No deep-research API keys were available in this container, so just deep-research
was not run and there is no -deep-research-<provider>.md file for this gene.
Literature research was done directly: the three cached publications already in
publications/ were read, PubMed was searched through the PubMed MCP tools
(IpaH4.5 Shigella, 11 hits, all inspected by title/abstract), and four further
papers were cached with just fetch-pmid (PMID:30414351, PMID:33117724,
PMID:34296983, PMID:1791758).
Of the papers cited here only PMID:33117724 has full_text_available: true; the
others are abstract-only. The GONUTS wiki could not be consulted: the live site
(gowiki.tamu.edu) returns a "site is temporarily offline while we perform
upgrades" notice, the Wayback availability API has no snapshot for
PMID:23083102 or PMID:26700764, and the one archived page that does exist
(SHIFL:IPA4, snapshot 2021-10-18) is on web.archive.org, which this container
blocks by egress policy. So the CACAO students' own evidence notes were not
readable, and the curator-deference rule was applied throughout: no experimental
CACAO row was removed.
ipaH4.5 is one of a family of ~9 ipaH genes in S. flexneri, five on the
large virulence plasmid (pWR100/pCP301, where P18009 is CP0079) and several
chromosomal copies. The paralogs share a conserved C-terminal catalytic region
and differ in their N-terminal leucine-rich-repeat substrate-recognition region:
This is the structural basis for treating substrate claims as paralog-specific
(IpaH9.8→NEMO/ABIN-1/GBPs, IpaH7.8→glomulin/NLRP1B/GBPs, IpaH4.5→p65, TBK1,
RPN13, NLRP3) while treating secretion and catalytic mechanism claims as
family-level. UniProt's own feature table matches: LRR 1–10 (res. 63–270),
"Interaction with target proteins" 1–284, NEL catalytic domain 295–574.
Because the family shares antigenic determinants and most anti-IpaH antisera are
not paralog-specific, any immuno-detection claim from an "IpaH" blot has to be
read as family-level. That is exactly the situation for the extracellular
region IDA (see below).
UniProt assigns the NEL (Novel E3 Ligase) domain to residues 295–574 with an
active site at residue 379:
[file:SHIFL/ipaH4.5/ipaH4.5-uniprot.txt "FT ACT_SITE 379 FT /note=\"Glycyl thioester intermediate\""]
Residue 379 of P18009 is a cysteine (checked directly against the SQ block:
…DATESCEDRVAL…, positions 374–385). A "glycyl thioester intermediate" on a
cysteine is the HECT-like mechanism: the ubiquitin C-terminal glycine is first
transferred onto the enzyme's own catalytic Cys before being passed to the
substrate lysine.
This is independently confirmed for IpaH4.5 itself. The NLRP3 paper uses a
C379A point mutant as its catalytic-dead control throughout:
Noted source defect. UniProt P18009 carries
AltName: Full=Probable RING-type E3 ubiquitin transferase ipaH4.5
{ECO:0000305}. That is inconsistent with the same record's NEL domain and
Cys379 glycyl-thioester active site, and with the C379A data above: NEL-domain
IpaH/SspH ligases are not RING-type (RINGs are scaffolds with no catalytic
residue and no thioester intermediate). Recorded as a source_defects entry
(resource: UNIPROT) on the GO:0061630 row.
| Substrate | Chain | Outcome | Reference |
|---|---|---|---|
| NF-κB p65 (RELA) | polyUb | dampened NF-κB transcriptional output, less MIP-2/TNF-α | PMID:23083102 |
| TBK1 | K48 | proteasomal degradation, IRF3 not activated | PMID:26700764 |
| RPN13 (ADRM1) | polyUb | 19S RP dysfunction, less MHC-I cross-presentation | PMID:30414351 |
| NLRP3 | K63 | NLRP3 stabilized, inflammasome/pyroptosis up | PMID:33117724 |
| Rab31 (reported as GAP, not ligase) | — | CD-MPR mistrafficking, lysosome attenuated | PMID:34296983 |
| ELMO2 | polyUb | RAC activation disrupted (preprint, not cited in review) | PMID:42538988 |
Note what this paper does not contain: any assay of host protein synthesis.
Everything on the output side is transcriptional (an NF-κB-response-element
luciferase reporter) or cytokine protein levels downstream of that. This matters
for the GO:0039604 symbiont-mediated suppression of host translation CACAO row —
see the CACAO audit below.
A second, independent ubiquitin-proteasome substrate; corroborates
GO:0043161 proteasome-mediated ubiquitin-dependent protein catabolic process
beyond the TBK1 paper.
This is the mechanistically cleanest IpaH4.5 paper available (substrate
interaction, chain-linkage mapping, C379A catalytic control, Δ mutant plus
complementation, and a ΔipaH4.5/ipaH4.5(C379A) strain). It is also the only
positive-sign activity: K63 chains stabilize NLRP3 rather than destroying it,
so the same ligase both suppresses NF-κB/IRF3 signalling and promotes
inflammasome signalling. One NEW row (GO:0141087 positive regulation of
inflammasome-mediated signaling pathway, IDA) is proposed from it; nothing in
the existing GOA set covers this arm.
A second molecular function (GTPase activator activity) in a protein whose only
recognized catalytic module is the NEL domain would be a significant claim.
It rests on one abstract-only paper from one group, UniProt records no TBC-like
motif in P18009, and PROSITE/Pfam match only LRR_TTSS and NEL. Not asserted as
a NEW annotation; raised instead as a suggested_questions entry and a
suggested_experiments entry.
A 2026 bioRxiv preprint identifies ELMO2 as an IpaH4.5 substrate with strict
paralog selectivity (no activity on ELMO1/ELMO3), mapped to the ELMO2 Armadillo
repeat region. Not cited in the review YAML because it is an un-peer-reviewed
preprint, but worth revisiting on publication — it would be the clearest
demonstration of LRR-encoded substrate specificity for this paralog.
Secretion/location evidence is family-level, not IpaH4.5-specific:
The title and most of the paper are about IpaH(9.8), including the nuclear
import work. UniProt nevertheless cites this paper on the P18009 entry under
SUBCELLULAR LOCATION, AND SECRETION VIA TYPE III SECRETION SYSTEM (reference
[6]), i.e. the UniProt curator read the Congo-red secretion data as covering the
IpaH proteins generically, which the quoted sentences support. So the
GO:0005576 extracellular region IDA is kept, but as non-core: the medium is
where a T3SS effector is caught in transit under artificial induction, not where
it works. Every known IpaH4.5 substrate (p65, TBK1, RPN13, NLRP3, Rab31) is a
host cytosolic protein, which is what makes GO:0030430 host cell cytoplasm the
functionally relevant compartment. Note that no paper reports nuclear IpaH4.5;
the nuclear-transport result belongs to IpaH9.8 and must not be propagated here.
ASSIGNED BY = CACAO)| GO term | Code | PMID | What the paper shows for ipaH4.5 | Verdict |
|---|---|---|---|---|
| GO:0000209 protein polyubiquitination | IDA | 23083102 | in vitro ligase activity "towards ubiquitin and p65"; K48 chains on TBK1 and K63 chains on NLRP3 in two later papers | matches — ACCEPT |
| GO:0051865 protein autoubiquitination | IDA | 23083102 | abstract reports activity "towards ubiquitin and p65", not towards itself; full text and GONUTS notes unreadable | plausible family property, peripheral — KEEP_AS_NON_CORE |
| GO:0039604 symbiont-mediated suppression of host translation | IDA | 23083102 | no protein-synthesis assay anywhere; the readout is an NF-κB-response-element luciferase reporter plus cytokine levels | wrong term — MODIFY → GO:0085034 |
| GO:0006511 ubiquitin-dependent protein catabolic process | IMP | 26700764 | "polyubiquitylated TBK1 undergoes proteasome-dependent degradation"; RPN13 likewise via the UPS | right but under-specific — MODIFY → GO:0043161 |
| GO:0052170 symbiont-mediated suppression of host innate immune response | IMP | 26700764 | TBK1 degraded, IRF3 phosphorylation/nuclear import/activation lost, host antibacterial response dampened | matches — ACCEPT |
| GO:0000209 / GO:0051865 / GO:0039604 all key to the same paper | — | — | three BP rows from one abstract-only paper | see individual rows |
The notable failure is GO:0039604. It is not a phenotype-read-as-process and
not a wrong-paper case; it is a wrong-mechanism substitution. The paper
repeatedly says "transcriptional activity of NF-κB … inhibited" and "inhibiting
the NF-κB signalling pathway", and the student (or the curation interface's term
search) landed on the translation sibling of the host-gene-expression branch.
Because the aspect (symbiont suppression of host gene expression) is right and
only the step is wrong, MODIFY to GO:0085034 symbiont-mediated suppression of
host NF-kappaB cascade is the correct action rather than REMOVE — the claim is
not false, it is misfiled, and GO:0085034 is precisely the mechanism the paper
demonstrates. For the CACAO failure taxonomy this is a fourth pattern worth
naming: right pathway, wrong molecular step — a term chosen from the correct
neighbourhood of the ontology but describing a different biochemical process
(translation vs transcription/signalling) than the one assayed. It is invisible
to every mechanical check (the PMID resolves, the paper is about the gene, the
evidence code fits the assay) and can only be caught by reading the term
definition against the figures.
Two of the six rows are also slightly generous with the IDA code:
GO:0039604/GO:0051865/GO:0000209 are all keyed to one paper whose relevant
panels we cannot see. None of that justifies removal under the curator-deference
rule, and the two IMP rows from PMID:26700764 are both well matched to the
abstract's own statements.
GO:0004842 ubiquitin-protein transferase activity (IEA, InterPro NEL) andGO:0061630 ubiquitin protein ligase activity (IEA, EC:2.3.2.27) are the sameGO:0016567 protein ubiquitination (IEA, InterPro NEL) is modified toGO:0000209, which the gene already carries.GO:0035821 modulation of process of another organism (IEA, ARBA) is anGO:0052170 and GO:0085034.GO:0030430 host cell cytoplasm (IEA, UniProt SubCell) is accepted as the