Epe1

Summary: This IBA annotation is not supported. Epe1 has a non-canonical active site (its Fe(II)-binding triad is H297-E299-Y370, so the third iron ligand, a His in canonical HX(D/E)...H JmjC demethylases, is replaced by Tyr370) and shows no detectable demethylase activity in vitro (Tsukada 2006, PMID:16362057; Raiymbek 2020, PMID:32195666). Its characterized anti-silencing mechanisms (Swi6 binding, SAGA association, and recruitment of the Bdf2 bromodomain protein to heterochromatin boundaries) have not been shown to require catalysis, although the JmjC domain is required for Epe1 function and Wang et al. 2015 and Sorida et al. 2019 interpret active-site mutant phenotypes as loss of enzymatic function (see reason).

Reason: What is removed here is the PAINT inference path, the propagation of canonical JHDM1-type histone demethylase activity from the PANTHER node to Epe1, not the proposition that Epe1 has histone demethylase activity. OLS lists GO:0032452 among the hierarchical ancestors of GO:0032454 histone H3K9 demethylase activity, whose two experimental rows (IDA and EXP, PMID:25838386) are kept UNDECIDED. If PomBase upholds either of them, GO:0032452 follows from it by the true-path rule, so this REMOVE applies only to the IBA row as an inference. The review names this parent/child tension explicitly rather than resolving it silently, as it does for the GO:0006325/GO:0006338 pair (generic ancestors kept as non-core) and the negated GO:0031507 against the accepted GO:0033696 rows (left UNDECIDED and raised with PomBase); the parallel is in how the relation is treated, not in the action taken. GO:0016491 and GO:0051213 are not ancestors of GO:0032454 (OLS hierarchicalAncestors of GO:0032454), so their REMOVE rows do not formally conflict with those children, although JmjC demethylation is 2-oxoglutarate-dependent dioxygenase chemistry, so a positive GO:0032454 finding would bear on them biologically. No demethylase activity has been detected for purified Epe1. Mass spectrometry assays using purified Epe1 with methylated H3K9 peptides showed no detectable removal of methyl groups, even with a five-fold excess of Swi6/HP1 (Raiymbek 2020). The JmjC domain lacks one of the conserved Fe(II)-binding residues (Tyr370 in place of the third His). The predicted Fe(II)-site mutant H297A is not a clean test of catalysis, and its phenotype depends on the assay. At endogenous expression it behaves like epe1 deletion in erasing tethering-induced H3K9me (Audergon 2015) and fails to remove established ectopic heterochromatin, yet still suppresses ectopic heterochromatin-mediated variegation and has reduced heterochromatin localization (Sorida 2019). When overexpressed, H297A still disrupts pericentric silencing (milder than wild type from the chromosomal nmt41 promoter, similar to wild type from a plasmid), and this effect depends on SAGA (Bao 2019). Raiymbek et al. report the same asymmetry for their cofactor-site mutants ("Hence, co-factor binding mutants of Epe1 can act as multi-copy suppressors of epigenetic silencing despite the presumptive loss of enzymatic activity."). Wang et al. 2015 report that epe1-H374A and epe1-Y307A cause sickness similar to epe1 deletion when combined with mst2 loss, and interpret this as redundancy between the enzymatic activities of Mst2 and Epe1 (PMID:25774602). Y307 matches the UniProt Y307A mutagenesis site, but residue 374 of O94603 (948 aa) is Thr, not His, and is not a triad residue (the triad is H297, E299 and Y370), so which histidine Wang et al. mutated cannot be determined from the cache. This sits uneasily with the H297A results above, which show part of Epe1's anti-silencing activity surviving H297A, whereas UniProt records Y307A as loss of function from Ayoub 2003. The papers do not reconcile the two. They concern different residues (Raiymbek et al. assign Y307 to the alpha-ketoglutarate site and H297 to the Fe(II) site, and H297 is a UniProt Fe ligand, rule-predicted from PRU00538), different assays and expression levels, and the cached Ayoub abstract does not describe the Y307A assay. These results show that part of Epe1's anti-silencing activity does not need an intact H297, but they do not show that no function needs it; the IBA is removed on the biochemical and active-site evidence, not on the H297A phenotype. The C-terminus alone (without JmjC) can disrupt heterochromatin, but only partly ("Therefore, the Epe1434-948 mutant is a hypomorphic allele that partially retains wild-type levels of Epe1 anti-silencing activity."), and it is expressed at lower levels ("The Epe1434-948 protein is expressed at levels that are at least 4-5 fold lower than full-length Epe1 which in part might explain its limited efficacy in cells."; Raiymbek 2020). Several results bear on whether catalysis contributes. The JmjC domain is required for Epe1 function. Complementation experiments show that "the jmjC domain is essential for Epe1 activity" (Ayoub 2003, PMID:12773576), and Zofall and Grewal find that Epe1's effect on Pol II accessibility requires the JmjC domain (PMID:16762840). A required domain could reflect folding or binding rather than catalysis. Against a gross folding defect, recombinant H297A has the same denaturation temperature as wild type ("Wild-type Epe1 and Epe1 H297A exhibit similar denaturation temperatures, implying that the mutation within the JmjC domain does not destabilize the protein or cause substantial alterations in protein structure"; PMID:32195666), although this was measured for H297A only and does not exclude a local change in the domain. For all three mutants, Raiymbek et al. also report that "We verified that the expression level of all Epe1 mutant proteins is equal relative to an actin loading control. Hence, neither overexpression artifacts nor changes in protein stability contribute to the maintenance-specific phenotype we observed in our genetic assays". That closes the expression and stability part of the objection for H297A, Y307A and Y370A; a local conformational change, as Sorida et al. speculate (below), stays open. Zofall and Grewal themselves note that the mechanism "might be distinct from other JmjC proteins that possess known demethylase activities", and Raiymbek et al. show that the JmjC-containing half of Epe1 binds H3K9 methylation directly ("Next, we expressed and purified a C-terminal truncation mutant of Epe1, MBP-Epe1-ΔC from Sf9 insect cells, which includes amino acids 1–600 and includes the putative catalytic JmjC domain. We found that Epe1-ΔC can also directly bind to an H3K9me3 peptide and specifically interacts with H3K9 methylated histones (Figure 5—figure supplement 1A,B)."). That fragment (amino acids 1–600) lacks the C-terminal 601-948 region, although it still contains the minimal Swi6-binding site ("The second fragment corresponds to only the minimal Swi6HP1 interaction site extending from 434 to 600 amino acids (Epe1434-600)."), and the peptide and histone assays contain no Swi6. H3K9me recognition by the JmjC-containing half is therefore Swi6-independent and shown by data, not only hypothesized. The authors' interpretation is that the JmjC domain itself "might be primarily responsible for H3K9 methylation recognition and binding", a non-catalytic role; the fragment tested also includes sequence outside the JmjC domain (233–434), so the assignment to the JmjC domain proper remains their inference. UniProt also records Y307A as loss of function, with experimental evidence from the same Ayoub paper (FT MUTAGEN 307; UniProt's reference line for PMID:12773576 reads "FUNCTION, SUBCELLULAR LOCATION, AND MUTAGENESIS OF TYR-307"). MUTAGEN 307 is the only mutagenesis recorded in O94603 and no domain deletion is recorded, so the domain-essential statement and the Y307A record may reflect the same Ayoub experiment; the cached abstract does not mention residue 307, so this cannot be settled here, and the Y307A record should not be counted as a separate, stronger line of evidence. UniProt has no BINDING feature at 307. Three cached statements bear on what Y307 is. Raiymbek et al. assign it to the 2-oxoglutarate site ("Alanine substitutions of amino acid residues involved in Fe (II) or α-ketoglutarate binding (epe1 H297A and epe1 Y307A, respectively) disrupt co-factor binding"), their alignment figure distinguishes the two sites ("Iron-binding residues are colored in green and α-ketoglutarate-binding residues are colored in purple"), and their grouping elsewhere of "residues that affect Fe(II) or α-ketoglutarate binding (H297A, Y307A, and Y370A)" is consistent with that. Sorida et al. describe Epe1Y307A as a JmjC mutant "which retains the metal-binding residues", citing a background reference ([12] in their text) that cannot be identified from the cache, which has no reference list (PMID:31206516). The three agree once the wording is read closely; Y307 is placed at the 2-oxoglutarate site, not among the Fe(II) ligands (H297, E299, Y370). The Y307A loss of function therefore implicates the 2-oxoglutarate-binding part of the JmjC cofactor pocket. That makes it residue-level evidence that the cofactor pocket matters, not only that the JmjC fold must be present, but it does not show that catalysis is what matters. Raiymbek et al. state that the substitutions "disrupt co-factor binding", but they assert this from the residues' assigned roles in the JmjC active site rather than from a measurement. The data shown for these mutants are reporter phenotypes ("When expressed at endogenous levels, these mutants form red or sectored colonies on +tetracycline-containing medium and resemble epe1Δ cells"), Swi6 co-immunoprecipitation and localization. The cached texts report no cofactor-binding or activity measurement on the Y307A protein, and elsewhere the authors call the co-factor mutants' loss of enzymatic activity "presumptive". Y307A also attenuates Swi6 binding in co-immunoprecipitation and "fails to co-localize with mCherry-Swi6HP1" in vivo (PMID:32195666), so a pocket mutation that alters the Swi6-binding surface or localization would give the same phenotype. That in vivo result conflicts with the reference Sorida et al. cite for Swi6 interacting with Epe1Y307A; the conflict is not only one of degree. Raiymbek's text says the mutations "significantly attenuate" the interaction, but their figure legend states that "The interaction between the two proteins is preserved in wild-type cells and is completely eliminated in all Epe1 JmjC mutants", and all three mutants lose dg occupancy ("Consistent with our co-immunoprecipitation studies, all co-factor binding mutants of Epe1 exhibit a significant reduction or completely fail to localize at the pericentromeric dg repeats (Figure 2B)."). The cached sources do not resolve it. Wang et al. 2015 use epe1-Y307A as an enzymatically dead allele and interpret the sickness of active-site mutants combined with mst2 loss as redundancy between the enzymatic activities of Mst2 and Epe1 (PMID:25774602). That reading is consistent with a 2-oxoglutarate-site mutation, but it is an inference from the residue's position, not a measurement. The divergent position itself is also required. Raiymbek et al. report that "Replacing the non-conserved tyrosine residue in Epe1 with alanine (epe1 Y370A) leads to a similar loss of function phenotype. Hence, despite the lack of conservation, a natural tyrosine substitution within the JmjC domain of Epe1 is essential for its anti-silencing function in cells." This is independent of Ayoub, and it bears on this removal's own premise. Tyr370 is not a decayed, dispensable position but a residue Epe1 needs. Required is not catalytic, though. H297A, Y307A and Y370A all weaken the Swi6 interaction, so the same alternative explanation applies to Y370A as to the other two. The triad argument therefore rests on Tyr370 differing from the canonical His ligand, and on the undetected activity, not on the site being functionless. Y370A's "similar loss of function phenotype" is the same red-or-sectored colony readout described above for H297A and Y307A. The direct readout for all three residues is H3K9me itself. The mutants establish ectopic H3K9me2 like wild type ("Both wild-type and Epe1 mutant strains exhibit high levels of H3K9me2 during establishment (Figure 1C)."), but unlike wild type they keep it after the initiator is released ("In contrast, Epe1 mutants that exhibit a red or sectored phenotype upon +tetracycline addition retain high levels of H3K9 methylation at the ectopic site (Figure 1D)."). That is an in vivo H3K9me measurement for H297A, Y307A and Y370A, independent of both Ayoub and Sorida. It shows that the three residues are needed for ectopic H3K9me to be lost, not that Epe1 removes it catalytically. The strongest genetic result is independent of Ayoub. Sorida et al. 2019 found that H297A, a UniProt Fe ligand (rule-predicted, PRU00538), separates two functions. The mutant still suppressed red-white variegation, the de novo arm that depends on the N-terminal activation domain, but entirely failed to remove already-established ectopic heterochromatin, which they attribute to the JmjC domain (PMID:31206516). The removal arm has a dosage condition. It is read at single-copy expression, where even wild-type Epe1 removes ectopic H3K9me incompletely ("We introduced a single copy of Epe1 into epe1Δ clones harboring ectopic heterochromatin, and found that Epe1 could reduce H3K9me from ectopic heterochromatin but some of the heterochromatin persisted."), and Sorida et al. report that "We found that re-introduction of single copy Epe1 did not erase ectopic heterochromatin when an H3K9me source existed nearby, while Epe1 overexpression completely erased it". So the H297A removal defect is observed at endogenous dose, where removal is already limited, and whether overexpressed H297A would remove established ectopic heterochromatin is not reported. This is one experiment read from its two arms. Prevention of de novo H3K9me is retained under H297A, which supports the point above that part of the anti-silencing activity does not need an intact H297; removal of established H3K9me is lost, which is the counter-evidence here. That is a different paper, residue and assay from Ayoub's; it is a separation of function rather than a loss of function, which a folding defect would not predict; and it reads out H3K9me, the mark a demethylase would act on ("Introduction of epe1H297A as well as epe1Δ into epe1Δ W-t1 resulted in diploid cells that retained the white phenotype and H3K9me at ade5*, while introduction of epe1+ complemented the white phenotype and almost depleted H3K9me of ade5* (Fig 4C and 4D)."). The in vivo H3K9me readout is not unique to Sorida, since Raiymbek report it for all three cofactor-site mutants (above); Sorida's distinct contribution is the separation of function. The retained arm is nearly but not completely retained. Sorida et al. report that "indeed, 96.2% of Epe1H297A cells formed WT-like red colonies, while 61.7% of epe1Δ cells did.", that "Epe1H297A cells generated a few pink colonies that were not generated by wild-type cells" and conclude "that JmjC-dependent demethylation contributes to full suppression to some extent". Their abstract also states that "Epe1 prevented ectopic H3K9me deposition independently of both its JmjC-mediated demethylation and heterochromatin association ability". The retained arm is therefore expected despite H297A's loss of heterochromatin localization, while the lost removal arm, which acts on established heterochromatin, remains open to either reading, loss of catalysis or loss of heterochromatin association. Wang 2015 reads the active-site mutants as catalysis-dead; no study has measured enzymatic demethylation directly, and the same mutations also weaken Swi6 binding and heterochromatin localization (Raiymbek, Sorida), which could explain the phenotypes without catalysis. Sorida et al. propose this mechanism themselves ("we speculate that conformational changes in the JmjC domain induced by perturbations in Fe2+ binding result in a slight alteration of the interaction surface for Swi6 binding, while severely disrupting the structure of a region essential for heterochromatin association"), while adding that "We also found that the H297A mutation slightly impaired the interaction of Epe1 with Swi6, although this may not fully explain the reduced heterochromatin localization of Epe1 (Fig 4H, S4G Fig).". Raiymbek et al. adopt the same reading ("These results suggest that mutations within the JmjC domain of Epe1 may induce a conformational change that attenuates Swi6HP1 binding (Sorida et al., 2019).") and state its limit ("Our co-immunoprecipitation, imaging, and ChIP experiments preclude us from making any conclusions as to whether the interaction between Epe1 and Swi6HP1 is mutation-dependent or requires the putative catalytic functions of Epe1 in vivo."). Three of their results point away from lost chemistry as the cause of the Swi6 defect. Recombinant Epe1 binds Swi6 without cofactors ("The interaction between Epe1 and Swi6HP1 is nearly identical in the presence or absence of EDTA (Figure 2E)."), yet recombinant H297A does not ("However, this type of interaction and increase in binding is not observed in the case of MBP-Epe1 H297A (Figure 2E)."). And H297A lets the deacetylase Clr3 back onto Swi6 ("We measured a weak interaction between Swi6HP1 and Clr3 in a wild-type Epe1 background that substantially increases in an Epe1 H297A strain background (Figure 6A)."), a non-catalytic route to the phenotype. That route is supported by their Clr3 data. Deleting clr3 raises Epe1 occupancy ("Consistent with our model, we observed a three-fold increase in Epe1 localization at the pericentromeric dg repeats and the mating type locus (mat) (Figure 6C, D)."), and tethering Clr3 overrides Epe1 ("Cells exhibit a red and sectored phenotype in cells where Clr3 is artificially tethered, despite Epe1 still being present (Figure 7B)."; "These results suggest that constitutively tethering Clr3 to sites of heterochromatin formation is sufficient to oppose Epe1 activity resulting in maintenance of H3K9 methylation even after +tetracycline addition."). In vitro, adding Fe(II), alpha-ketoglutarate and ascorbate "did not alter the extent of interaction between Epe1 and Swi6HP1", so Swi6 binding does not depend on demethylase reaction conditions, although Raiymbek et al. note that these assays "fail to capture any effect that co-factor binding itself may have on the interaction between Epe1 and Swi6HP1" (PMID:32195666). The removal rests on the undetected in vitro activity and the non-canonical triad; this genetic evidence is what the UNDECIDED GO:0032454 rows reflect. OpenScientist independently supported removal, tracing the live IBA to PANTHER:PTN000564171 and confirming the H297-E299-Y370 active-site divergence from canonical H-D-H JmjC demethylases.

Summary: Chromatin remodeling is true for Epe1 but general. It is an ancestor of heterochromatin boundary formation (GO:0033696), which Epe1 already carries on seven ACCEPTed experimental rows.

Reason: The row is correct because OLS lists GO:0006338 among the hierarchical ancestors of GO:0033696 heterochromatin boundary formation, which Epe1 carries on seven ACCEPTed experimental rows; the IBA is therefore implied by those rows and adds no information of its own, and the earlier MODIFY to GO:0033696 would only have duplicated them. It is kept as a correct, non-core generic term, on the same grounds as its parent GO:0006325. It is not grounded on SAGA recruitment or nucleosome turnover. Acetylation is performed by Gcn5 in SAGA (Epe1's part is the GO:0062070 SAGA complex binding row), and turnover by chaperones and remodelers such as FACT (see core function 4). The supporting quote is the boundary-formation result that the ancestry argument rests on.

Summary: Epe1 does regulate transcription at heterochromatic repeats by recruiting SAGA complex and promoting RNA Pol II occupancy. It enables transcription of centromeric repeats that feed into the RNAi pathway for heterochromatin establishment.

Reason: Overexpressed Epe1 recruits SAGA to heterochromatin to promote histone acetylation and transcription of the underlying repeats (PMID:30573453). Studies show increased Pol II occupancy and transcript production from dg/dh repeats when Epe1 is overexpressed. This transcription is essential for generating RNAi substrates that maintain heterochromatin in a regulated manner. Evidence not tied to overexpression comes from Zofall and Grewal 2006, who report that Epe1 promotes Pol II accessibility at heterochromatin (PMID:16762840); their paper is cached as an abstract only, which does not state whether that result used endogenous or overexpressed Epe1.

Summary: Epe1 functions as a transcriptional coregulator by recruiting the SAGA histone acetyltransferase complex to heterochromatin sites, promoting transcriptional activation through histone acetylation.

Reason: Mass spectrometry of affinity-purified overexpressed Epe1 identified many SAGA components (PMID:30573453), and overexpressed Epe1 recruits this co-activator complex to heterochromatin. The N-terminal region contains a transcriptional activation domain that contributes to anti-silencing activity.

Summary: Nuclear localization of Epe1 is well established; the PomBase IDA row (PMID:12773576) supports it.

Reason: Ayoub et al. 2003 describe Epe1 as a conserved nuclear protein (PMID:12773576); the cached abstract does not state the method used. Epe1 also localizes to heterochromatin, which is nuclear, consistent with its function in heterochromatin regulation.

Summary: Chromatin organization is true for Epe1 but is the most generic of its chromatin process terms.

Reason: The row is correct because OLS lists GO:0006325 among the hierarchical ancestors of GO:0033696 heterochromatin boundary formation, which Epe1 carries on seven ACCEPTed experimental rows. Within this review the only process the row covers is that boundary formation, so it adds nothing the GO:0033696 rows do not already say. It does not stand in for Epe1's role in heterochromatic nucleosome turnover, which core function 4 describes without any process term because Epe1 is required for turnover but not shown to perform it. The row is kept as non-core rather than removed, on the same grounds as its child GO:0006338 chromatin remodeling.

Summary: This annotation is based on JmjC domain homology. No oxidoreductase or demethylase activity has been detected for Epe1, and its JmjC Fe(II)-binding triad is non-canonical, with Y370 in place of the third histidine ligand.

Reason: What is removed is the InterPro-keyword inference of canonical JmjC enzyme activity, not a claim that latent or in vivo activity is excluded (see the UNDECIDED GO:0032454 rows). No demethylase or other oxidoreductase activity has been detected for purified Epe1 (PMID:16362057, PMID:32195666). The JmjC domain has a non-canonical Fe(II)-binding triad (H297-E299-Y370; Tyr370 replaces the third, His, iron ligand of canonical JmjC demethylases), although Tyr370 is required for Epe1 function (PMID:32195666). GO:0016491 is not an ancestor of GO:0032454 histone H3K9 demethylase activity (OLS hierarchicalAncestors of GO:0032454), so this REMOVE does not formally conflict with the UNDECIDED GO:0032454 rows, although JmjC demethylation is 2-oxoglutarate-dependent dioxygenase chemistry and a positive GO:0032454 finding would bear on it biologically. The genetic evidence on whether the JmjC domain contributes catalysis (the domain and the Y307A and Y370A residues are required for function, H297A, Y307A and Y370A retain ectopic H3K9me, and H297A separates prevention from removal) and the non-catalytic explanations for it are weighed in full in the GO:0032452 row and are not repeated here. None of that evidence measures an enzymatic activity, and this removal rests on the undetected activity and the non-canonical triad.

Summary: Metal ion binding by Epe1 has not been measured. UniProt asserts Fe cation binding at H297 and E299, which are retained, but the third canonical iron ligand is replaced by Y370, so whether Epe1 binds Fe(II) is unresolved.

Reason: This IEA comes from the UniProt Metal-binding keyword (KW-0479), which rests on UniProt's BINDING 297 and BINDING 299 features; in the entry, each of those two BINDING lines is followed by /ligand="Fe cation", /ligand_note="catalytic" and a PROSITE-ProRule PRU00538 evidence line. Those features are themselves predicted from the JmjC PROSITE ProRule (ECO:0000255, PRU00538), not measured, so the chain is JmjC domain, then PRU00538, then the BINDING features, then KW-0479, then GO:0046872. It is therefore a domain-derived inference, although UniProt places it on specific residues. Both residues are retained, and Audergon et al. treat H297 as the Fe(II) site (PMID:25838386). The Epe1 JmjC domain has tyrosine at position 370 in place of the third iron-coordinating histidine of active JmjC enzymes, so canonical three-ligand coordination is absent. Metal binding has not been measured, and a two-ligand or Tyr-assisted site cannot be excluded. Y370A causes a loss of function similar to H297A and Y307A ("a natural tyrosine substitution within the JmjC domain of Epe1 is essential for its anti-silencing function in cells"), so the divergent Tyr is required, though not shown to be a metal ligand. Mutating H297 or Y370 attenuates Swi6 binding in pull-downs (PMID:32195666), which is consistent with, but does not demonstrate, a metal-dependent role for the site. There is no positive evidence that the site cannot bind iron, so the row is left UNDECIDED, alongside the experimental GO:0032454 rows, until metal binding is tested directly.

Summary: Inferred from JmjC domain presence. No dioxygenase activity has been detected for Epe1, and its Fe(II)-binding triad is non-canonical; it is a probable pseudo-enzyme.

Reason: What is removed is the InterPro-keyword inference of canonical JmjC dioxygenase activity, not a claim that latent or in vivo activity is excluded (see the UNDECIDED GO:0032454 rows). No dioxygenase or demethylase activity has been detected for purified Epe1 (PMID:32195666), and the Fe(II) triad is non-canonical, with Y370 in place of the third iron-coordinating histidine, although Tyr370 is required for Epe1 function (PMID:32195666). GO:0051213 is not an ancestor of GO:0032454 histone H3K9 demethylase activity (OLS hierarchicalAncestors of GO:0032454), so this REMOVE does not formally conflict with the UNDECIDED GO:0032454 rows, although JmjC demethylation is 2-oxoglutarate-dependent dioxygenase chemistry and a positive GO:0032454 finding would bear on it biologically. The genetic evidence on whether the JmjC domain contributes catalysis (the domain and the Y307A and Y370A residues are required for function, H297A, Y307A and Y370A retain ectopic H3K9me, and H297A separates prevention from removal) and the non-catalytic explanations for it are weighed in full in the GO:0032452 row and are not repeated here. None of that evidence measures an enzymatic activity, and this removal rests on the undetected activity and the non-canonical triad.

Summary: This H3K36-specific demethylase annotation is not supported. The H3K36 specificity comes from JHDM1-family homologs such as budding-yeast Jhd1, which is a distinct protein; no demethylase activity has been detected for Epe1, and H3K36 has not been tested.

Reason: What is removed is the electronic inference (GO_REF:0000120) of canonical JHDM1-type H3K36 demethylase activity, not a claim that latent or in vivo activity is excluded (see the UNDECIDED GO:0032454 rows). The H3K36 specificity is transferred by family membership from enzymes such as budding-yeast Jhd1, which demethylates H3K36 and is distinct from Epe1. No cited study has tested Epe1 on H3K36-methylated substrates. Purified Epe1 showed no demethylation of di- or tri-methyl H3K9 peptides (PMID:32195666), and the in vivo evidence bearing on Epe1 catalysis concerns H3K9 methylation only. With no activity detected on any tested substrate and a non-canonical Fe(II) triad (H297-E299-Y370), there is no basis for the H3K36 specificity. Tyr370 is required for Epe1 function (Y370A causes a loss of function phenotype, PMID:32195666), but that shows the residue is needed, not that it supports catalysis.

Summary: This IntAct IPI row records Epe1 binding Cdt2 (UniProtKB:Q10990), the substrate receptor of the Cul4-Ddb1(Cdt2) ubiquitin ligase that targets Epe1 for ubiquitylation and degradation. The interaction is real, but protein binding is too generic; ubiquitin protein ligase binding describes it.

Reason: The WITH partner of this IPI annotation is Q10990, which UniProt identifies as cdt2. PMID:21215368 shows that the Cul4-Ddb1(Cdt2) complex directly recognizes Epe1 and promotes its ubiquitylation and degradation, which restricts Epe1 to heterochromatin boundaries. The replacement GO:0031625 ubiquitin protein ligase binding stands in for this Q10990 interaction. Epe1's other partners come from other references and are represented elsewhere in the review, not by this row. Swi6 binding is the NEW GO:0070087 row, SAGA association the NEW GO:0062070 row, and Bdf2 recruitment the NEW GO:0030674 row; each also has a core-function entry.

Proposed replacements: ubiquitin protein ligase binding

Summary: This is one of Epe1's core functions - establishing and maintaining heterochromatin boundaries, in part through recruitment of the anti-silencing factor Bdf2.

Reason: Extensive evidence supports this annotation. It recruits the Bdf2 bromodomain protein to IRCs, the inverted repeats flanking centromeres (PMID:24013502), and disruption of epe1 lets heterochromatin spread across barriers (PMID:12773576). This is a well-characterized core function.

Summary: PomBase IDA for H3K9 demethylase activity from Audergon et al. (2015). The full text (PMC4397586, now cached) contains no in vitro demethylase assay. The evidence is in vivo - epe1 deletion, and the Fe(II)-site (H297A) and 2-oxoglutarate-site (K314A) point mutants, allow tethered-Clr4-initiated H3K9me and silencing to persist after Clr4 release, and re-introducing epe1+ removes the persistent H3K9me. The authors interpret this as Epe1 acting as an H3K9 demethylase, while stating that demethylase activity has not been detected.

Reason: The paper shows that Epe1 is required to erase H3K9me from an ectopic heterochromatin site and that this requires its predicted Fe(II)/2-oxoglutarate-binding residues, which is real experimental evidence the curators used. It does not show that Epe1 itself catalyses demethylation, and IDA is not the matching code for a genetic result. Other work reports no detectable demethylase activity for purified Epe1 and finds that the H297A mutation compromises Swi6 binding in pull-downs (Raiymbek 2020, PMID:32195666), while Sorida 2019 (PMID:31206516) finds only a slight binding defect but a large loss of heterochromatin localization; and the Fe(II) triad is non-canonical (H297-E299-Y370, with Tyr370 in place of the third His ligand). The authors note that the PHF2 JmjC domain carries a similar anomaly yet has latent, phosphorylation-activated H3K9 demethylase activity, so catalysis is disputed rather than refuted. UniProt also records Y307A as loss of function (MUTAGEN 307, from Ayoub 2003), which fits a JmjC-residue-dependent function. Raiymbek et al. assign Y307 to the 2-oxoglutarate site, and Sorida et al. describe Epe1Y307A as retaining the metal-binding residues, so the record implicates the 2-oxoglutarate part of the cofactor pocket; Y307A also fails to co-localize with Swi6 in vivo (PMID:32195666), so it does not isolate catalysis. The UniProt Y307A record and Ayoub's statement that the jmjC domain is essential may come from the same experiment, and neither measures catalysis. Sorida et al. 2019 give independent support for a JmjC-dependent function (PMID:31206516). H297A still suppressed variegation but entirely failed to remove already-established ectopic heterochromatin, a separation of function read out as H3K9me at ectopic heterochromatin. Raiymbek et al. give the same kind of in vivo H3K9me readout for all three cofactor-site mutants ("In contrast, Epe1 mutants that exhibit a red or sectored phenotype upon +tetracycline addition retain high levels of H3K9 methylation at the ectopic site (Figure 1D)."; PMID:32195666), so Sorida's distinct contribution is the separation of function. Prevention is nearly but not fully retained (a few pink colonies), which Sorida et al. read as JmjC-dependent demethylation contributing "to full suppression to some extent". The removal arm is read at single-copy dose, where even wild-type Epe1 removes ectopic H3K9me incompletely and "We found that re-introduction of single copy Epe1 did not erase ectopic heterochromatin when an H3K9me source existed nearby, while Epe1 overexpression completely erased it" (Sorida), so it is conditional on dose. This is the kind of result the annotation rests on; it does not measure enzymatic demethylation, and H297A also reduces Epe1's heterochromatin localization. The experimental annotation is therefore not removed; it should be revisited with PomBase (evidence code, and whether a process term such as heterochromatin boundary formation or negative regulation of heterochromatin formation captures this result better).

Summary: This negative annotation (NOT|involved_in) captures the curator's intent that Epe1 opposes rather than promotes heterochromatin assembly, but it contradicts the seven accepted GO:0033696 heterochromatin boundary formation rows under the true-path rule, so it is left undecided.

Reason: The biology behind the NOT row is sound. Epe1 is an anti-silencing factor, and the cited paper (Ragunathan 2015) reports that Epe1 and Clr4 play opposing roles in maintaining silent H3K9me domains. The problem is logical, not biological. OLS lists GO:0031507 among the hierarchical ancestors of GO:0033696 heterochromatin boundary formation, and Epe1 carries seven ACCEPTed experimental GO:0033696 rows. By the true-path rule those rows entail GO:0031507, which this row denies, so the two cannot both stand as written. This review does not overrule the curator either way; the row is UNDECIDED, and the question of whether the NOT row or the placement of GO:0033696 should change is raised with PomBase in the suggested questions.

Summary: Correct annotation - Epe1 is essential for heterochromatin boundary formation and maintenance, preventing spreading of silent chromatin.

Reason: Multiple studies support this core function, which PomBase curated from PMID:25831549. The cached text of that paper mentions Epe1 only once, as opposing Clr4 in maintaining silent H3K9me domains, so the boundary-specific data cannot be checked there (see its reference_review). Other papers show the mechanism. Epe1 recruits Bdf2, whose bromodomains protect acetylated histone H4 from Sir2-mediated deacetylation (PMID:24013502), and Epe1 is preferentially recruited to the inverted repeats flanking centromeres to restrain the spread of pericentromeric heterochromatin (PMID:16762840).

Summary: Correct cellular component annotation - Epe1 localizes to heterochromatin through its interaction with HP1/Swi6 bound to H3K9me.

Reason: The cited paper reports that Epe1 is recruited to heterochromatic loci by Swi6/HP1 (its cached abstract does not state the method). The Epe1 C-terminus binds Swi6 directly, and H3K9 methylation stimulates this interaction (PMID:32195666). Localization is essential for its boundary function.

Summary: Epe1 localizes to pericentric heterochromatin where it regulates boundaries and enables repeat transcription for RNAi-mediated silencing.

Reason: The cited abstract reports that Epe1 is preferentially recruited to the inverted repeats flanking centromeres to restrain the spread of pericentromeric heterochromatin, and that it promotes Pol II accessibility at heterochromatic repeats. Its recruitment to heterochromatic loci depends on Swi6/HP1. The abstract does not state the method.

Summary: Epe1 localizes to mating-type region heterochromatin.

Reason: The PMID:16762840 abstract states that Epe1 is distributed across all major heterochromatic domains, which in fission yeast include the mating-type region; it does not name the mating-type locus specifically. Functional evidence at the mat locus exists. Ayoub et al. 2003 identified epe1 in a screen for silencing beyond the mat IR-L barrier, and conclude from epitype switching in a KΔ::ade6+ background and the meiotic stability of silencing beyond IR-L that Epe1 negatively controls heterochromatin stability (PMID:12773576); Trewick et al. 2007 show that loss of Epe1 causes variegated silencing there (PMID:17948055). No cached paper directly localizes Epe1 at the mat locus. The row rests on the curator's reading of the full paper.

Summary: Epe1 localizes to subtelomeric heterochromatin.

Reason: The PMID:16762840 abstract states that Epe1 is distributed across all major heterochromatic domains, which in fission yeast include the subtelomeres; it does not name subtelomeric heterochromatin specifically. Its Bdf2-dependent boundary function has been mapped at the IRC elements flanking centromeres, not at subtelomeres, so this row rests on localization only.

Summary: Epe1 is found at heterochromatin islands.

Reason: Zofall et al. 2012 report their own result placing Epe1 at heterochromatin islands (PMID:22144463). The curator's reference, PMID:16762840, is available here only as an abstract, which does not mention islands but places Epe1 "at certain meiotic genes"; since fission yeast heterochromatin islands are largely meiotic-gene domains, that clause is plausibly the curator's basis, but this cannot be confirmed from the cache.

Summary: Epe1 regulates facultative heterochromatin formation that can occur independently of the RNAi pathway, preventing excessive silencing.

Reason: The cited study shows that heterochromatin formation at meiotic loci can occur independently of RNAi, and that heterochromatin assembly at many such nucleation sites is suppressed by factors such as Epe1. This is consistent with its anti-silencing role.

Summary: Same claim as the IDA row from Audergon et al. (2015), with the EXP code. The full text contains genetic evidence (epe1 deletion and H297A/K314A point mutants allow H3K9me to persist at an ectopic site) but no direct demethylase assay.

Reason: As for the IDA row - an experimental annotation from a paper whose full text supports a JmjC-residue-dependent role for Epe1 in removing ectopic H3K9me, while direct catalysis remains disputed (non-canonical H297-E299-Y370 triad, no in vitro activity detected). UniProt's Y307A loss-of-function record (from Ayoub 2003, possibly the same experiment as Ayoub's domain-essential statement) likewise points to a JmjC-residue-dependent function without measuring catalysis; Raiymbek et al. place Y307 at the 2-oxoglutarate site and Sorida et al. describe Epe1Y307A as retaining the metal-binding residues, so it implicates the 2-oxoglutarate part of the cofactor pocket, and Y307A also loses co-localization with Swi6 (PMID:32195666). Sorida et al. 2019 independently show that H297A leaves the de novo arm intact but entirely fails to remove established ectopic heterochromatin (PMID:31206516), and Raiymbek et al. show that H297A, Y307A and Y370A retain ectopic H3K9me after the initiator is released ("In contrast, Epe1 mutants that exhibit a red or sectored phenotype upon +tetracycline addition retain high levels of H3K9 methylation at the ectopic site (Figure 1D)."), neither measuring demethylation. Not removed; flagged for discussion with PomBase alongside the IDA row.

Summary: Another correct annotation for heterochromatin boundary formation, a core Epe1 function demonstrated by multiple studies.

Reason: Sorida et al. 2019 show that Epe1 suppresses ectopic heterochromatin formation, and that the epe1Δ strain develops heterochromatin islands and ectopic heterochromatin. This boundary function is central to Epe1 biology.

Summary: Epe1 enables transcription of tandem repeats that generate RNAi substrates for heterochromatin formation, paradoxically promoting RNAi-mediated silencing.

Reason: Recent study shows Epe1 is required for efficient transcription of tandemly repeated genes that trigger RNAi-dependent heterochromatin. The authors propose that Epe1 generates enough transcripts to activate RNAi without disrupting heterochromatin, so that it reinforces silencing. This represents a regulatory feedback mechanism.

Summary: Ayoub et al. 2003 describe Epe1 as a conserved nuclear protein. This experimental row supersedes the IEA annotation.

Reason: The cached abstract of Ayoub et al. 2003 describes Epe1 as a conserved nuclear protein but does not state the method; the IDA code implies a direct localization assay in the full paper. This experimental row is stronger than the IEA prediction.

Summary: The original paper identifying Epe1 as a heterochromatin boundary factor. Foundational evidence for this core function.

Reason: This seminal paper first characterized Epe1 as modulating heterochromatization and preventing silencing spread. Demonstrated that Epe1 mutation affects position effect variegation and heterochromatin boundaries. This established the boundary function that has been confirmed by numerous subsequent studies.

Summary: Further evidence that Epe1 prevents unregulated heterochromatin assembly and maintains boundaries.

Reason: The paper demonstrates Epe1 prevents both unregulated assembly and disassembly of heterochromatin, maintaining proper boundaries. Shows Epe1 is required for heterochromatin homeostasis and boundary integrity. Core function with strong experimental support.

Summary: Epe1 localizes to and regulates heterochromatin islands, preventing their inappropriate formation in euchromatin.

Reason: Study shows Epe1 is present at heterochromatin islands and regulates their formation. In its absence, ectopic heterochromatin islands form inappropriately. This cellular component annotation accurately reflects Epe1 localization and function at these specialized chromatin structures.

Summary: Heterochromatin localization curated by PomBase from a 19S proteasome study; the localization itself is well supported by other papers.

Reason: This is a curator-made IDA annotation and is kept. The cached record for PMID:29214404 is abstract-only and does not mention Epe1, so the specific data the curator used cannot be checked here (see its reference_review). Epe1's heterochromatin localization is independently established, for example by Swi6-dependent recruitment of Epe1 to constitutive heterochromatin (PMID:17449867).

Summary: Another confirmation of heterochromatin localization, demonstrating Epe1 presence at silent chromatin domains.

Reason: Trewick et al. 2007 show GFP-tagged Epe1 colocalising with Swi6 at heterochromatin, dependent on Swi6, Clr4 and Rik1. Consistent and well-validated cellular component annotation.

Summary: Genetic interaction between epe1 and mst2 limits heterochromatin domain expansion.

Reason: Wang et al. 2015 find a strong negative genetic interaction between mst2 and epe1 deletions, and H3K9me2 domains expand at centromeres in mst2∆ epe1∆ swi6∆ cells even when boundary elements are present. The authors note that the interaction does not involve Epe1's Bdf2-dependent boundary activity, so the IGI supports a role in limiting heterochromatin spread rather than in the IRC boundary mechanism specifically.

Summary: Key paper showing Epe1 recruits Bdf2 bromodomain protein to establish heterochromatin boundaries at IRCs.

Reason: Wang 2013 demonstrates Epe1 recruits BET family protein Bdf2 to heterochromatin boundaries, particularly at inverted repeat centromeric (IRC) boundaries. Bdf2 recognizes acetylated H4 and antagonizes Sir2-mediated deacetylation, preventing heterochromatin spreading. Essential boundary mechanism.

Summary: Study of Epe1 interaction with heterochromatin assembly pathway confirms its heterochromatin localization.

Reason: Isaac et al. 2007 report that Swi6 recruits Epe1 to constitutive heterochromatin domains, and their expression data suggest that Epe1 interacts functionally with the heterochromatin assembly pathway at the histone deacetylation step.

Summary: Recombinant Epe1 binds Swi6/HP1 in vitro, and the interaction depends on the Swi6 chromoshadow domain and is stimulated by H3K9 methylation.

Reason: Raiymbek et al. 2020 show binding between recombinant Epe1 and Swi6 in vitro, reduced by the chromoshadow-domain mutation L315E, so the term for the bound domain fits. In cells the interaction requires H3K9 methylation (lost in clr4 deletion and H3K9R mutants), and H3K9me3 peptides stimulate it in vitro. GO:0140030 modification-dependent protein binding was used earlier but does not fit, because its definition requires the modification to be on the bound protein, and here the methylation is on histone H3, not on Swi6.

Summary: Overexpressed Epe1 co-purifies with the SAGA complex and co-immunoprecipitates with its histone acetyltransferase subunit Gcn5, probably through other SAGA subunits.

Reason: Bao et al. 2019 identified many SAGA components by mass spectrometry of affinity-purified overexpressed Epe1, with no SAGA subunits in control purifications, and confirmed by co-immunoprecipitation that overexpressed Epe1 interacts with Gcn5 (UniProtKB Q9UUK2, the supporting entity). The interaction is reduced in tra1 deletion cells, which the authors read as Tra1 mediating it, so Epe1 probably contacts the complex rather than Gcn5 directly. GO:0062070 SAGA complex binding describes that association; GO:0035035 histone acetyltransferase binding, used in an earlier draft of this row, would assert binding to the enzyme itself, which the data do not show. Every result was obtained with overexpressed Epe1, so whether endogenous Epe1 associates with SAGA is not established. The association is functionally relevant to the overexpression phenotype, which gcn5 deletion rescues, although the authors note that gcn5 deletion might instead act by reducing Epe1 recruitment to heterochromatin.

Summary: Epe1 binds the BET bromodomain protein Bdf2 and is required to recruit it to heterochromatin boundaries (IRCs).

Reason: Wang et al. 2013 show that Bdf2 and Epe1 interact by co-immunoprecipitation from endogenous loci and by yeast two-hybrid, and that Bdf2 is recruited to IRC boundaries in an Epe1-dependent manner. Epe1 itself is localized to heterochromatin through Swi6, so it brings Bdf2 into contact with boundary chromatin, which fits the adaptor definition. GO has no bromodomain-protein binding term, and bare protein binding is discouraged. The evidence code is IMP rather than IPI because the adaptor claim rests on the epe1 mutant phenotype (loss of Bdf2 recruitment to IRCs); the co-IP and two-hybrid data establish the binding half of the claim, but binding alone would only support a binding term.

Summary: Purified Epe1 binds histone H3 tails, preferring H3K9-methylated peptides and histones.

Reason: Raiymbek et al. 2020 show in in vitro binding assays that purified MBP-Epe1 preferentially associates with an H3K9me3 peptide over an unmodified H3K9 peptide, and selectively interacts with H3K9-methylated over H3K4-methylated histones. This is direct evidence that Epe1 binds histones. The more specific GO:0062072 H3K9me2/3 reader activity is not proposed here; see the knowledge gap on the nucleosome-turnover core function.