcile with that itinerary, and overexpressed fusions of secretory proteins frequently mislocalise when the signal sequence is occluded or targeting is saturated. Flagged rather than removed: this is an IDA, I have not seen the primary images, and a minor unsecreted intracellular pool cannot be excluded - indeed the reported eIF4E interaction would require one. |
Loading supporting content…
Download this section (compressed HTML)Q: Should GOA record the negative result explicitly, as NOT|enables GO:0004252 serine-type endopeptidase activity with IDA evidence from PMID:41985786, rather than simply dropping the InterPro IEA? A NOT annotation would protect the gene against re-assertion the next time the S1 signature is re-mapped.
Q: What does the secreted PRSS23 protease-homology domain actually bind? The 2026 study frames the options - receptor ligand, non-catalytic cofactor or decoy in a protease cascade, or scaffold - but does not discriminate among them, and no molecular function can be annotated until one is established.
Q: How can a signal-peptide-bearing, glycosylated, secreted protein bind cytosolic eIF4E (PMID:39920289)? Is there a genuine unsecreted intracellular pool of PRSS23, or is that interaction an artefact of overexpression - and does the same question explain the LIFEdb nuclear localisation?
Q: Is the zebrafish EndoMT/Snail requirement (PMID:23213106) also protease-independent? The morpholino experiments predate the activity assays; repeating the rescue with a catalytic-serine mutant would settle whether the developmental role is non-proteolytic too.
Q: Do the paralogous S1 members that share the dismantled activation switch - PRSS35 is named in the 2026 paper, and in most species also substitutes Thr for the catalytic Ser - warrant the same reclassification, and how many other InterPro-driven S1 protease annotations rest on triad conservation alone?
Experiment: Interactome of the secreted, endogenously tagged PRSS23 protease-homology domain: anti-HA immunoprecipitation from conditioned media (the same JHOC-5 system used for the activity-based probe work) followed by quantitative LC-MS/MS against the Ser316Ala line and an untagged control, to identify partners of the fold and discriminate the receptor-ligand, cofactor and scaffold models.
Experiment: Reconstitute the activation switch. Introduce the Ile16 and Asp194 equivalents into recombinant PRSS23 (alone and combined) and re-assay with the activity-based probe and the chromogenic panel. Work on other pseudoproteases predicts that restoring switch residues will not restore activity; a negative result would show the fold has been remodelled beyond the switch, and a positive one would date the loss of catalysis precisely.
Experiment: Separate the protease-homology domain from the prodomain functionally: test whether purified secreted domain added exogenously to PRSS23-knockdown ovarian carcinoma spheroids rescues anoikis resistance and proliferation. This asks whether PRSS23 acts as a diffusible extracellular ligand at all, which is the premise behind proposing to intercept its secretion therapeutically.
Experiment: Surface-binding screen for a receptor: label the recombinant domain and assay binding to mesothelial and ovarian carcinoma cell surfaces, with and without pre-treatment by heparinase and phospholipase C, then deconvolute any specific binding by proximity labelling. This directly tests the ligand-pseudoprotease model exemplified by HGF and MSP.
Experiment: Endogenous catalytic-serine knock-in mouse (Prss23 S-to-A) compared with a Prss23 null, phenotyped for ovarian follicular atresia and cardiac valve morphogenesis. If the catalytic mutant is normal and the null is not, the protease-independent model holds in normal physiology and not only in tumour cells.
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)