PAX4 is a vertebrate paired-box transcription factor of the Pax4/6 class, encoded on chromosome 7q32. It binds DNA through an N-terminal paired domain (PAI and RED subdomains) and a paired-type homeodomain, and it acts mainly as a sequence-specific transcriptional repressor of RNA polymerase II genes. Its repression of the insulin and glucagon promoters and of its own promoter has been shown in human reporter assays, and it also has a weaker, context-dependent activating potential. In human beta-cell models it occupies more than a thousand genomic sites, including the insulin promoter and regulatory regions of islet developmental genes. PAX4 is expressed transiently in NEUROG3-derived endocrine progenitors of the developing pancreas and in intestinal enteroendocrine progenitors, and at low levels in adult islets and gut mucosa. In the pancreas it biases endocrine progenitors toward the beta-cell fate and keeps alpha-cell genes such as ARX and GCG repressed. Unlike in mouse, it is not essential for human beta-cell formation. PAX4-null human stem-cell-derived islets still form but contain polyhormonal cells with less insulin. Children with complete PAX4 loss have transient neonatal diabetes that remits and later relapses. In the human small intestine PAX4 is the effector through which ZNF800 restrains serotonin-producing enterochromaffin cells: PAX4 promotes enterochromaffin differentiation and suppresses ARX-dependent enteroendocrine subtypes such as L cells. Heterozygous coding variants (for example rs2233580, reported as R192H) raise type 2 diabetes risk in East Asians. Its earlier designation as the cause of MODY9 is not supported by recent population-genetic analyses. An alternatively spliced isoform found in insulinoma antagonises the full-length repressor.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0000122 negative regulation of transcription by RNA polymerase II | IEA GO_REF:0000108 | ACCEPT | Summary: Inferred from the IDA repressor-activity annotation (GO:0001227) by the GO inter-ontology link between repressor activity and negative regulation of Pol II transcription. Reason: The inference is sound and the process is the core transcriptional output of PAX4. Human PAX4 represses the insulin and glucagon promoters in reporter assays (a MODY-associated variant lost this activity), represses its own promoter through high-affinity sites, and knockdown in the human beta-cell line EndoC-betaH1 derepresses GCG. This matches the mouse review, where the same term is accepted on IDA. Supporting Evidence: PMID:17426099 R164W segregated with diabetes in the family of the proband and in vitro studies showed that it impairs the repressor activity of PAX4 on the insulin and glucagon promoters PMID:10967107 PAX4 itself has at least two high affinity binding sites within the promoter through which it exerts a strong negative autoregulatory effect PMID:37777536 loss of PAX4 increased GCG transcript by 2.7-fold in shPAX4 EndoC-βH1 cells (Fig. 2h), consistent with PAX4 being a repressor of GCG expression |
| GO:0000785 chromatin | ISA GO_REF:0000113 | ACCEPT | Summary: TFClass-based assignment of a sequence-specific DNA-binding transcription factor to chromatin. Reason: PAX4 is a sequence-specific DNA-binding factor, and human data now show chromatin occupancy directly. CUT&RUN of tagged PAX4 in EndoC-betaH1 cells identified about 1,700 bound regions, many in accessible chromatin of beta cells and pancreatic progenitors. Chromatin is where the core activity acts. Supporting Evidence: PMID:40614820 Analysis using SEACR [30] identified 1,673 binding peaks PMID:40614820 Of the 1,673 CUT&RUN peaks, 612 overlapped with ATAC-seq peaks in these datasets |
| GO:0000785 chromatin | ISS GO_REF:0000024 | ACCEPT | Summary: Curator transfer of chromatin location from mouse Pax4 (MGI:97488). Reason: Consistent with the ISA row and with direct human CUT&RUN occupancy data. The mouse review likewise places the core repressor activity on chromatin. Supporting Evidence: PMID:40614820 The binding of PAX4 to the INS gene promoter is direct evidence for the importance of the gene in insulin secretion |
| GO:0000978 RNA polymerase II cis-regulatory region sequence-specific DNA binding | IBA GO_REF:0000033 | ACCEPT | Summary: Sequence-specific binding of Pol II cis-regulatory regions, inherited from the PAX family root node PTN002552508, whose seeds include mouse Pax4 (MGI:97488). Reason: Sequence-specific DNA binding through the paired domain and homeodomain is the ancestral PAX activity, and PAX4 keeps both DNA-binding domains intact. In human cells PAX4 binds the insulin promoter (CUT&RUN) and high-affinity sites in its own promoter, both Pol II cis-regulatory regions. This is the term to which the broader DNA-binding rows below are refined, as in the mouse review. Supporting Evidence: PMID:40614820 The binding of PAX4 to the INS gene promoter is direct evidence for the importance of the gene in insulin secretion PMID:10967107 PAX4 itself has at least two high affinity binding sites within the promoter through which it exerts a strong negative autoregulatory effect PMID:10567552 identified a bipartite consensus DNA binding sequence consisting of a homeodomain binding site separated from a paired domain binding site by 15 nucleotides |
| GO:0000981 DNA-binding transcription factor activity, RNA polymerase II-specific | IBA GO_REF:0000033 | ACCEPT | Summary: Pol II-specific DNA-binding transcription factor activity inherited from the PAX root node. Human PAX4 (UniProtKB:O43316) is itself among the seeds. Reason: The ancestral PAX activity applies. This sign-neutral term is kept as the umbrella, as in the mouse review, because PAX4 represses most documented targets (GO:0001227, accepted) but has also been reported to transactivate the Bcl-xL and c-myc promoters. Human PAX4 among the seeds marks experimental grounding on the target, not circularity. Supporting Evidence: PMID:15596543 Pax4 transactivated the Bcl-xL and c-myc promoters, whereas its diabetes-linked mutant was less efficient PMID:17426099 R164W segregated with diabetes in the family of the proband and in vitro studies showed that it impairs the repressor activity of PAX4 on the insulin and glucagon promoters |
| GO:0000981 DNA-binding transcription factor activity, RNA polymerase II-specific | IEA GO_REF:0000002 | ACCEPT | Summary: InterPro2GO mapping from the homeobox conserved site. Reason: Correct for a paired-type homeodomain protein with demonstrated regulation of Pol II promoters, and kept as the sign-neutral umbrella, as for the IBA row. Supporting Evidence: file:human/PAX4/PAX4-uniprot.txt Mainly acts as a transcription repressor that binds to a common element in the glucagon, insulin and somatostatin promoters and inhibits their expression PMID:15596543 Pax4 transactivated the Bcl-xL and c-myc promoters |
| GO:0000981 DNA-binding transcription factor activity, RNA polymerase II-specific | ISA GO_REF:0000113 | ACCEPT | Summary: TFClass assignment of PAX4 (paired-domain class 3.2.1) as a DNA-binding transcription factor. Reason: Correct and consistent with the IBA and IEA rows on the same term. The more specific repressor activity is accepted separately. Supporting Evidence: PMID:10567552 PAX4 can function as a transcriptional repressor PMID:15596543 Pax4 transactivated the Bcl-xL and c-myc promoters |
| GO:0001227 DNA-binding transcription repressor activity, RNA polymerase II-specific | IDA PMID:37883554 Unbiased transcription factor CRISPR screen identifies ZNF80... | ACCEPT | Summary: From the ZNF800 CRISPR study in adult human small-intestinal organoids. PAX4 is the effector of ZNF800 and suppresses ARX-dependent enteroendocrine lineages. Reason: The cached paper is abstract-only. UniProt's curated FUNCTION text, tied to this paper, says PAX4 represses ARX in this system, and the authors' commentary describes a unilateral inhibition of ARX by PAX4. Neither source shows PAX4 binding at ARX, so the curator's full-text reading is deferred to. The activity itself is independently established for the human protein. PAX4 represses the insulin and glucagon promoters in reporter assays and its own promoter through high-affinity binding sites, and its loss derepresses GCG in human beta cells. This is the defining molecular activity of PAX4 and matches the mouse IDA rows. Supporting Evidence: file:human/PAX4/PAX4-uniprot.txt Prevents differentiation of other intestinal enteroendocrine cell types by repressing the expression of ARX, the master transcription factor for intestinal type-L, -N, -D, -I and -G enteroendocrine cell differentiation PMID:38279896 Furthermore, we unveiled a unilateral inhibition of PAX4 towards ARX PMID:10967107 PAX4 itself has at least two high affinity binding sites within the promoter through which it exerts a strong negative autoregulatory effect PMID:17426099 in vitro studies showed that it impairs the repressor activity of PAX4 on the insulin and glucagon promoters |
| GO:0003309 type B pancreatic cell differentiation | IBA GO_REF:0000033 | ACCEPT | Summary: Beta-cell differentiation inherited from the Euteleostomi node PTN002900580, seeded by mouse Pax4, mouse and rat Pax6 and zebrafish pax6b. Reason: The node placement is sound here, as in the mouse review. Both paralogs below the node have beta-cell evidence, so the term holds for PAX4 whichever way the PAX4/PAX6 split is drawn. Human data now support it directly but differ from mouse in degree. PAX4-null hiPSC-derived islets still form beta-like cells, but they show derepressed alpha-cell genes, polyhormonal cells and low insulin. Re-analysis of the knockout progenitors shows ARX up-regulation and loss of beta-cell gene expression. Children with complete PAX4 loss have transient, not permanent, neonatal diabetes. So human PAX4 biases progenitors toward the beta-cell fate without being essential for beta-cell formation. Participation is still met: PAX4 is a DNA-binding repressor acting in the progenitors whose fate it sets. Bgee calls PAX4 present in islet of Langerhans but absent in its type B pancreatic cell samples (gold quality), where PAX6 scores 99. That absence is the developmental-term artifact: Bgee's calls merge all stages, and PAX4 expression is highest in the embryonic pancreas and markedly reduced after birth (PMID:42210309), so these calls cannot test a differentiation role. They did not affect the action. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002900580 · PTN002900580 SUPPORTS TRANSFER Euteleostomi speciation node. Pax4 and Pax6 below it both have experimental beta-cell evidence, and human PAX4 has its own stem-cell and patient data. MGI:MGI:97488 · mouse Pax4 SUPPORTS TRANSFER Mouse Pax4 knockout (PMID:9121556) is the ortholog seed. In human the role is retained but not essential. Supporting Evidence: PMID:40614820 human PAX4 expression is important for endocrine cell development and has a key role in directing progenitor cells toward a beta cell fate PMID:37777536 we found that PAX4 deficiency and/or loss-of-function to result in derepression of alpha cell genes, leading to the formation of polyhormonal endocrine cells in vitro, with reduced total insulin content PMID:40614820 demonstrating that in contrast to mouse models, PAX4 is not essential for the development of human pancreatic beta-cells PMID:9121556 We propose that the early expression of Pax4 in a subset of endocrine progenitors is essential for the differentiation of the beta and delta cell lineages. PMID:42210309 Pax4 is highly expressed in embryonic pancreas and its expression is markedly reduced after birth file:human/PAX4/PAX4-bioinformatics/RESULTS.md | GO:0003309 type B pancreatic cell differentiation | type B pancreatic cell | CONSISTENT (UBERON:0000006 islet of Langerhans (50.6)) | ENRICHED (CL:0000169 type B pancreatic cell (99.3)) | file:human/PAX4/PAX4-bioinformatics/RESULTS.md type B pancreatic cell absent (gold, 49); body of pancreas absent (gold, 30); islet of Langerhans present (gold, 51) |
| GO:0003309 type B pancreatic cell differentiation | ISS GO_REF:0000024 | ACCEPT | Summary: Curator transfer from mouse Pax4 (P32115), whose knockout lacks beta and delta cells. Reason: The transfer is supported by direct human evidence (hiPSC knockout, risk alleles corrected by gene editing, and complete-loss patients). The human role is contributory rather than essential, which the description and core functions state. The row is consistent with the IBA row. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB:P32115 · mouse Pax4 SUPPORTS TRANSFER The mouse loss- and gain-of-function data are sound. Human data confirm the role but show it is not essential. Supporting Evidence: PMID:37777536 unlike in mouse, PAX4 is not essential for human beta cell differentiation in vitro PMID:40614820 human PAX4 expression is important for endocrine cell development and has a key role in directing progenitor cells toward a beta cell fate PMID:25483960 it did reduce the number of glucagon-positive cells and increased the number of unihormonal insulin-positive cells |
| GO:0003677 DNA binding | IEA GO_REF:0000002 | MODIFY | Summary: InterPro2GO mapping from the homeodomain, paired domain and PAX family signatures. Reason: Correct but general. The paired domain and homeodomain bind specific Pol II cis-regulatory sequences (insulin promoter, the PAX4 promoter), and the gene already carries GO:0000978 by IBA. This is the same refinement as in the mouse review. Proposed replacements: RNA polymerase II cis-regulatory region sequence-specific DNA binding Supporting Evidence: file:human/PAX4/PAX4-uniprot.txt Mainly acts as a transcription repressor that binds to a common element in the glucagon, insulin and somatostatin promoters and inhibits their expression PMID:40614820 The binding of PAX4 to the INS gene promoter is direct evidence for the importance of the gene in insulin secretion |
| GO:0003677 DNA binding | TAS PMID:9480859 Isolation of full-length cDNA of mouse PAX4 gene and identif... | MODIFY | Summary: TAS from the 1998 paper that cloned full-length mouse Pax4 and identified the human homologue, noting its paired domain and paired-type homeodomain. Reason: The DNA-binding claim rests on domain content in a cloning paper. It is correct but general. The specific, experimentally supported activity is sequence-specific binding of Pol II cis-regulatory regions, already carried by IBA. Proposed replacements: RNA polymerase II cis-regulatory region sequence-specific DNA binding Supporting Evidence: PMID:9480859 Two DNA binding motifs, a 128-amino acid paired domain and a 61-amino acid paired-type homeodomain exhibit the highest amino acid homology with PAX6 PMID:10967107 PAX4 itself has at least two high affinity binding sites within the promoter |
| GO:0003690 double-stranded DNA binding | IEA GO_REF:0000117 | MODIFY | Summary: ARBA machine-learning prediction of double-stranded DNA binding. Reason: Correct but general. GO:0000978 is a descendant of GO:0003690, is already carried by IBA, and is supported by human promoter-binding data. This matches the mouse refinement of its double-stranded DNA binding rows. Proposed replacements: RNA polymerase II cis-regulatory region sequence-specific DNA binding Supporting Evidence: PMID:40614820 The binding of PAX4 to the INS gene promoter is direct evidence for the importance of the gene in insulin secretion |
| GO:0005634 nucleus | IBA GO_REF:0000033 | ACCEPT | Summary: Nuclear activity inherited from the PAX family root node. Reason: PAX4 is a DNA-binding repressor of Pol II genes. Tagged human PAX4 is nuclear in every overexpression system tested (EndoC-betaH1, MIN6, AD293, HEK293). No study has localized the endogenous protein, because ChIP-grade antibodies are lacking. Supporting Evidence: PMID:40614820 Western blotting and fluorescent microscope images confirmed nuclear localization and the overexpression of PAX4 in the EndoC-βH1 cells PMID:41475885 The results showed that WT PAX4 was uniformly distributed in the nucleus |
| GO:0005634 nucleus | IEA GO_REF:0000044 | ACCEPT | Summary: UniProt subcellular-location mapping to nucleus. Reason: Consistent with the IBA row and with direct observation of tagged human PAX4 in the nucleus. Supporting Evidence: PMID:37777536 Overexpression of the PAX4 variants in AD293 cells confirmed that the p.His192 allele does not prevent nuclear localization PMID:33009794 The aberrant PAX4 variants were shown to retain stability and nuclear localization. |
| GO:0005654 nucleoplasm | TAS Reactome:R-HSA-210886 | ACCEPT | Summary: Reactome places the PAX4 protein product in the nucleoplasm in its NEUROG3-dependent synthesis reaction. Reason: Nucleoplasm is Reactome's compartment for the soluble pool of a nuclear protein, and it is consistent with the observed nuclear localization of human PAX4. GO defines nucleoplasm as the nuclear content other than the chromosomes and nucleolus, so the two are sibling locations under nucleus, not nested ones. Chromatin, where the repressor activity acts, is carried separately by its own accepted row. Supporting Evidence: Reactome:R-HSA-210886 the protein product is transported to the nucleus PMID:41475885 The results showed that WT PAX4 was uniformly distributed in the nucleus |
| GO:0005694 chromosome | IEA GO_REF:0000044 | MODIFY | Summary: UniProt subcellular-location mapping to chromosome, itself by similarity to mouse. Reason: UniProt's 'Chromosome' location is an inference from DNA binding. For a sequence-specific transcription factor the precise component is chromatin (GO:0000785, part_of chromosome), which the gene already carries by ISA and ISS and which human CUT&RUN supports. This is the same refinement as in the mouse review. Proposed replacements: chromatin Supporting Evidence: file:human/PAX4/PAX4-uniprot.txt SUBCELLULAR LOCATION: Nucleus {ECO:0000250|UniProtKB:P32115}. Chromosome {ECO:0000250|UniProtKB:P32115} PMID:40614820 Analysis using SEACR [30] identified 1,673 binding peaks |
| GO:0006355 regulation of DNA-templated transcription | IEA GO_REF:0000002 | MODIFY | Summary: InterPro2GO mapping to regulation of transcription. Reason: Correct but general. PAX4 regulates RNA polymerase II genes, predominantly by repression, and the gene carries GO:0000122 (accepted). The sign-neutral Pol II umbrella GO:0006357 is retained on its own IBA row. This matches the mouse review. Proposed replacements: negative regulation of transcription by RNA polymerase II Supporting Evidence: PMID:17426099 in vitro studies showed that it impairs the repressor activity of PAX4 on the insulin and glucagon promoters |
| GO:0006357 regulation of transcription by RNA polymerase II | IBA GO_REF:0000033 | ACCEPT | Summary: Regulation of Pol II transcription, inherited from the PAX family root node. Reason: Correct ancestral process, kept as the sign-neutral umbrella, as in the mouse review. PAX4 mostly represses (GO:0000122) but has been reported to transactivate Bcl-xL and c-myc promoters, so the direction depends on target and context. Supporting Evidence: PMID:15596543 Pax4 transactivated the Bcl-xL and c-myc promoters, whereas its diabetes-linked mutant was less efficient PMID:10967107 PAX4 itself has at least two high affinity binding sites within the promoter through which it exerts a strong negative autoregulatory effect |
| GO:0007420 brain development | IBA GO_REF:0000033 | REMOVE | Summary: Brain development, inherited from the Bilateria node PTN008693041. Its seeds are Drosophila eyeless and twin of eyeless and vertebrate Pax6 genes (mouse, rat, zebrafish pax6a and pax6b). No PAX4 gene is a seed. Reason: PAX4 descends from PTN008693041 under any topology, so the question is whether the PAX4 lineage kept the ancestral neural developmental role. The evidence says it diverged, and nothing says it kept it. (1) No study of any PAX4 ortholog reports a brain phenotype or function. In human, two children with complete PAX4 loss, followed into childhood, had no reported extra-pancreatic features. This is evidence on the target species itself, which the mouse review could only borrow. (2) Pax4 and Pax6 are not interchangeable where both are expressed, and comparative work describes lineage-specific Pax4 roles against a conserved Pax6 program. (3) Human Bgee data differ from mouse. PAX4 has low fetal present calls in ganglionic eminence (RNA-seq, fourth and fifth lunar months, scores about 42-48) and ventricular zone, which give a CONSISTENT tier. All stage-merged forebrain calls are absent where PAX6 is present. Low fetal presence is expression, not participation (projects/BGEEDB_EXPRESSION.md rules 1-2), so it does not change the decision. `just paint-losses` finds no IRD on the lineage, so the cached IBA is current. This is an over-propagated IBA, and the same action is taken in the mouse review. An IRD on the mammalian PAX4 clade PTN000676639 is suggested (see suggested_questions). Propagation Review Root cause: PROPAGATION BAD Failure modes: FUNCTIONAL DIVERGENCE Sources checked: PANTHER:PTN008693041 · PTN008693041 SUPPORTS SOURCE BUT NOT TARGET Bilateria speciation node seeded by Drosophila ey and toy and vertebrate Pax6 genes. Human PAX4 descends from it through the mammalian PAX4 clade PTN000676639, which has no brain evidence. MGI:MGI:97490 · mouse Pax6 SUPPORTS SOURCE BUT NOT TARGET Pax6 brain development is well established. Pax4 cannot substitute for Pax6 (PMID:10364449). Supporting Evidence: PMID:40614820 No extra pancreatic features were reported in either patient. PMID:10364449 The gene-targeting experiments revealed that Pax-4 and Pax-6 cannot substitute for each other in tissue with overlapping expression of both genes. PMID:24951566 This indicates evolutionarily conserved roles of Pax6 and lineage-specific differences in the roles of Pax4 during pancreas development among vertebrates. file:human/PAX4/PAX4-bioinformatics/RESULTS.md | GO:0007420 brain development | brain | CONSISTENT (UBERON:0004023 ganglionic eminence (45.2)) | ENRICHED (UBERON:0003053 ventricular zone (99.3)) | file:human/PAX4/PAX4-deep-research-claude-code.md The existing IBA annotations to brain, forebrain, retina and sensory organ development probably reflect PAX6-clade biology and deserve scrutiny for PAX4. |
| GO:0007423 sensory organ development | IBA GO_REF:0000033 | MARK AS OVER ANNOTATED | Summary: Sensory organ development, inherited from the Bilateria node PTN008693041. The only Pax4 among its seeds is rat Pax4 (RGD:620433), whose relevant annotation is an IEP for retina development (PMID:19012751). Reason: Apart from rat Pax4, the seeds resolved in the mouse review are Pax6-lineage genes and zebrafish pax10. The one Pax4 seed is an expression pattern: rat Pax4 mRNA appears in photoreceptors only after birth, and the authors themselves argue against a developmental role. No eye or other sense-organ phenotype has been reported for any PAX4 ortholog, and comparative work states that mammalian and teleost Pax4 genes are not implicated in eye development. Human evidence adds two surviving complete-loss patients with no reported extra-pancreatic features. In Bgee, PAX4 is absent from the sampled sense-organ tissues (stage-merged, so consistency only). The row is marked rather than removed, as in mouse, because the clinical report describes no eye examination and a late photoreceptor-maturation role has never been tested. The claim exceeds the evidence but is not contradicted. Propagation Review Root cause: PROPAGATION BAD Failure modes: FUNCTIONAL DIVERGENCE SOURCE EVIDENCE WEAK Sources checked: PANTHER:PTN008693041 · PTN008693041 SUPPORTS SOURCE BUT NOT TARGET Bilateria node. The sense-organ role is carried by the Pax6 and Pax10 lineages, and no functional evidence places it in the PAX4 lineage. RGD:620433 · rat Pax4 SOURCE WEAK OR INFERRED The seed annotation is an IEP (postnatal photoreceptor expression, PMID:19012751), whose authors argue against a developmental role. UniProtKB:P26367 · human PAX6 SUPPORTS SOURCE BUT NOT TARGET The PAX6 eye and sense-organ program is the source biology and does not transfer to its repressor paralog. Supporting Evidence: PMID:24951566 The Pax4 genes of mammals and teleost fishes are not implicated in the development of eyes PMID:19012751 The finding that Pax4 expression is highest in the adult argues against a role in development, as is the case with Pax6. PMID:40614820 No extra pancreatic features were reported in either patient. file:human/PAX4/PAX4-bioinformatics/RESULTS.md | GO:0007423 sensory organ development | sense organ | ABSENT_DEVELOPMENTAL (absent in UBERON:0005384 nasal cavity epithelium, UBERON:0006919 tongue squamous epithelium) | ENRICHED (UBERON:0001812 palpebral conjunctiva (99.5)) | |
| GO:0009887 animal organ morphogenesis | TAS PMID:9480859 Isolation of full-length cDNA of mouse PAX4 gene and identif... | MARK AS OVER ANNOTATED | Summary: TAS from the 1998 cDNA cloning paper, which frames PAX4 as a regulator of islet endocrine progenitor fate. Reason: The cloning paper reports sequence, gene structure and chromosomal location, and cites PAX4's role in progenitor fate. It does not show organ morphogenesis. PAX4's documented role is allocation of endocrine cell fates within the pancreatic and intestinal epithelium (GO:0003309, GO:7770020), not generation of organ shape. GO:0009887 is not an ancestor of those terms, so TOO_GENERAL does not apply. The same action is taken on the mouse TAS row for this term. Supporting Evidence: PMID:9480859 PAX4, a member of the paired box (PAX) gene family, is involved in the mechanism regulating the fate of pancreatic islet endocrine progenitor cells PMID:40614820 human PAX4 expression is important for endocrine cell development and has a key role in directing progenitor cells toward a beta cell fate |
| GO:0030900 forebrain development | IBA GO_REF:0000033 | REMOVE | Summary: Forebrain development, inherited from the Euteleostomi node PTN002900580. Its seeds are mouse and rat Pax6 and zebrafish pax6a and pax6b only. Reason: Every seed is a Pax6 ortholog. Human PAX4 inherits the term only because PANTHER 19 draws the PAX4/PAX6 split below this node twice, at a Mammalia duplication (PTN000676674, above the human PAX4 clade PTN000676639) and separately at a Neopterygii duplication for fish pax4. This is contrary to the single 2R origin of Pax4, Pax6 and Pax10 shown by synteny. In a species-reconciled tree the PAX4 orthology group would branch above the node that carries this Pax6-seeded IBD. Independently, no PAX4 ortholog has a forebrain phenotype, and human complete-loss patients have no reported extra-pancreatic features. Human Bgee has low fetal ganglionic-eminence presence (CONSISTENT tier) and absence from all stage-merged forebrain entities where PAX6 is present. This shows expression, not function. The same action is taken in the mouse review. Propagation Review Root cause: PROPAGATION BAD Failure modes: WRONG ORTHOLOG OR PARALOG FUNCTIONAL DIVERGENCE Sources checked: PANTHER:PTN002900580 · PTN002900580 SUPPORTS SOURCE BUT NOT TARGET Euteleostomi node whose children are a Tetrapoda and a Neopterygii duplication. PAX4 and PAX6 are separated independently in each, so the Pax6-seeded IBD reaches PAX4. MGI:MGI:97490 · mouse Pax6 SUPPORTS SOURCE BUT NOT TARGET Pax6 forebrain patterning is well established, but it belongs to the paralog. Supporting Evidence: PMID:24951566 Our synteny analysis showed that Pax4, -6, and -10 originated in the 2R genome duplications early in vertebrate evolution. file:mouse/Pax4/Pax4-bioinformatics/panther_tree_paths.md PTN002900580[Euteleostomi/SPE] has children: PTN009179984[Tetrapoda/DUP], PTN009180006[Neopterygii/DUP] file:mouse/Pax4/Pax4-bioinformatics/panther_tree_paths.md | Homo sapiens | PAX4 | PAIRED BOX PROTEIN PAX-4 | HGNC:8618 | PMID:40614820 No extra pancreatic features were reported in either patient. file:human/PAX4/PAX4-bioinformatics/RESULTS.md | GO:0030900 forebrain development | forebrain | CONSISTENT (UBERON:0004023 ganglionic eminence (45.2)) | ENRICHED (UBERON:0002476 lateral globus pallidus (93.2)) | |
| GO:0045892 negative regulation of DNA-templated transcription | IEA GO_REF:0000117 | MODIFY | Summary: ARBA prediction of negative regulation of transcription. Reason: Correct but general. PAX4 represses RNA polymerase II promoters (insulin, glucagon, its own promoter), so the Pol II-specific term GO:0000122, already carried and accepted, is the right one. This matches the mouse review. Proposed replacements: negative regulation of transcription by RNA polymerase II Supporting Evidence: PMID:11263967 PAX4v reversed the wild-type PAX4 mediated repression of the insulin promoter in cotransfection assays PMID:10967107 it exerts a strong negative autoregulatory effect |
| GO:0048513 animal organ development | IEA GO_REF:0000117 | TOO GENERAL | Summary: ARBA machine-learning prediction of animal organ development. Reason: True but uninformative. GO:0003309 type B pancreatic cell differentiation, accepted in this review, is part_of endocrine pancreas development and pancreas development, which are under animal organ development. The accepted specific term therefore already entails this one (ancestry checked against the GO release used by the repository's OAK adapter). TOO_GENERAL rather than MODIFY because the entailing term is a cell differentiation term reached over part_of, not a refinement of the same claim, and the umbrella spans organ biology unrelated to PAX4. The MODIFY rows in this review (regulation of transcription, DNA binding) differ, because their replacements are is_a descendants that sharpen the same claim. Supporting Evidence: PMID:40614820 human PAX4 expression is important for endocrine cell development and has a key role in directing progenitor cells toward a beta cell fate |
| GO:0048731 system development | IEA GO_REF:0000117 | TOO GENERAL | Summary: ARBA machine-learning prediction of system development. Reason: True but uninformative. GO:0003309 (accepted) is part_of endocrine pancreas development, which is part_of endocrine system development, a system development term. The accepted specific term entails this one. TOO_GENERAL rather than MODIFY for the reason given on the animal organ development row. Supporting Evidence: PMID:37777536 we found that PAX4 deficiency and/or loss-of-function to result in derepression of alpha cell genes, leading to the formation of polyhormonal endocrine cells in vitro, with reduced total insulin content |
| GO:0060041 retina development in camera-type eye | IBA GO_REF:0000033 | MARK AS OVER ANNOTATED | Summary: Retina development, inherited from the Euteleostomi node PTN002900580. Its seeds are mouse and rat Pax6, zebrafish pax6a, pax6b and pax10, and rat Pax4 (RGD:620433), whose seed annotation is an IEP. Reason: The only Pax4 seed is the rat IEP from PMID:19012751. That paper found Pax4 absent from the fetal eye and present in photoreceptors only postnatally, highest in the adult. Its authors concluded that this timing argues against a developmental role. The node placement is also suspect for the reason given on the forebrain row, since the PAX4/PAX6 split is drawn twice below this node. No PAX4 ortholog has an eye phenotype, and comparative work states that mammalian and teleost Pax4 are not implicated in eye development. Human data are consistent. Two complete-loss patients surviving into childhood had no reported extra-pancreatic features. Bgee has PAX4 absent from retinal pigmented layer and palpebral conjunctiva where PAX6 is present at the top of its profile; these are stage-merged, so they show consistency only. No human neural-retina or pineal sample has a PAX4 call. As in mouse, the row is marked rather than removed because a late photoreceptor role has not been tested, and the patients' report describes no eye examination. A blinded OpenScientist run, given only this hypothesis, the GO term and the IBA reference (no papers, no review decision, no expression data), independently judged the term over-annotated and pointed to a PAINT-level fix, from the same rat photoreceptor and complete-loss evidence it found itself. Propagation Review Root cause: PROPAGATION BAD Failure modes: WRONG ORTHOLOG OR PARALOG FUNCTIONAL DIVERGENCE SOURCE EVIDENCE WEAK Sources checked: PANTHER:PTN002900580 · PTN002900580 SUPPORTS SOURCE BUT NOT TARGET Euteleostomi node below which PANTHER 19 splits PAX4 from PAX6 twice (Mammalia PTN000676674, Neopterygii PTN009180006), against the 2R origin. RGD:620433 · rat Pax4 SOURCE WEAK OR INFERRED The IEP seed is postnatal and adult photoreceptor expression (PMID:19012751), and the authors argue against a developmental role. ZFIN:ZDB-GENE-081022-10 · zebrafish pax10 SUPPORTS SOURCE BUT NOT TARGET Resolved as pax10 in the mouse review. Pax10 is a distinct 2R paralog whose retinal expression says nothing about PAX4. MGI:MGI:97490 · mouse Pax6 SUPPORTS SOURCE BUT NOT TARGET Pax6 is the retinal developmental regulator, and the role belongs to the paralog. Supporting Evidence: PMID:19012751 Pax4 expression is not detectable in the foetal eye; however, postnatal Pax4 transcript levels rapidly increase. PMID:19012751 The finding that Pax4 expression is highest in the adult argues against a role in development, as is the case with Pax6. PMID:24951566 The Pax4 genes of mammals and teleost fishes are not implicated in the development of eyes file:human/PAX4/PAX4-bioinformatics/RESULTS.md | GO:0060041 retina development in camera-type eye | retina | ABSENT_DEVELOPMENTAL (absent in UBERON:0001782 pigmented layer of retina) | ENRICHED (UBERON:0001782 pigmented layer of retina (88.7)) | file:human/PAX4/PAX4-hypotheses/function-hypothesis-go-0060041/openscientist.md Over-annotated / weakly supported — the term should NOT be retained as a confident PAX4 function. |
| GO:1990837 sequence-specific double-stranded DNA binding | IDA PMID:28473536 Impact of cytosine methylation on DNA binding specificities ... | MODIFY | Summary: From the methyl-SELEX/HT-SELEX survey of human transcription factor specificities. The PAX4 motif is in the supplementary data rather than the main text. Reason: The SELEX result on random oligonucleotides is sound and is exactly what GO:1990837 describes. The human evidence as a whole supports the more specific descendant GO:0000978, which the gene already carries by IBA. PAX4 occupies the insulin promoter in human beta cells and high-affinity sites in its own promoter. The refinement matches the mouse review, where the ISO copy of this row was modified to GO:0000978. Proposed replacements: RNA polymerase II cis-regulatory region sequence-specific DNA binding Supporting Evidence: PMID:28473536 In total, HT-SELEX models now exist for 632 of the ~1400 (23) human TFs. PMID:40614820 The binding of PAX4 to the INS gene promoter is direct evidence for the importance of the gene in insulin secretion |
| GO:7770019 negative regulation of enteroendocrine cell differentiation | IDA PMID:37883554 Unbiased transcription factor CRISPR screen identifies ZNF80... | ACCEPT | Summary: In ZNF800-deficient human small-intestinal organoids, PAX4 knockout shifts enteroendocrine differentiation from enterochromaffin cells to other subtypes, including L cells, through loss of PAX4's inhibition of ARX. Reason: The GO term was created from this paper. The cached text is abstract-only, but the authors' commentary and UniProt's curated FUNCTION text (ECO:0000269 to this paper) describe the result. This is subtype-selective negative regulation within a program PAX4 also drives positively. It agrees with mouse intestine, where Pax4 loss expands L and ghrelin cells with Arx up-regulation. In the human pancreas, whose endocrine cells GO classes as enteroendocrine, PAX4 loss derepresses alpha-cell genes. The term is part of the core fate-allocation mechanism, as in the mouse review. Bgee scores this term ABSENT_DEVELOPMENTAL (PAX4 absent in the type B pancreatic cell samples); as for beta-cell differentiation, stage-merged calls cannot test a developmental role, and they did not affect the action. Supporting Evidence: PMID:38279896 indicating PAX4 is responsible for EC lineage specification by suppressing other lineages file:human/PAX4/PAX4-uniprot.txt Prevents differentiation of other intestinal enteroendocrine cell types by repressing the expression of ARX PMID:22570716 Inactivation of Pax4 leads to the up-regulation of Arx and the differentiation of these progenitors into GLP1-expressing cells file:human/PAX4/PAX4-bioinformatics/RESULTS.md | GO:7770019 negative regulation of enteroendocrine cell differentiation | enteroendocrine cell | ABSENT_DEVELOPMENTAL (absent in CL:0000169 type B pancreatic cell) | ENRICHED (CL:0000169 type B pancreatic cell (99.3)) | |
| GO:7770020 intestinal enterochromaffin enteroendocrine cell differentiation | IDA PMID:37883554 Unbiased transcription factor CRISPR screen identifies ZNF80... | ACCEPT | Summary: The ZNF800 screen identified a PAX4-centred network controlling enterochromaffin differentiation in human small-intestinal organoids. PAX4 knockout reverses the enterochromaffin expansion caused by ZNF800 loss. Reason: The human organoid result and the mouse knockout (serotonin cells exclusively Pax4-dependent) agree. This is a core gut role, and the term was defined from this paper. Bgee shows PAX4 among its highest scores in rectum and colon mucosa (ENRICHED tier for this term), which corroborates but does not establish the role. Supporting Evidence: PMID:37883554 restricts enterochromaffin cell differentiation by directly controlling an endocrine TF network centered on PAX4 PMID:38279896 only the knockout of PAX4 resulted in the reversal of EEC differentiation, shifting from ECs to other rare EEC subtypes PMID:22570716 Serotonin-producing cells are exclusively Pax4-dependent file:human/PAX4/PAX4-bioinformatics/RESULTS.md | GO:7770020 intestinal enterochromaffin enteroendocrine cell differentiation | type EC enteroendocrine cell | ENRICHED (UBERON:0001052 rectum (61.8)) | CONSISTENT (UBERON:0001826 nasal cavity mucosa (78.8)) | |
Loading supporting content…
Download this section (compressed HTML)Q: PAINT node placement for PTHR45636. PANTHER 19 separates PAX4 from PAX6 twice below the Euteleostomi node PTN002900580, at a Mammalia duplication (PTN000676674, above the human PAX4 clade PTN000676639) and at a Neopterygii duplication (PTN009180006). Synteny instead places one origin of Pax4, Pax6 and Pax10 in the 2R duplications (PMID:24951566). As a result, the Pax6-seeded IBDs for retina development (GO:0060041) and forebrain development (GO:0030900) at PTN002900580, and for brain development (GO:0007420) and sensory organ development (GO:0007423) at the Bilateria node PTN008693041, all reach human PAX4. Should PAINT add IRDs for these terms on the PAX4 clades (PTN000676639 in mammals, PTN009180007 in ray-finned fishes) while keeping GO:0003309, which both paralogs support? Or should the tree be reconciled so that one PAX4 orthology group branches above PTN002900580?
Q: Does human PAX4 repress ARX directly? ARX rises in PAX4-null hiPSC endocrine progenitors and falls with PAX4 overexpression, but falls with PAX4 knockdown in EndoC-betaH1 cells, and the intestinal organoid study reports a unilateral inhibition without, as far as the cached text shows, direct binding.
Q: Why is PAX4 essential for beta-cell formation in mouse but not in human, where complete loss gives transient neonatal diabetes with remission and later relapse (PMID:40614820)? Does the relapse reflect a role in adult beta-cell maintenance or expansion, as the R192H knock-in mouse suggests (PMID:42210309)?
Q: Is the dominant-negative isoform 2 (PAX4v, PMID:11263967) produced in normal human islet or intestinal development, or only in insulinoma?
Q: Should MODY9 remain a disease designation for PAX4, given that variant- and gene-level population evidence does not support PAX4 as a high-penetrance MODY gene (PMID:35108381, PMID:36208030)?
Experiment: CUT&RUN for endogenously epitope-tagged PAX4 in hiPSC-derived pancreatic endocrine progenitors and in ZNF800-deficient human small-intestinal organoids, integrated with single-cell RNA-seq of PAX4-null lines, and deletion or mutation of candidate PAX4 sites at the ARX locus.
Hypothesis: Human PAX4 directly represses ARX and alpha-cell genes in endocrine progenitors and in intestinal enteroendocrine progenitors.
Type: genome-wide occupancy and enhancer mutagenesis
Experiment: Differentiate PAX4-null and isogenic control hiPSCs into retinal and cortical organoids, with PAX6-null lines as positive controls, and score lamination, photoreceptor maturation and regional identity. In parallel, record ophthalmological findings in any further complete-loss patients.
Hypothesis: PAX4 has no role in human eye or forebrain development, so the inherited IBA terms reflect paralog biology only.
Type: organoid phenotyping
Experiment: Transplant PAX4-null, R192H and corrected hiPSC-derived islets into immunodeficient mice, challenge the hosts with high-fat feeding, and follow insulin secretion, hormone co-expression and alpha-cell gene derepression over time.
Hypothesis: PAX4 sustains adult human beta-cell identity under metabolic stress.
Type: xenotransplant functional assay
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: Whether human PAX4 represses ARX by binding ARX regulatory DNA, and why ARX responds in opposite directions to PAX4 loss in endocrine progenitors and in a mature beta-cell line, is unknown.
OPEN BIOLOGY RESIDUAL_SUBGAP
What is known: PAX4 is a sequence-specific repressor of the insulin, glucagon and PAX4 promoters. ARX rises in PAX4-null hiPSC endocrine progenitors and falls with PAX4 overexpression in hESC-derived progenitors. In EndoC-betaH1 cells, PAX4 knockdown lowers ARX. In intestinal organoids PAX4 inhibits ARX unilaterally.
Significance: It decides whether the PAX4-ARX antagonism that allocates endocrine fates in mouse is a direct repressor-target relationship in human cells or an indirect network effect.
What would resolve it: Occupancy of endogenously tagged PAX4 at the ARX locus in human progenitors and enteroendocrine cells, with site mutagenesis (see suggested_experiments).
Provenance (the field's own admissions):
Gap: What allows human beta cells to form without PAX4, and what PAX4 contributes after birth that explains diabetes relapse in complete-loss patients, is unknown.
OPEN BIOLOGY RESIDUAL_SUBGAP
What is known: Complete PAX4 loss in two children gave transient neonatal diabetes that remitted and relapsed in childhood. PAX4-null hiPSC-derived islets form beta-like cells with polyhormonal features.
Significance: It defines the human-specific role of PAX4 that differs from mouse, where the gene is essential for beta-cell formation.
What would resolve it: Longitudinal functional study of PAX4-null hiPSC-derived islets in vivo under metabolic stress (see suggested_experiments).
Provenance (the field's own admissions):
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)