ABRAXAS1 (Q6UWZ7) — Evaluation of "microtubule binding" (GO:0008017, IBA) OpenScientist openscientist-autonomous 2 artifacts 2026-07-06T18:19:51.723880

OpenScientist final report (text/html)

OpenScientist final report (application/pdf)

ABRAXAS1 (Q6UWZ7) — Evaluation of "microtubule binding" (GO:0008017, IBA)

Focus: function-assignment hypothesis — ABRAXAS1 directly has microtubule binding (GO:0008017)
Source: genes/human/ABRAXAS1/ABRAXAS1-ai-review.yaml → existing_annotations[2].function_hypothesis
Evidence type under review: IBA (GO_REF:0000033, PAINT phylogenetic inference)

Executive Judgment

Verdict: REFUTED / OVER-ANNOTATED (paralog carry-over).

The IBA "microtubule binding" annotation on ABRAXAS1 is not supported by any primary
experimental evidence for ABRAXAS1 itself. Programmatic provenance tracing shows the
term was propagated by PAINT from a shared PANTHER ancestral node
(PTN001272083) whose experimental support comes from the paralog ABRAXAS2
(ABRO1 / FAM175B, UniProt Q15018) — the scaffold of the cytoplasmic BRISC complex,
which genuinely localizes to the spindle pole and has direct (IDA/IMP) microtubule and
spindle annotations. ABRAXAS1, by contrast, is an exclusively nuclear scaffold of the
BRCA1-A complex whose experimentally supported molecular function is
polyubiquitin-modification-dependent protein binding (GO:0031593, IDA).

Two co-propagated terms from the same node/source — GO:0008608 (spindle-microtubule
attachment to kinetochore) and GO:0090307 (mitotic spindle assembly) — are part of the
same over-annotation cluster and share the fate of GO:0008017.

Most important caveat: This is a decision about direct annotation carry-over, not a
claim that ABRAXAS1 has zero mitotic role. ABRAXAS1 does participate in mitotic G2/M DNA
damage checkpoint signalling (properly annotated IMP/NAS), but that is checkpoint
regulation, not physical microtubule binding, and does not rescue GO:0008017.

Evidence Matrix

#	Citation	Evidence type	Supports/Refutes	Claim tested	Key finding	Context	Confidence / limitations
1	QuickGO provenance (GO_REF:0000033)	computational / database	Refutes (as direct)	Is GO:0008017 experimental for ABRAXAS1?	GO:0008017 on Q6UWZ7 is IBA, assignedBy GO_Central, `withFrom` = PANTHER PTN001272083 + UniProtKB:Q15018	GO annotation record	High; definitively an inferred, not experimental, term
2	UniProt Q15018 (ABRAXAS2/ABRO1)	direct assay + localization	Competing source	Which protein actually binds microtubules?	ABRAXAS2 has GO:0008017 IDA, GO:0008608 IMP, GO:0090307 IMP; localizes to spindle pole / cytoskeleton	Human, BRISC complex	High; ABRAXAS2 is the true experimental source of the propagated cluster
3	Computed NW alignment (this run)	structural / evolutionary	Qualifies	Are ABRAXAS1 & ABRAXAS2 paralogs sharing a PANTHER family?	Global identity ~40.8% (shorter-seq) / ~45.4% (aligned core); both = MPN domain (7–160) + coiled-coil scaffold	Sequence computation	High; explains why IBA co-clusters them
4	UniProt Q6UWZ7 features/comments	database + localization	Refutes	What is ABRAXAS1's function/location?	Function = BRCA1-A scaffold recognizing K63-Ub histones; Loc: Nucleus only; MF IDA = GO:0031593 (polyUb-dependent binding)	Human	High
5	PMID 31253574 (Rabl et al., 2019, Mol Cell)	structural / review-level synthesis	Refutes	Are ABRAXAS & ABRO1 distinct scaffolds?	"…BRCC36 subunit that is functionalized by scaffold subunits ABRAXAS and ABRO1, respectively" — nuclear BRCA1-A vs cytoplasmic BRISC	Human, cryo-EM/biochem	High; clarifies paralog division of labour
6	PMID 28009280 (Kyrieleis et al., 2016)	structural	Refutes	Where does ABRAXAS1 act?	Recruited to damaged chromatin; cleaves K63-Ub on histones H2A/H2AX	Human, negative-stain EM	High; nuclear/chromatin, not spindle
7	PMIDs 17525340, 19261746, 19261748, 19261749	direct assay (IDA)	Refutes (context)	ABRAXAS1's real MF/CC	Foundational BRCA1-A papers; basis for GO:0070531 (IDA nucleus) and GO:0031593 (IDA polyUb binding)	Human cells	High
8	PMID 34272385 / 37198153 / 31630195	mutant phenotype / patient cells	Qualifies	ABRAXAS1 loss-of-function phenotypes	Genome-stability/HR pathway-choice, BRCA1 mislocalization, breast-cancer predisposition — all nuclear DNA-repair	Human patient cells	High; no spindle phenotype reported

GO Curation Implications

Lead (requires curator verification):

GO:0008017 microtubule binding (MF, IBA) → REMOVE / do not accept for ABRAXAS1.
It is an IBA carry-over whose only experimental support lies in the paralog ABRAXAS2
(Q15018). No ABRAXAS1-specific experimental evidence exists.
Co-propagated from the same node/source — treat identically:
GO:0008608 attachment of spindle microtubules to kinetochore (BP, IBA) → remove/flag.
GO:0090307 mitotic spindle assembly (BP, IBA) → remove/flag.
Retain the informative, experimentally grounded MF: GO:0031593
polyubiquitin modification-dependent protein binding (IDA) — this is the
appropriate, more-specific molecular function and should be the annotation curators
keep. (Do not fall back to bare "protein binding".)
Retain CC: GO:0070531 BRCA1-A complex (IDA), GO:0005634 nucleus (IDA).

This is an MF-term removal driven by paralog-based IBA over-annotation, with a
better-supported MF already in place.

Mechanistic Scope

Immediate molecular activity actually tested/supported for ABRAXAS1: a
scaffold/adaptor that reads K63-linked polyubiquitin (via associated RAP80/BRCC36)
and physically tethers BRCA1–BARD1 to ubiquitinated H2A/H2AX at DNA double-strand
breaks. Molecular function = polyubiquitin-dependent protein binding; cellular
component = nuclear BRCA1-A complex.
Microtubule binding is a molecular activity of a different gene product (ABRAXAS2),
not a downstream phenotype of ABRAXAS1. There is no direct-binding, localization, or
mutant evidence placing ABRAXAS1 on microtubules or the spindle.
ABRAXAS1's only mitosis-adjacent role is G2/M DNA-damage checkpoint signalling — a
regulatory/pathway consequence, distinct from GO:0008017.

Conflicts and Alternatives

Paralog confusion (primary explanation): ABRAXAS1 (Q6UWZ7, nuclear BRCA1-A) vs
ABRAXAS2 (Q15018, cytoplasmic BRISC, spindle pole). ~41% identical; same PANTHER
family → PAINT propagated ABRAXAS2's experimental spindle/microtubule terms to
ABRAXAS1. The withFrom field explicitly names Q15018 as the source.
Node separation is telling: the genuine ubiquitin-binding term (GO:0031593)
propagated from a different ancestral node (PTN001272081) and is independently IDA
on ABRAXAS1; the microtubule cluster propagated from PTN001272083 and is IDA/IMP
only on ABRAXAS2. The annotations that are real are internally corroborated; the
microtubule ones are not.
No isoform/organism rescue found: UniProt lists ABRAXAS1 localization as Nucleus
only; no spindle/cytoskeletal isoform is annotated.

Knowledge Gaps

Gap	What was checked	Why it matters	What would resolve it
Any direct ABRAXAS1–tubulin/microtubule interaction	Literature search (multiple queries) + UniProt + QuickGO	If a real interaction existed, removal would be wrong	An in vitro microtubule co-sedimentation/pelleting assay with purified ABRAXAS1; IF co-localization with tubulin in mitosis
Whether ABRAXAS1 ever localizes to spindle/centrosome	UniProt subcellular location (Nucleus only); HPA nuclear body	Localization would be a prerequisite for the term	High-resolution mitotic IF / live imaging of endogenous ABRAXAS1
PAINT node review status	Traced node PTN001272083 + Q15018	Curators may already have flagged the node	Inspect the PANTHER family tree annotation and whether a NOT/qualifier was applied

Discriminating Tests

Microtubule co-sedimentation assay with purified recombinant ABRAXAS1 ± taxol-stabilized
microtubules — direct test of GO:0008017; ABRAXAS2 as positive control.
Mitotic immunofluorescence / live imaging of endogenous ABRAXAS1 vs ABRAXAS2 to test
spindle-pole localization (expected: ABRAXAS2 yes, ABRAXAS1 no).
Depletion phenotype comparison: siABRAXAS1 vs siABRAXAS2 spindle-assembly / chromosome
mis-attachment scoring — the spindle phenotype should segregate to ABRAXAS2.
PANTHER tree audit of PTN001272083 to confirm the experimental basis is exclusively
ABRAXAS2-lineage and apply an experimental-evidence-based override for ABRAXAS1.

Curation Leads (require curator verification)

Action change: Remove/reject IBA GO:0008017 (and co-clustered GO:0008608,
GO:0090307) on ABRAXAS1 as paralog over-annotation; keep GO:0031593 (IDA) as the
informative MF.
Candidate reference to cite for the over-annotation source (verify snippet):
PMID 31253574 — "…the lysine-63 linkage-specific BRCC36 subunit that is functionalized
by scaffold subunits ABRAXAS and ABRO1, respectively."
Candidate reference for ABRAXAS1's true nuclear function (verify snippet):
PMID 28009280 — "It is recruited to damaged chromatin as a component of the BRCA1-A
deubiquitinase, which cleaves K63-linked ubiquitin chains attached to histone H2A and H2AX."
Suggested curator question: Does GO_Central want to add a NOT qualifier or
experimental-override for ABRAXAS1 at node PTN001272083, given that Q15018 (ABRAXAS2) is
the sole experimental source and ABRAXAS1 is nucleus-restricted?
Suggested experiment: microtubule co-pelleting of purified ABRAXAS1 (definitive).

Provenance (computed this run)

UniProt Q6UWZ7: 409 aa; MPN domain 7–160; coiled-coil 206–260; Loc = Nucleus; MF IDA = GO:0031593.
QuickGO: GO:0008017 = IBA / GO_REF:0000033 / assignedBy GO_Central / withFrom PANTHER:PTN001272083 + UniProtKB:Q15018.
UniProt Q15018 (ABRAXAS2): microtubule binding IDA; spindle attachment IMP; spindle assembly IMP; Loc = spindle pole/cytoskeleton.
Needleman–Wunsch ABRAXAS1 vs ABRAXAS2: ~40.8% identity (shorter seq) / ~45.4% (aligned core) → true paralogs in one PANTHER family.