OTX2

Summary: OTX2 is a sequence-specific DNA-binding transcription factor that occupies chromatin, shown directly by promoter ChIP in human RPE cells and genome-wide by enhancer occupancy.

Reason: The TFClass-based inference is confirmed by direct human experimental data: OTX2 is recovered at an endogenous promoter by ChIP and, genome-wide, sits predominantly at distal enhancers. Chromatin is therefore a correct and informative location for this protein.

Summary: Phylogenetic inference that OTX2 binds RNA polymerase II cis-regulatory regions, in agreement with direct human ChIP and reporter evidence.

Reason: The IBD node placement for the bicoid/OTX clade is consistent with target-level experiments on the human protein, so the inherited assertion adds phylogenetic breadth to a function that is independently demonstrated. The presence of P32243 itself among the WITH/FROM sources reflects its own experimental annotation contributing to the node, which is expected.

Summary: Direct evidence that human OTX2 occupies an RNA polymerase II cis-regulatory region: ChIP at the endogenous DCT promoter in retinal pigment epithelial cells, with a mapped OTX2 site.

Reason: This is the most informative molecular-function-adjacent DNA-binding term available and it is supported by occupancy at a defined endogenous promoter element in human cells rather than by in vitro binding alone.

Summary: OTX2 is a homeodomain DNA-binding transcription factor acting on RNA polymerase II genes; the phylogenetic assertion matches direct human promoter-reporter data.

Reason: This direction-neutral term is the best statement of OTX2's core molecular activity, since OTX2 activates some targets (DCT, BEST1, CRX) and represses others depending on cofactors.

Summary: InterPro-based electronic assignment of RNA polymerase II-specific DNA-binding transcription factor activity from the homeodomain signature.

Reason: The domain-based inference is correct and is redundant with, rather than in conflict with, the manually curated IBA and ISA annotations for the same term.

Summary: TFClass-based assignment of OTX2 to the sequence-specific DNA-binding transcription factor class (bicoid-related homeodomain).

Reason: The classification is correct and consistent with the experimental activator assays; it supplies the same core molecular function from an orthogonal source.

Summary: OTX2 activates the human DCT promoter through its cognate binding site in an RPE-restricted manner, and antisense knockdown lowers DCT protein - direct transcriptional activator activity.

Reason: A promoter-reporter assay with a mapped OTX2 site, combined with loss-of-function in the native cell type, establishes activator activity rather than mere binding. This is one of the two human experiments that anchor the whole molecular-function block.

Summary: OTX2 acts as a transcriptional activator at the human BEST1 promoter, synergising with SOX9 and MITF.

Reason: A second, independent human RPE target with direct promoter-level activation. The combinatorial requirement is informative: OTX2's activator output at this promoter depends on partner factors, which is the general pattern for this transcription factor.

Summary: Generic DNA binding, inferred electronically from the homeodomain.

Reason: Correct but uninformative: it is an ancestor of the specific terms already annotated (GO:0000978 RNA polymerase II cis-regulatory region sequence-specific DNA binding and GO:1990837). Retained because it is not wrong, but it should not be read as a core function when the more precise children are present.

Summary: Generic DNA-binding transcription factor activity, inferred electronically from the homeodomain signature.

Reason: The direct parent of GO:0000981, which is annotated three times over from better sources. Accurate but adds no resolution; kept as non-core for the same reason as GO:0003677.

Summary: OTX2 physically interacts with SOX9 and MITF during synergistic activation of the BEST1 promoter.

Reason: The interaction is real and well evidenced, but 'protein binding' conveys nothing about what OTX2 does. The partners here are themselves transcription factors, and the functional content of the interaction is captured precisely by GO:0008134 transcription factor binding. Proposing that replacement in line with project guidance to avoid the uninformative parent.

Proposed replacements: transcription factor binding

Summary: OTX2 protein is localised to nuclei of human fetal retinal pigment epithelium.

Reason: Direct experimental localisation in human tissue, in the very cell type where OTX2's best-characterised human targets (DCT, BEST1) are regulated. The nucleus is where the core transcription-factor activity is exercised.

Summary: Phylogenetic inference that OTX2 is active in the nucleus.

Reason: Nuclear activity is the conserved state across the OTX/bicoid clade and is confirmed experimentally for the human protein; is_active_in is the appropriate qualifier for the compartment in which the DNA-binding activity is executed.

Summary: Nuclear localisation of OTX2 reported in a human neural-crest-derived dopaminergic differentiation protocol.

Reason: The location is correct and independently established. Noting for the record that OTX2 is used here as a lineage marker rather than being the subject of a localisation study, so this row adds little beyond the stronger EXP and IBA annotations for the same term.

Summary: Automated multi-method assignment of nuclear localisation.

Reason: Consistent with the experimental and phylogenetic annotations for the same compartment.

Summary: Immunofluorescence-based localisation to the nucleoplasm.

Reason: A legitimate refinement of the nuclear annotation and consistent with a chromatin-associated transcription factor; no conflict with the chromatin location, which describes the functionally engaged fraction. No supporting_text is attached, deliberately. The basis is the GO_REF:0000052 immunofluorescence record itself, which carries no quotable sentence, and the one citable human localisation text available ("OTX2 was localized to the nuclei of retinal pigment epithelium", PMID:19414065) reaches the parent GO:0005634 nucleus rather than the nucleoplasm refinement. Quoting it here would have attached a snippet that does not support the term it sits under.

Summary: Generic regulation of DNA-templated transcription, inferred from the homeodomain.

Reason: An ancestor of GO:0006357, which is annotated from phylogeny and matches the human promoter-level evidence. Correct but superseded in resolution.

Summary: OTX2 regulates RNA polymerase II transcription - the process counterpart of its core molecular function.

Reason: Direction-neutral and correct: OTX2 both activates (DCT, BEST1, CRX) and represses targets depending on context. This is the process term that should carry the core function.

Summary: Exogenous OTX2 protein applied in a gradient to Xenopus retinal explants weakly attracted both nasal and temporal growth cones, but did not produce the divergent turning response that constitutes guidance.

Reason: OTX2 was the specificity control in this study, not its subject, and the authors' explicit conclusion is that OTX2 does not elicit the opposite turning responses that define Engrailed-2's guidance activity. A weak, non-directional attraction was measured, so the term is not flatly false and REMOVE would overstate the case - but annotating human OTX2 to axon guidance on the strength of the experiment whose published result is the negative control inverts what the paper shows. Downgraded rather than deleted.

Summary: OTX2 binds eukaryotic initiation factor 4E. The activity is real and demonstrated for the human protein, but not by the reference cited here.

Reason: Direct GST pull-down shows human OTX2 binding eIF4E, placing it with the homeoproteins that engage the cap-binding machinery; the construct provenance confirms the human cDNA was used. The activity is genuine but peripheral to OTX2's nuclear transcription-factor role, so it is kept as non-core. The citation is the problem: PMID:15705863 is a medulloblastoma copy-number study that cannot support this claim. Recorded via reference_review on that entry and by supplying the canonical source here rather than by editing the GOA-supplied identifier.

Summary: Regulation of smoothened signalling, asserted as a traceable author statement from a medulloblastoma copy-number paper whose full text is not accessible.

Reason: The cited paper's abstract makes no statement about hedgehog or smoothened signalling, and the article is not open access (Europe PMC: isOpenAccess N, inEPMC N), so the intro/discussion in which a traceable statement would sit cannot be read. Project policy is not to overrule a curator on evidence that cannot be inspected, and there is a plausible indirect route (the OTX2 caudal border positions the isthmic organizer, where hedgehog and FGF signalling centres are established), but I found no human experiment establishing OTX2 regulation of smoothened signalling. Left UNDECIDED pending access to the full text; it should not be treated as core.

Summary: FITC-labelled OTX2 protein is internalised by retinal growth cones, consistent with the homeoprotein transfer behaviour OTX2 shares with Engrailed-2.

Reason: The localisation result for OTX2 is positive and was genuinely assayed in this study despite its Engrailed-2 title, so this is not a paralog mis-attribution. It reflects the non-cell-autonomous, extranuclear side of OTX2 biology - the same property that underlies its transfer into cortical parvalbumin interneurons - rather than its core nuclear function, and it rests on exogenously applied protein in a heterologous system.

Summary: OTX2 is required for forebrain development; its expression domain prefigures the forebrain and its loss deletes the structure entirely in mouse.

Reason: The term is unambiguously correct and is one of OTX2's two defining developmental roles. Flagging an evidence problem in the opposite direction from the usual one: the citation is a TAS from a copy-number study, whereas the real support is three independent mouse null alleles showing loss of the forebrain through failed anterior neuroectoderm specification, now complemented by human in vitro evidence that OTX2-positive anterior neural ectoderm is already lineage committed to forebrain/midbrain fate. The strongest function of this gene carries its weakest evidence code.

Summary: OTX2 is required for midbrain development, and its caudal expression boundary sets the midbrain-hindbrain organizer position.

Reason: Correct and central. Mouse nulls lose the midbrain along with the forebrain, and the mutual-repression relationship with GBX2 shows the OTX2 border is instructive for organizer placement: ectopic GBX2 shifts the OTX2 border rostrally and a new organizer forms there. As with GO:0030900, the cited TAS source is much weaker than the available evidence.

Summary: OTX2 is part of a protein-containing complex with SOX9 and MITF on the BEST1 promoter.

Reason: Accurate but maximally generic: it restates the SOX9/MITF co-activation result that is already captured with far more information by GO:0001228 and by the proposed GO:0008134. Retained because the interaction is genuine; not core.

Summary: Positive regulation of embryonic development, transferred by curator judgement from mouse Otx2 (UniProtKB:P80206).

Reason: The orthology assignment is sound and the mouse evidence is strong, but the term sits at a very low resolution for a gene whose null deletes the entire forebrain and midbrain. The informative content of this claim is carried by GO:0030900 and GO:0030901; keeping as non-core rather than removing a legitimate transfer.

Summary: Regulation of fibroblast growth factor receptor signalling. The cited reference cannot be checked, but an independent human mechanism for this claim exists.

Reason: Unlike the smoothened annotation from the same TAS block, this one has concrete support elsewhere: patient-iPSC organoids show that hypothalamic OTX2 controls FGF10 expression, which induces LHX3 in oral ectoderm and sustains pituitary progenitors, with CRISPR correction reversing the phenotype. OTX2 therefore does regulate FGF-receptor signalling, but indirectly, by transcriptional control of the ligand in an adjacent tissue - hence non-core. The cited source remains flagged as unverifiable in its reference_review.

Summary: Positive regulation of DNA-templated transcription, transferred from mouse Otx2.

Reason: Correct in direction but an ancestor of GO:0045944, which is annotated twice by IDA from human promoter assays. Kept as a valid but redundant transfer.

Summary: OTX2 positively regulates RNA polymerase II transcription of DCT through a defined promoter element in human RPE cells.

Reason: Direct human evidence combining promoter occupancy, reporter activation through a mapped site, and loss of target protein on knockdown. This is the process counterpart of the GO:0001228 activator activity from the same study.

Summary: OTX2 positively regulates BEST1 transcription in synergy with SOX9 and MITF.

Reason: A second independent human target supporting the same positive-regulation process, and one that illustrates the combinatorial nature of OTX2's activator output.

Summary: Protein-containing complex assembly, inferred from the demonstration that OTX2, SOX9 and MITF interact and synergise at the BEST1 promoter.

Reason: Showing that OTX2 is present in a co-activating complex is not the same as showing OTX2 drives assembly of that complex. The experiments establish interaction and functional synergy; no assembly kinetics, order-of-addition, or OTX2-dependence of complex formation is demonstrated. The underlying result is already represented by GO:0032991 and GO:0001228, so this row adds an unsupported causal claim on top of evidence that is otherwise well used.

Summary: Otx2 acts combinatorially with Lmx1a and FoxA2 to enhance midbrain dopaminergic neuron differentiation from embryonic stem cells.

Reason: A genuine genetic-interaction result in which OTX2 is one of the three co-expressed factors tested, so the IGI code is used correctly. Biologically this is downstream of OTX2's midbrain patterning role - midbrain dopaminergic neurons arise within the territory OTX2 specifies - rather than a separate core activity, and the evidence is from mouse ESCs.

Summary: Dopaminergic neuron differentiation, asserted from a review of Wnt signalling in midbrain dopaminergic development.

Reason: Review-derived TAS that duplicates the term already supported by the primary IGI from PMID:19951692. Not incorrect, but it contributes no independent evidence and inherits the same non-core standing.

Summary: Primitive streak formation, transferred by curator judgement from mouse Otx2.

Reason: The mouse basis is real - one null allele produces severe gastrulation and axial-mesoderm defects, placing the Otx2 requirement before neural plate stages. I would not have chosen this term myself, since OTX2's documented gastrulation-stage function is anterior specification in epiblast and visceral endoderm rather than streak formation as such, but the curator transferred from an experimentally grounded mouse annotation whose full text I have not read, so it is kept rather than challenged. Not core.

Summary: Methyl-SELEX across 542 human transcription factors measures OTX2 binding to double-stranded sequence-specific sites and places it among the extended homeodomain factors that prefer CpG-methylated sequences.

Reason: A careful in vitro binding measurement, but a more general term than GO:0000978, which already records binding at RNA polymerase II cis-regulatory regions with in vivo support. The genuinely novel content of this study - the preference for methylated CpG within the binding site - is not captured by any term currently annotated; raised as a question rather than asserted, since the in vivo consequence for OTX2 specifically is not established here.

Summary: Positive regulation of gastrulation, transferred by curator judgement from mouse Otx2.

Reason: Supported by the mouse null phenotype, in which the requirement for Otx2 is already evident at gastrulation stages. Legitimate as a transfer, but a coarse description of a gene whose specific gastrulation-stage job is to specify anterior neuroectoderm; that content lives in the forebrain/midbrain terms.

Summary: Proposed new annotation. The caudal border of the OTX2 expression domain is where the midbrain-hindbrain (isthmic) organizer forms, and the border is positionally instructive rather than a readout: transiently expressing GBX2 inside the caudal OTX2 domain shifts that border rostrally and a normal-appearing organizer assembles at the new position. Human GOA records the anatomical outcomes on either side of the boundary (forebrain and midbrain development) but nothing about the boundary itself. Transferred by similarity from zebrafish Otx2, which carries the term, and at that gene's own strength: acts_upstream_of_or_within, not involved_in.

Reason: Proposed after working the project's NEW checks, including one that overturned an earlier candidate. (1) What OTX2 contributes is the position of the boundary, not the boundary itself: the organizer forms at the OTX2 caudal border and relocates when that border is moved, and in zebrafish Otx morphants the presumptive mesencephalon becomes an extended rhombomere 1. The structure is built by the isthmic organizer genes; OTX2 sets where. That is what acts_upstream_of_or_within says, and it is why this row does not claim involved_in - per the project's NEW rule, evidence that a gene is necessary for a process is not evidence that it performs a step of it. (2) The comparator check supports the term. Queried against QuickGO: GO:0030917 has 8 experimental annotations in mouse (Gbx2, Fgf8, Wnt1, En1, Lrp6 by IMP) and 138 in zebrafish, and it is annotated to the OTX2 orthologues themselves - zebrafish otx2 (UniProtKB:Q91981, Swiss-Prot), otx2a and otx2b all carry it by IGI from PMID:16611693. So this is not a term GO withholds from the OTX-class selector; human OTX2 simply has no annotation for it, as is the case for the other human regional selectors whose perturbation experiments were done in mouse and fish. The ISS therefore transfers from Q91981, the reviewed zebrafish Otx2 entry. An earlier revision of this row cited UniProtKB:P80206 (mouse Otx2) in supporting_entities, which was wrong: mouse Otx2 returns zero hits for GO:0030917, so the WITH pointed at a gene that does not carry the term. The mouse papers remain the mechanistic grounding in supported_by; they are not the similarity donor. Note that original_reference_id here is a PMID where OTX2's four existing ISS rows use GO_REF:0000024 (manual transfer by curator judgment of sequence similarity). The departure is deliberate, not an oversight: a GO_REF names the method, which is the right citation for a transfer a curator has already made, whereas this row is a proposal, and the useful thing to show a curator weighing it is the paper that put the term on the donor in the first place. (3) The qualifier follows the donor rather than the surrounding IEA rows. Every one of the 146 experimental GO:0030917 annotations in mouse and zebrafish uses acts_upstream_of_or_within; the involved_in rows on this term are all IEA/ARBA electronic inferences. An ISS must not strengthen the assertion it transfers, and reading the donor's acts_upstream_of_or_within as involved_in would do exactly that. (4) Redundancy check: GO:0030917 is neither an ancestor nor a descendant of any term OTX2 already carries - checked against all 37 of its annotations in the part_of-inclusive closure, which returns only the self-match. GO:0030900 and GO:0030901 are siblings of it under GO:0007420 brain development, not ancestors, and OTX2 does not carry GO:0007420 itself. Two alternatives were rejected by the same checks. GO:0009952 anterior/posterior pattern specification looked like the obvious choice, and mouse Otx2 does carry it by IMP and IGI, but OTX2 already carries GO:0090009 primitive streak formation by ISS, and GO:0090009 sits under GO:0009952 by part_of - so the term would have added an ancestor of an existing annotation rather than new coverage. Stating that relation precisely, because it is the single edge the whole withdrawal rests on and an earlier wording of this reason called it a descendant without qualification: in the GO 2026-07-26 release GO:0090009 has exactly one is_a parent, GO:0048646 anatomical structure formation involved in morphogenesis, and its is_a closure contains no term in the regionalization branch. GO:0009952 is reached only through BFO:0000050 part_of (and RO:0002131 overlaps). That is still decisive, because GO annotation propagation follows part_of as well as is_a, so an annotation to GO:0090009 already entails GO:0009952; the withdrawal holds, but on a part_of edge rather than an is_a one. Verified against the local oaklib sqlite:obo:go entailed_edge closure, the same source the module validator resolves labels through. GO:0021999 neural plate anterior/posterior regionalization, the term the triggering paper is nominally about, was rejected because its annotations are CELSR2, NOG and SSBP3 in mammals and zebrafish axin1, bptf, smad2 and gsc experimentally - morphogen and chromatin machinery, not regional selector transcription factors. GO:0021978 telencephalon regionalization, which mouse Otx2 carries by IMP, reaches GO:0030900 forebrain development - which human OTX2 already carries - by BFO:0000050 part_of, and was rejected for the same redundancy reason as GO:0009952. It is not under GO:0009952 at all.