Bacillus phage phi29 gene 17 / protein p17 (UniProtKB:P03686)

Curation notes for the CACAO review project (batch 3, screen 1: adjacent-locus /
one-paper-many-genes).

Method note. No deep-research provider API keys exist in this container, so
just deep-research was not run and no -deep-research-<provider>.md file was
created. Everything below comes from the cached publications/PMID_*.md records,
NCBI E-utilities abstracts (papers then cached with just fetch-pmid), the
UniProt flat file, QuickGO and OLS.

GONUTS is now retired, not merely offline. https://gowiki.tamu.edu/ returns
HTTP 410 with the page "GONUTS, the GO wiki, has been retired ... The GO wiki that
was at gowiki.tamu.edu is no longer online." The Wayback Machine has no snapshot of
BPPH2:GP17 or of the PMID:820555 paper page. The student-evidence check the
CACAO project plan relies on is therefore permanently unavailable, and the
abstract-only CACAO rows have to be judged under curator deference for good.

What p17 is

A 166-residue early protein encoded by the first gene of phi29's right-side early
operon (the C2-promoter operon, which also contains gene 16.7), expressed very early
after infection and repressed later by p6 PMID:17526715. It is a homodimer with a
putative coiled-coil region and interacts with the viral architectural/origin-binding
protein p6 PMID:12480935. It is conserved across all phi29-like phages and was used as a
phylogenetic marker, described there as "an accessory component of bacteriophage DNA
replicating machinery" PMID:9929388.

p17 is not itself a membrane protein. Within the right-side early operon only
gene 16.7 encodes one PMID:17526715, and the
membrane-associated early replication protein of phi29 is p1, not p17
PMID:15576790. p17 participates in processes that are
membrane-associated (in vivo phi29 replication, the pull step of ejection) without
being a membrane component itself, and UniProt P03686 carries no membrane keyword,
transmembrane feature or subcellular-location block. Phrasings that call p17
"membrane-associated" conflate it with p1 or p16.7.

Role 1: viral DNA genome replication

The oldest evidence is the 1976 survey of phi29 gene products and functions, a bulk
nonsense-mutant screen: PMID:820555. The abstract does not name the six, but the same laboratory named them seven
years later while describing the mutant extracts it used: PMID:6402761. With gene 1, which also carries a DNA-replication
annotation from the 1976 paper, that is the set of six. Gene 17 is therefore genuinely
one of the DNA-synthesis cistrons of PMID:820555, and the attribution is independent of
anything in the abstract. Later work restates it as settled: PMID:10650217 in DNA
replication; and PMID:9858710 viral DNA replication in vivo.

The same 1983 in vitro work shows the role is not in the initiation reaction itself:
the p3-dAMP initiation complex forms without p17 PMID:6402761.

The direct evidence is Crucitti et al. 1998, which cloned gene 17, purified p17 and
showed a stimulatory effect in vitro: PMID:9858710, specifically
"starting with a low amount of input DNA". In vivo the requirement is conditional on
multiplicity of infection — PMID:9858710 dispensable at high moi — with the proposal that p17 recruits
limiting initiation factors at the origins early in infection and becomes dispensable
once the other replication proteins reach optimal concentration.

The mechanism proposed from the 2003 interaction study is that p17 acts through p6:
PMID:12480935 favouring initiation of replication and negatively modulating its own
transcription. In vivo this is supported by chromatin immunoprecipitation in
sus17-infected cells, where p6 binding is reduced along the whole genome; the authors'
own reading is topological and therefore indirect PMID:15576790.

Importantly, the replication role survives controlling for the ejection defect below:
PMID:17526715.

Role 2: the pull step of phi29 genome ejection into the host

Not represented in GOA at all, and the better-documented of p17's two roles. phi29
injects its linear genome right-to-left in two steps; the first ~65% is pushed in by
capsid pressure, and the remainder is actively pulled in by machinery that includes p17:
PMID:15066038 molecular machinery that pulls the remaining DNA inside the cell. The pull step
needs an electrochemical proton gradient, and p17 is where the energy enters:
PMID:16517096,
in a motor "that comprises at least viral protein p17, encoded by the right early
operon". X-ChIP measurements of left-end internalisation in non-p6-producing cells
confirm it and rank it against p16.7: PMID:17526715, and the absence of p16.7 "affects the pull step less drastically than the
absence of protein p17".

This is participation, not mere necessity: p17 is a component of the pulling machinery
and the energy-transduction step is attributed to it, with azide having no effect in its
absence.

Reading of the four GOA rows

GO:0006260 is not an ancestor of GO:0039693 (checked against QuickGO's is_a/part_of
ancestor set for GO:0039693: GO:0006259, GO:0071897, GO:0019079, GO:0141187 and up, no
GO:0006260), so the keyword row is a lateral, less informative statement of the same
claim rather than a strict parent restatement. MODIFY to GO:0039693, which the gene
already carries three times.

The predicted failure shape (screen 1), tested

Prediction: PMID:820555 is a multi-gene survey cited across most of phi29's numbered
genes, so gene 17's CACAO row might be the T4 gp5.4 shape — one paper's finding
misattributed to a neighbouring numerically-named gene.

Absent. The paper's "six genes" really do include gene 17, named by the same
laboratory in PMID:6402761, and the claim is confirmed by three later papers. The
abstract is the only thing that is silent; the attribution is not a guess.

What the full CACAO row set for phi29 does show (QuickGO, assignedBy=CACAO,
taxonId=2884424, 24 rows) is a single Feb–Apr 2025 campaign that read PMID:820555
once and distributed it across the phage:

• GO:0039693 IMP on genes 1, 2, 3, 6 and 17 — the DNA-synthesis cistrons, correct;
• GO:0044659 viral release from host cell by cytolysis IMP on genes 4 and 15, from a
  paper whose abstract reports one such cistron ("A cistron, probably involved in
  the final lysis of the infected bacteria, has been found"). Gene 15 is the endolysin;
  gene 4 is the late-gene transcription activator. That pair, not gene 17, is where this
  paper's one-to-many distribution looks over-extended; it belongs to the gene 4 review.
• Every 2025-dated phi29 row (10 of them) carries
  acts_upstream_of_or_within_positive_effect, while the same organism's 2012–2017 rows
  use involved_in/enables/part_of. So screen 3's qualifier shape is present here
  as a campaign-wide habit — but on this gene it is the defensible choice, not a
  defect: the evidence is a bulk nonsense-mutant screen with no demonstration of direct
  participation, and the modern literature shows a large part of the sus17 replication
  defect is indirect (failed genome internalisation, altered DNA topology, reduced p6
  binding). The conservative qualifier is more honest here than involved_in would be.