15

Summary: Combined electronic assertion of the gene's demonstrated catalytic activity, from the GH24 family signatures, the EC number and the HAMAP endolysin rule. It agrees with the experimental result on the purified protein.

Reason: This is the core molecular function. The electronic route is sound - IPR002196 is the glycoside hydrolase family 24 (phage lysozyme) family entry, PF00959 matches the catalytic domain, and the entry carries EC 3.2.1.17 - and it is independently confirmed by experiment: PMID:3469652 purified the product of gene 15 to near homogeneity, confirmed its identity by N-terminal sequencing, and found lysozyme activity comparable to the T4 gene e and P22 gene 19 lysozymes. The HAMAP CAUTION that the protein lacks the conserved Asp active site is a sequence observation about one of the two canonical catalytic residues; it does not put the activity in doubt, because the activity was assayed directly on the purified enzyme, and Glu15 is annotated as the proton donor/acceptor.

Summary: CACAO row (2012). Right gene, right paper, right term: PMID:3469652 cloned gene 15, purified its product, confirmed identity by N-terminal sequencing and measured lysozyme activity. The evidence code is the one thing wrong - the paper contains no mutant, so this is IDA, not IMP.

Reason: CACAO assignment, experimental, and what the paper shows for this gene is exactly this term. The whole paper is about the product of gene 15: a DNA fragment encoding the entire gene was expressed in Escherichia coli, the 26 kDa product purified, its identity established by N-terminal amino acid sequencing, and its lysozyme activity compared with two other phage lysozymes. Nothing here rests on a phenotype: there is no sus15 mutant in the phi29 collection, and the only suppressor-sensitive mutant in this part of the genome is sus14(1241) in the neighbouring holin gene (PMID:3027653). So the student chose the term and the paper correctly, and used an experimental code that misdescribes a purified-protein assay; the mismatch is recorded as a source defect rather than an action change, following the batch-2 treatment of the SWP7 and T5 D15 rows. The cached record is abstract-only (full_text_available: false) and no figure-level check was possible: PMC304338 is not in the open-access subset, and the GONUTS wiki page that would hold the student's reasoning now returns HTTP 410 Gone with "GONUTS, the GO wiki, has been retired" (queried 2026-11-01), so the project's standing fallback for abstract-only CACAO rows is permanently unavailable. No REMOVE is entertained for an experimental row whose full text has not been read.

Summary: Root molecular-function term projected from keyword KW-0081 (Bacteriolytic enzyme). True and uninformative: the specific activity, GO:0003796, is asserted twice in this review.

Reason: GO:0003824 is the root of the catalytic branch and an ancestor of GO:0003796 lysozyme activity, which this review accepts on both an electronic and an experimental row. The term is being judged here, not the keyword: KW-0081 (Bacteriolytic enzyme, "Enzyme, e.g. lysozyme or endopeptidase, essential for lysis of bacterial cell walls", checked live at rest.uniprot.org/keywords) is correct for this protein. What the mapping adds is nothing - it restates that an enzyme is an enzyme - and the same keyword's useful content reaches GOA through GO:0031640 and, wrongly, through GO:0042742 below.

Summary: Own-process metabolic term for a host-directed activity. The chemistry is right and the direction of the interaction is wrong: GO:0009253 describes an organism turning over its own peptidoglycan, and phi29 has none.

Reason: Ancestry checked in QuickGO: GO:0009253 sits wholly inside the metabolic branch - peptidoglycan metabolic process, aminoglycan catabolic process, macromolecule catabolic process, all under GO:0009987 cellular process - with no interspecies parent anywhere above it. As written the row asserts that Bacillus phage phi29 catabolises peptidoglycan of its own, which is false; gp15 hydrolyses the sacculus of its Bacillus subtilis host. GO has the matching host-directed term, GO:0052009 symbiont-mediated disruption of host cell wall, under GO:0098933 symbiont-mediated disruption of host cell envelope and GO:0044419, so replacing the term keeps the substance and puts it on the right branch. This is the third instance of the shape the CACAO project found on PRD1 endolysin P15 (GO:0044277, batch 1) and phage K LysK (GO:0009253 from InterPro2GO, batch 2), and the sharpest of the three, because here one of the two sources is HAMAP rule MF_04110 / UniRule UR000433564, whose name is ENDOLYSIN_T4 and whose scope is phage endolysins exclusively: every protein the rule fires on acts on another organism's wall, so the own-process term is wrong for all of them, not merely for a cross-species minority of a shared domain.

Proposed replacements: symbiont-mediated disruption of host cell wall

Summary: Root hydrolase term from KW-0378. True, and subsumed by the glycosidase activity asserted elsewhere in this review.

Reason: GO:0016787 is an ancestor of GO:0003796 lysozyme activity via GO:0016798 hydrolase activity acting on glycosyl bonds. The claim is correct - gp15 hydrolyses a glycosidic bond - but uninformative once the specific activity is asserted. Judging the term, not the keyword, which is accurate for this entry.

Summary: Glycosidase class term from KW-0326, the immediate ancestor neighbourhood of the accepted lysozyme activity.

Reason: GO:0016798 is an ancestor of GO:0003796, which this review accepts on two rows, so the term adds nothing beyond what is already asserted. It is nonetheless correct: the enzyme is a glycosidase of family GH24, and the record's catalytic activity line gives the bond it cleaves. Judging the term, not the keyword mapping, which is accurate.

Summary: Second own-process term for the same host-directed activity, from InterPro2GO on the GH24 family entry. GO:0016998 is a cell-wall-organization term describing the annotated organism's own wall, which a phage does not have.

Reason: Ancestry checked in QuickGO: GO:0016998 sits under GO:0044036 cell wall macromolecule metabolic process and GO:0071554 cell wall organization or biogenesis, all within GO:0009987 cellular process, with no interspecies parent. It therefore makes the same false assertion as GO:0009253 above, one step more generally, and takes the same host-directed replacement, GO:0052009 symbiont-mediated disruption of host cell wall. This term had not been seen in this shape before: the PRD1 instance was GO:0044277 cell wall disassembly and the LysK instance was GO:0009253, so GO:0016998 is a third own-process term reached by the same route. Note that this is not a case of the "correct but general" rule - the problem is not that the term is broad but that it is on the wrong branch, so the replacement is not a descendant of what it replaces.

Proposed replacements: symbiont-mediated disruption of host cell wall

Summary: The only cellular-component row, from the curated subcellular-location keyword. Correct: the endolysin is synthesised in and retained in the host cytoplasm, and reaches the murein only once the holin has breached the membrane.

Reason: gp15 has no signal peptide and no transmembrane segment in the record, and the curated location note states that it is cytoplasmic and depends on the holin to reach the periplasmic space - which is precisely the holin-endolysin mechanism demonstrated for this pair in Escherichia coli, where protein 14 alone kills without lysing and lysis follows only when the phi29 lysozyme is co-expressed. Host cell cytoplasm is the right compartment for the protein even though it is not where the catalysis happens; GO has no host periplasmic space term (checked by QuickGO search), so the site of action cannot be annotated and is noted in suggested_questions instead. Kept as core rather than non-core because it is the compartment the enzyme occupies for the whole of its functional life until the moment of lysis.

Summary: Keyword-derived host-killing term. True, and an ancestor of the specific term GO:0044659 that this review accepts on two rows.

Reason: Ancestry checked in QuickGO: GO:0044659 viral release from host cell by cytolysis descends from GO:0031640 through GO:0001897 symbiont-mediated cytolysis of host cell, GO:0001907 and GO:0051715 cytolysis in another organism. The broad term is correct but loses both of the things that matter - that the killing is lysis of the cell wall, and that the organism killed is this phage's own host rather than an incidental target. Judging the term, not the row's evidence; KW-0081 (Bacteriolytic enzyme) and KW-0204 (Cytolysis) are both appropriate for this entry, and this is the one GO:0031640 mapping of KW-0081 that is sound, in contrast to GO:0042742 below.

Summary: KW-0081 (Bacteriolytic enzyme) mapped to a host-defence process term. The direction of the interspecies interaction is inverted: phi29 attacks the bacterium, it does not defend itself against one.

Reason: GO:0042742 denotes a reaction mounted to protect the annotated organism from a bacterium. gp15 degrades the peptidoglycan of its own host in order to release progeny virions; there is no defence and no bacterium being resisted by phage phi29. This is a demonstrably wrong electronic keyword-to-GO mapping, which the guidelines list as legitimate ground for REMOVE, and the real content of the keyword is already carried correctly by GO:0031640 and GO:0044659. The identical row was removed on PRD1 endolysin P15 in CACAO batch 1 and on phage K LysK in batch 2 and reported upstream both times; this is a third instance, so the defect is systematic across phage-encoded lytic enzymes. One correction to those two reports: both describe KW-0081 as "Antimicrobial", but the live UniProt keyword record gives KW-0081 as Bacteriolytic enzyme, a child of KW-0929 Antimicrobial. That makes the mapping worse rather than better - the keyword means specifically a wall-lysing enzyme, a class for which host defence is never the right framing, so the defect cannot be excused as a broad keyword that also covers antimicrobial peptides.

Summary: The core biological process, asserted by the HAMAP endolysin rule. gp15 performs the peptidoglycan hydrolysis that ruptures the host cell and releases progeny virions.

Reason: Unlike the two own-process terms from the same rule, this one is on the correct branch: GO:0044659 descends from GO:0019076 viral release from host cell and from GO:0001897 symbiont-mediated cytolysis of host cell under GO:0044419 (checked in QuickGO). The claim is also participation rather than mere necessity - gp15 itself catalyses the murein hydrolysis that is the lytic step - and it is supported experimentally for this gene pair: co-expression of phi29 protein 14 with the phi29 lysozyme lyses Escherichia coli, whereas protein 14 alone kills without lysis, which is the holin-endolysin division of labour the term describes. This is the row that carries the claim the misattributed CACAO row below also states, so the correction of that row's provenance does not cost the entry a true annotation.

Summary: CACAO row (2025) on the 1976 phi29 sus-mutant survey. The term is true of gp15 and is carried independently by the UniRule row above, but the mutant phenotype the paper reports belongs to the adjacent gene 14, the holin, and the qualifier is wrong for a protein that executes the lytic step.

Reason: Kept, and the curator deferred to, because the claim is true and independently established: gp15 is the endolysin whose hydrolysis of the host sacculus releases progeny virions, which the GO_REF:0000104 row above asserts on sound grounds. What is wrong is the reference-to-gene pairing, recorded as a source defect. PMID:820555 reports one lysis cistron, unnamed in the abstract: "A cistron, probably involved in the final lysis of the infected bacteria, has been found", whose mutants show delayed lysis with a normal or better burst after artificial lysis. Five lines place that cistron at gene 14 rather than gene 15. (i) The only suppressor-sensitive mutant in this region of the phi29 collection is sus14(1241) (PMID:3027653); there is no sus15. (ii) Gene 14's product is the one required for efficient lysis of Bacillus subtilis (PMID:8432697). (iii) Gene 14 encodes the holin and gene 15 the lysozyme (PMID:7831803), and a holin mutant gives exactly the delayed-lysis-with-normal-burst phenotype the 1976 abstract describes, because the endolysin accumulates in the cytoplasm but cannot reach the murein. (iv) gp15 was characterised eleven years later by cloning and purification (PMID:3469652), not by any mutant, so a 1976 mutant-phenotype paper cannot ground an IMP on it. (v) Decisively, the same curator made a second GO:0044659 IMP row from the same sentence on gene 4, the late-genes transcription activator p4 (UniProtKB:P03682, 2025-02-20, in the project's CACAO snapshot): one cistron in the paper, two genes annotated, and the actual lysis cistron carries no row at all. The choice between those two rows is therefore a false one - the answer is neither, because the cistron is gene 14 - and the later phi29 literature attributes the 1976 abstract's exact phenotype, delayed lysis with a normal or larger burst after artificial lysis, to the alleles sus14(1241) and sus14(1242) by name. This is the adjacent-locus shape that batch 3's first screen was built to test, and it behaves as it did on phage T4 gp5.4 - the term happens to be true of the entry it landed on, so an inventory of terms scores the row as a success while its stated evidence is about a different protein. Removal is not the remedy: the full text could not be read (abstract-only cache, Eur J Biochem 1976 is not in PMC), and the GONUTS page that would hold the student's reasoning now returns HTTP 410 Gone with "GONUTS, the GO wiki, has been retired" (queried 2026-11-01).