Focused Hypothesis Report — *S. pombe* Mus81 mitochondrial localization & mtDNA function OpenScientist openscientist-autonomous 3 citations 2 artifacts 2026-08-31T19:57:27.618676 citations file

OpenScientist final report (text/html)

OpenScientist final report (application/pdf)

Focused Hypothesis Report — S. pombe Mus81 mitochondrial localization & mtDNA function

Gene: mus81 (UniProt P87231, MUS81_SCHPO), Schizosaccharomyces pombe (NCBITaxon:284812)
Source review: genes/SCHPO/mus81/mus81-ai-review.yaml — free-text hypothesis
Seed hypothesis: Mus81 has a genuine mitochondrial localization and a direct role in mitochondrial DNA metabolism, rather than the mitochondrial HDA signal being a high-throughput localization artifact.

Executive Judgment

Verdict: Refuted (the mitochondrial/mtDNA claim is over-annotated; the HDA mitochondrion signal is best treated as a high-throughput localization artifact).

Four independent lines of evidence converge against the seed hypothesis:

Annotation provenance. QuickGO shows the mitochondrion term (GO:0005739, HDA) and the nucleus term (GO:0005634, HDA) for Mus81 both trace to a single reference, PMID:16823372 (Matsuyama et al. 2006 — genome-wide C-terminal YFP tagging of ~90% of the S. pombe proteome). This is one high-throughput imaging dataset that reported a dual/ambiguous signal. In contrast, every low-throughput direct assay (IDA) for Mus81 comes from dedicated papers and places it in nuclear complexes (Holliday junction resolvase complex GO:0048476, PMID:11719193/14527419; nuclear replication fork GO:0043596, PMID:17363897), and the curated UniProt subcellular-location statement is Nucleus only.
Sequence (with positive control). The Mus81 N-terminus has no canonical mitochondrial targeting presequence, shown directly by side-by-side comparison with the bona fide mitochondrial resolvase Cce1/Ydc2 (UniProt Q10423, curated Mitochondrion):

Protein (first 30 aa)	Arg	R+K	D+E	Net charge	1st acidic pos	GRAVY
Cce1/Ydc2 (Q10423, genuine mito) `MATVKLSFLQHICKLTGLSRSGRKDELLRR`	4	7	2	+5	25	−0.25
Mus81 (P87231, claimed mito) `MDCGNPLFLQWIQEWMEESTRRFPKSYQTW`	2	3	4	−1	2	−0.92

Cce1/Ydc2 has the textbook Arg-rich, net-positive presequence with no acidic residue until position 25; Mus81 is net-negative with an acidic residue (Asp) immediately at position 2 — the opposite of an import signal.

Biological redundancy. S. pombe already possesses a dedicated mitochondrial Holliday-junction resolvase, Ydc2/SpCce1 (RuvC/CCE1 family), which localizes exclusively to mitochondria and is the enzyme responsible for mtDNA maintenance (loss → mtDNA aggregation/depletion). Mus81 is a structurally unrelated nuclear XPF-family endonuclease. There is no functional niche that requires Mus81 in mitochondria, and no primary literature reports any mitochondrial or mtDNA role for Mus81.
No conservation. Direct QuickGO cross-check: S. cerevisiae MUS81 (Q04149) and human MUS81 (Q96NY9) carry only nuclear cellular-component annotations and no mitochondrion term. The mitochondrion call is unique to S. pombe Mus81 and to one HDA reference — not a conserved feature of the MUS81 family.

Most important caveat: This assessment is negative/inferential. An HDA call is weak positive evidence, not proof of absence; a genuine minor mitochondrial pool cannot be excluded without a targeted, endogenously-tagged localization experiment. The sequence check is a heuristic (TargetP/MitoFates could not be executed programmatically here). But the balance of evidence is strongly against a genuine, functional mitochondrial role.

Evidence Matrix

Citation (PMID)	Evidence type	Stance	Claim tested	Key finding	Context	Confidence / limitations
UniProt P87231 (database)	Review/database	Qualifies	Where is Mus81 annotated?	Curated location = Nucleus only; mitochondrion = HDA; nucleus complexes = IDA	S. pombe, curated record	High for provenance; HDA is low-specificity
QuickGO / PMID:16823372 (Matsuyama 2006)	Localization (high-throughput)	Refutes (artifact source)	Origin of the mito call	Mitochondrion and nucleus HDA calls both come from the same genome-wide C-terminal YFP study; a single ambiguous HT dataset	S. pombe proteome screen	High for provenance; HT dual-localization is low-specificity
This report (computed, N-term MTS + positive control)	Computational	Refutes	Does Mus81 carry an MTS?	Mus81 first-30: net −1, acidic at pos 2, GRAVY −0.92 → no MTS; positive control Cce1/Ydc2 (Q10423): net +5, first acidic pos 25 → canonical MTS	Sequence heuristic + control	Medium-high; heuristic, but validated against a true mito protein
PMID 11719193 (Boddy 2001)	Direct assay / interaction	Refutes (context)	Mus81 identity/locale	Mus81-Eme1 identified as nuclear structure-specific endonuclease (basis of IDA GO:0048476)	S. pombe	High
23584455	Direct assay / mutant	Refutes (context)	Mus81 core function	Mus81-Eme1 is a nuclear HJ resolvase activated by CDK/ATR (Eme1 phospho) to prevent chromosomal rearrangements	S. pombe	High
22855558	Mutant phenotype	Refutes (context)	Mus81 substrate/locale	Resolves meiotic nuclear joint molecules (crossovers); regulated by Smc5/6	S. pombe meiosis	High
19470480	Mutant phenotype	Refutes (context)	Mus81 role	Mus81-Eme1 generates nuclear meiotic crossovers at DSBs	S. pombe meiosis	High
23982516	Structural / biochem	Refutes (context)	Mus81 mechanism	WH domain of MUS81 binds branched DNA; yeast phenocopy links to nuclear DNA-damage sensitivity	Human + S. pombe	High
10954073	Localization + mutant	Competing	Which enzyme handles mtDNA junctions?	SpCCE1/Ydc2-GFP localizes exclusively to mitochondria; ydc2Δ → mtDNA aggregation	S. pombe	High — identifies the true mt resolvase
12823554	Mutant / rescue	Competing	Ydc2 mtDNA role	ydc2Δ → mtDNA depletion, rescued only by active full-length Ydc2; mutants localize to mitochondria	S. pombe	High
11726496	Structural	Competing	Nature of mt resolvase	Ydc2 = eukaryotic RuvC-family mt HJ resolvase (crystal structure)	S. pombe	High
9421521 / 9343409	Biochemical	Competing	Ydc2 identity	Ydc2 is the functional S. pombe CCE1 homolog, a mt HJ resolvase	S. pombe	High
QuickGO orthologs Q04149 (S. cerevisiae), Q96NY9 (human)	Structural/evolutionary (comparative)	Refutes	Is the mito call conserved?	Both orthologs have exclusively nuclear CC terms; no GO:0005739 in either — mito call is S. pombe-only & single-study	Budding yeast + human	High for provenance; annotation completeness caveat
PubMed "Mus81 mitochondria/mtDNA"	Absence of evidence	Refutes	Any mt role for Mus81?	No papers returned	—	Absence, not proof

GO Curation Implications (leads — require curator verification)

GO:0005739 mitochondrion (CC), evidence HDA:PomBase — lead: consider REMOVE or DO-NOT-PROPAGATE / NOT-core. It is unsupported by any focused assay, contradicted by the curated Nucleus location, by the absence of an MTS, and by the existence of a dedicated mitochondrial resolvase (Ydc2). At minimum it should not be treated as representing a biological function. If retained, flag as high-throughput-only and non-core.
Retain nuclear CC terms: GO:0048476 Holliday junction resolvase complex (IDA) and GO:0043596 nuclear replication fork (IDA); GO:0005634 nucleus is well supported. These are the genuine location.
MF: retain crossover junction endonuclease activity / structure-specific (5'-flap / branched DNA) endonuclease activity.
BP: retain nuclear DNA repair / recombination / replication-fork processing / meiotic recombination. Do not add mtDNA-metabolism BP terms to Mus81.
Avoid a bare "protein binding" fallback — the informative, supported terms are the nuclear resolvase MF/CC/BP set above.

Mechanistic Scope

Direct molecular activity (supported): Mus81, in a heterodimer with Eme1, is a Mg²⁺-dependent structure-specific endonuclease that incises branched DNA (3'-flaps, D-loops, replication forks, nicked Holliday junctions) in the nucleus, resolving recombination/replication intermediates. Activity is cell-cycle/DNA-damage regulated via CDK/ATR phosphorylation of Eme1.

Claimed but unsupported (seed hypothesis): import into mitochondria and direct action on mtDNA. This is not an observed direct activity; it is inferred only from a high-throughput localization dataset. The corresponding mtDNA-junction-processing function in S. pombe is performed by the unrelated Ydc2/SpCce1.

Conflicts and Alternatives

Dedicated-enzyme alternative (strongest): Ydc2/SpCce1 already fulfills mitochondrial HJ resolution/mtDNA maintenance; a second, unrelated nuclease for the same niche is biologically unnecessary and unreported.
Artifact of high-throughput tagging: N-terminal or over-expression YFP/GFP tagging commonly yields low-level spurious mitochondrial signals; the HDA code is exactly this class of data.
Paralog/ortholog carry-over (checked directly): QuickGO CC annotations for S. cerevisiae MUS81 (Q04149) and human MUS81 (Q96NY9) are exclusively nuclear (nucleus, nuclear chromosome, nuclear replication fork, HJ resolvase complex, telomeric/chromosomal region, nucleoplasm) with no mitochondrion (GO:0005739) term in either. The mitochondrion call exists only for S. pombe Mus81 and only via the single HDA reference PMID:16823372 — i.e., it is species-specific and not conserved, arguing against genuine, conserved mitochondrial biology.
Not excluded: a small, functionally minor mitochondrial pool, or a stress-specific relocalization, cannot be ruled out by these data alone.

Knowledge Gaps

Provenance of the exact HDA dataset — RESOLVED. Checked: QuickGO → the mitochondrion HDA (and nucleus HDA) call both derive from PMID:16823372 (Matsuyama et al. 2006), a C-terminal (3′) YFP genome-wide screen. Residual gap: raw image for Mus81 in that dataset not inspected here; a C-terminal tag would not mask an N-terminal MTS, so the mito signal most plausibly reflects HT imaging ambiguity/overexpression. Resolve: inspect the original localization image/record.
Endogenous localization. Checked: curated Nucleus statement + IDA nuclear complexes. Gap: no endogenously-tagged, mitochondria-co-stained Mus81 image was located. Resolve: mNeonGreen knock-in Mus81 + MitoTracker/mtDNA co-staining and fractionation.
Formal MTS prediction. Checked: manual N-terminal charge/hydropathy heuristic (no MTS). Gap: TargetP2/MitoFates/DeepMito not runnable programmatically here. Resolve: run these predictors on P87231.
mtDNA phenotype of mus81Δ. Checked: literature — none reported. Gap: no direct test of mtDNA copy number/integrity in mus81Δ. Resolve: qPCR mtDNA copy number and mtDNA topology in mus81Δ vs ydc2Δ.

Discriminating Tests

Endogenous knock-in localization + fractionation: tag chromosomal mus81 with a fluorophore; co-image with a mitochondrial marker and quantify Mn-fold enrichment; biochemically fractionate and immunoblot for Mus81 in purified mitochondria (with nuclear/cytosolic controls).
mtDNA functional assay: measure mtDNA copy number, aggregation, and petite/respiration phenotypes in mus81Δ vs ydc2Δ vs WT. Ydc2Δ shows mtDNA aggregation/depletion; a genuine Mus81 role predicts a (possibly redundant) mtDNA phenotype, whereas an artifact predicts none.
MTS predictors (TargetP2, MitoFates, DeepMito) on P87231 and a fused N-terminal-Mus81(1–40)-GFP import assay.
Epistasis: mus81Δ ydc2Δ double mutant — synthetic mtDNA defect would argue for a real (backup) mitochondrial role; no aggravation argues against.

Curation Leads (require curator verification)

Action change: Mark GO:0005739 mitochondrion (HDA) as non-core / candidate removal for Mus81; keep nuclear CC/MF/BP terms as the core annotation.
Candidate supporting references (verify snippets):
PMID 10954073: "a SpCCE1-GFP fusion localises exclusively to the mitochondria of S. pombe" — identifies the true mitochondrial resolvase (competes with the Mus81 mt claim).
PMID 12823554: "Cells that lacked Ydc2 showed a significant depletion of mtDNA content" — the mtDNA-maintenance function is Ydc2's, not Mus81's.
PMID 23584455: DNA-damage-induced activation of the nuclear Mus81-Eme1 resolvase — Mus81's genuine role/location.
UniProt P87231: curated SUBCELLULAR LOCATION = Nucleus; mitochondrion term is HDA only.
Suggested curator questions: Which high-throughput study underlies the HDA mitochondrion call, and what tag/expression was used? Is there any endogenous, non-HT evidence for Mus81 in mitochondria?
Suggested experiments: endogenous-tag co-localization + fractionation; mtDNA copy-number/topology in mus81Δ; MTS predictions and N-terminal import assay.

Provenance note

Computational analyses run here (code + raw outputs in the iteration transcript): (1) live UniProt REST fetch of P87231 (length 608; curated Nucleus; GO-CC evidence codes as tabulated); (2) QuickGO annotation query for P87231 → both mitochondrion and nucleus HDA terms map to PMID:16823372; nuclear IDA terms map to PMID:11719193/14527419/17363897; (3) N-terminal targeting-signal feature calculation for Mus81 vs the mitochondrial positive control Cce1/Ydc2 (Q10423): Mus81 first-30 net −1, acidic at pos 2, GRAVY −0.92; Cce1/Ydc2 net +5, first acidic pos 25, GRAVY −0.25. Writing to a provenance directory was blocked by sandbox permissions; results are reproducible from the recorded code. No result was fabricated; dedicated MTS predictor tools (TargetP2/MitoFates/DeepMito) could not be executed in this environment and are listed as gaps.