LysK (Q6Y7T6) — Staphylococcus phage K endolysin — curation notes

Provenance note. No deep-research API keys are available in this container, so
just deep-research was not run and there is no -deep-research-<provider>.md file for
this gene. All of the material below was assembled by hand from the cached publications
in publications/, the UniProt flat file lysK-uniprot.txt, and OLS/QuickGO term
lookups. The GONUTS wiki (gowiki.tamu.edu), which would record the CACAO students'
per-paper evidence, was queried for PMID:19493008 on 2026-10-29 and returned only
"This site is temporarily offline while we perform upgrades" — the same state the
CACAO project page records for batch 1.

Identity and naming

UniProt entry ENLYS_BPPGK / Q6Y7T6, 495 aa, "Endolysin LysK", from Staphylococcus
phage K (NCBITaxon:221915, genus Kayvirus; known host Staphylococcus aureus,
OH NCBI_TaxID=1280). The entry carries no gene name: GN ORFNames=ORF30/ORF32, PhageK_071 {ECO:0000312|EMBL:AHB79986.1}, and GOA therefore prints ORF30/ORF32 in its
SYMBOL column. The literature (O'Flaherty 2005, Becker 2009, Horgan 2009,
Sanz-Gaitero 2014) uniformly calls the gene lysK and the protein LysK, so this review
uses gene_symbol: lysK, matching the folder name. ORF30/ORF32 is an ORF designation,
not a symbol, and the stub's placeholder Q6Y7T6 is an accession.

Domain architecture

Three modules, N→C PMID:19493008:

Two catalytic activities, two bonds

Both catalytic domains were shown active on purified peptidoglycan by MS
PMID:19493008, with the cut sites identified as
PMID:19493008. The structure
paper restates the division of labour
PMID:25064136.

So: amidase domain = EC 3.5.1.28 (amide bond, MurNAc–L-Ala), well served by
GO:0008745. CHAP domain = EC 3.4.22.- , a D-alanyl-glycyl endopeptidase; GO has no
term for that bond specificity, and the nearest honest MF term is GO:0004197 cysteine-type endopeptidase activity, grounded in the papain fold and the triad.

The CHAP domain is the one that matters against live cells
PMID:19493008 whereas PMID:19493008 — hence the paper's title.
Independently confirmed by PMID:19047377.

Metals

The CHAP domain structure (PDB 4CSH, 4CT3, 1.69-1.79 Å) resolves two metals
PMID:25064136
and PMID:25064136. Both
were confirmed by X-ray fluorescence
PMID:25064136. The calcium is functionally required
PMID:25064136, the zinc is not pinned down
PMID:25064136. UniProt additionally places
a Zn site in the amidase domain at 214/324/332 by similarity to D6QY02, with the note
"Zn(2+) is required for amidase activity" (by similarity to P00806). So calcium ion
binding is experimentally established; zinc ion binding is well supported but inferential.

Activity against live staphylococci (host range)

PMID:16199588 — a 54 kDa
lytic zone on a zymogram containing heat-killed MRSA DPC5645 (Fig. 1A, lane 5), matching
the predicted mass. Kill curve: PMID:16199588. Genus-restricted:
PMID:16199588. Confirmed for untreated cells in
PMID:19493008.

Note the lysK transcript timing: PMID:16199588 — late expression, consistent with a lysis-timing role,
but "data not shown".

Direction of the interaction: what the terms must respect

LysK is encoded by the phage and acts on its host's peptidoglycan. Two annotation
shapes in GOA get this backwards, both of them electronic:

  1. GO:0009253 peptidoglycan catabolic process (IEA, InterPro2GO from IPR002502 /
    IPR036505). The term sits entirely in the metabolic branch — peptidoglycan metabolic
    process → aminoglycan → macromolecule catabolic process, all under GO:0009987 cellular process (checked in OLS). It describes an organism turning over its own
    peptidoglycan. A phage has none. The host-directed counterpart is GO:0052009 symbiont-mediated disruption of host cell wall, under the interspecies-interaction
    branch. This is the same failure the CACAO batch-1 review found on PRD1 endolysin P15
    with GO:0044277 cell wall disassembly, arriving here by a different route
    (InterPro2GO rather than a student), which suggests the own-process/host-directed
    confusion is a property of the ontology's two parallel branches rather than of any one
    annotation pipeline.
  2. GO:0042742 defense response to bacterium (IEA from KW-0081 Antimicrobial). The
    phage attacks the bacterium; it does not defend itself against one. Identical to the
    PRD1 P15 case, which was reported upstream as a keyword-mapping defect.

The correct in-vivo frame is the lytic end of the infection cycle:
GO:0001897 symbiont-mediated cytolysis of host cell, and below it
GO:0044659 viral release from host cell by cytolysis. UniProt curates the release role
as an inference — "Endolysin that degrades host peptidoglycans and participates in the
sequential events which lead to the programmed host cell lysis releasing the mature
viral particles (Probable)" — and the entry carries both the Host cell lysis by virus
and Viral release from host cell keywords, yet no GO:0044659 row reached GOA. Neither
cached paper assays progeny release during infection (both apply purified or
lysate-borne enzyme from the outside), so GO:0001897 is the better-supported of the two
and GO:0044659 is added here as an IC inference rather than claimed as demonstrated.

CACAO rows (ASSIGNED BY = CACAO in lysK-goa.tsv)

Term Evid PMID Verdict
GO:0008745 N-acetylmuramoyl-L-alanine amidase activity IDA 19493008 matches the MS assay exactly
GO:0008233 peptidase activity IMP 19493008 right claim, too general — CHAP is a cysteine endopeptidase
GO:0001897 symbiont-mediated cytolysis of host cell IDA 16199588 correct, and correctly on the interspecies branch
GO:0001897 symbiont-mediated cytolysis of host cell IDA 19493008 same, from the live-cell lysis assays

Notable: the students chose the host-directed GO:0001897 where the automated pipelines
chose own-process and host-defence terms. On this gene the CACAO rows are the better
annotations; the errors are all electronic.

PMID:19493008 is abstract-only in the cache (full_text_available: false), but its
abstract states the amidase assay, the cut sites, the CHAP-domain truncation results and
the live-cell lysis directly, so no row here rests on evidence that could not be seen.
No REMOVE is applied to any experimental row.

Open questions