ENDOG

Summary: ENDOG-family nucleases use catalytic magnesium for phosphodiester cleavage.

Reason: Mouse ENDOG and conserved active-site features support Mg(2+) binding by the human nuclease; the metal is part of the nuclease catalytic center rather than a generic structural annotation.

Summary: ENDOG-family nucleases use catalytic magnesium for phosphodiester cleavage.

Reason: Mouse ENDOG and conserved active-site features support Mg(2+) binding by the human nuclease; the metal is part of the nuclease catalytic center rather than a generic structural annotation.

Summary: The InterPro nucleic-acid-binding row is a broad proxy for ENDOG nuclease activity.

Reason: The InterPro records mark the DNA/RNA nonspecific endonuclease fold, so this IEA binding proxy is replaced by what that fold actually confers. Two of the three replacements are a lateral correction rather than a refinement. GO:0004520 and GO:0004521 sit under GO:0004518 nuclease activity and GO:0003824 catalytic activity, not under GO:0005488 binding, so neither is a descendant of GO:0003676; CLAUDE.md sanctions a MODIFY whose replacement is not a descendant, and this is one. The third, GO:0051880, is a genuine descendant by way of GO:0003677 DNA binding, and is kept so that narrowing the binding proxy does not discard binding altogether, because a specific binding activity is separately demonstrated for human ENDOG and carries its own row by IDA on PMID:36394256. GO:0004520 and GO:0004521 are rows the gene already carries in GOA; GO:0051880 is proposed as a NEW row in this review rather than already annotated. The ancestries were checked against the GO build rather than assumed.

Proposed replacements: DNA endonuclease activity RNA endonuclease activity G-quadruplex DNA binding

Summary: Human ENDOG inherited DNA endonuclease activity from the ENDOG/NUC1 nuclease family.

Reason: The PAINT assertion is consistent with the ENDOG-specific PANTHER subfamily and with direct human and mouse experiments showing that EndoG cleaves double-stranded DNA, including 5hmC-modified DNA and mitochondrial DNA.

Summary: ENDOG directly cleaves DNA substrates.

Reason: Human and mouse experiments support ENDOG as a DNA endonuclease: it cleaves 5hmC-modified double-stranded DNA, cleaves damaged mtDNA in a nuclease-dependent mitochondrial genome-maintenance response, and retains nuclease-dependent signaling to autophagy.

Summary: ENDOG directly cleaves DNA substrates.

Reason: Human and mouse experiments support ENDOG as a DNA endonuclease: it cleaves 5hmC-modified double-stranded DNA, cleaves damaged mtDNA in a nuclease-dependent mitochondrial genome-maintenance response, and retains nuclease-dependent signaling to autophagy.

Summary: ENDOG directly cleaves DNA substrates.

Reason: Human and mouse experiments support ENDOG as a DNA endonuclease: it cleaves 5hmC-modified double-stranded DNA, cleaves damaged mtDNA in a nuclease-dependent mitochondrial genome-maintenance response, and retains nuclease-dependent signaling to autophagy.

Summary: ENDOG directly cleaves DNA substrates.

Reason: Human and mouse experiments support ENDOG as a DNA endonuclease: it cleaves 5hmC-modified double-stranded DNA, cleaves damaged mtDNA in a nuclease-dependent mitochondrial genome-maintenance response, and retains nuclease-dependent signaling to autophagy.

Summary: RNA endonuclease activity is inherited in the DNA/RNA nonspecific ENDOG/NUC1 family.

Reason: ENDOG belongs to the conserved DNA/RNA nonspecific endonuclease family; the PAINT assertion captures this catalytic substrate breadth even though the best human process evidence centers on DNA cleavage. Mouse work gives a direct RNA substrate. Recombinant EndoG cleaves the 3'-end region of the mitochondrial H-strand transcript downstream of mt-tRNA(Thr) in vitro, and Endog-null liver accumulates that precursor (PMID:39752519).

Summary: ENDOG is a DNA nuclease, but the endonuclease child term is more precise.

Reason: The rat-derived Compara transfer captures the right catalytic function at a parent level; human ENDOG is specifically a DNA endonuclease.

Proposed replacements: DNA endonuclease activity

Summary: The UniProtKB:Q8WW22 interaction does not establish an informative ENDOG molecular function.

Reason: This is a generic protein-binding annotation from a large interaction screen. The DNAJA4 and ITLN2 edges are not part of the established ENDOG nuclease, mitochondrial DNA-maintenance, or 14-3-3-dependent autophagy mechanisms, so the rows add no actionable molecular-function information.

Summary: The UniProtKB:Q8WWU7 interaction does not establish an informative ENDOG molecular function.

Reason: This is a generic protein-binding annotation from a large interaction screen. The DNAJA4 and ITLN2 edges are not part of the established ENDOG nuclease, mitochondrial DNA-maintenance, or 14-3-3-dependent autophagy mechanisms, so the rows add no actionable molecular-function information.

Summary: The UniProtKB:Q8WWU7 interaction does not establish an informative ENDOG molecular function.

Reason: This is a generic protein-binding annotation from a large interaction screen. The DNAJA4 and ITLN2 edges are not part of the established ENDOG nuclease, mitochondrial DNA-maintenance, or 14-3-3-dependent autophagy mechanisms, so the rows add no actionable molecular-function information.

Summary: A nuclear ENDOG pool can cleave genomic DNA.

Reason: Nuclear ENDOG is consistent with the PAINT transfer and with direct evidence for ENDOG-dependent 5hmC-modified DNA cleavage and conservative recombination on nuclear plasmid substrates.

Summary: ENDOG is a mitochondrial nuclease.

Reason: A cleaved mitochondrial precursor and high-throughput mitochondrial proteome support the major mitochondrial ENDOG pool; full-text human work directly places ENDOG in mitochondria and tests its mtDNA cleavage and starvation-release functions.

Summary: ENDOG is a mitochondrial nuclease.

Reason: A cleaved mitochondrial precursor and high-throughput mitochondrial proteome support the major mitochondrial ENDOG pool; full-text human work directly places ENDOG in mitochondria and tests its mtDNA cleavage and starvation-release functions.

Summary: ENDOG is a mitochondrial nuclease.

Reason: A cleaved mitochondrial precursor and high-throughput mitochondrial proteome support the major mitochondrial ENDOG pool; full-text human work directly places ENDOG in mitochondria and tests its mtDNA cleavage and starvation-release functions.

Summary: The ENDOG family includes inner-membrane-associated mitochondrial nucleases.

Reason: The membrane claim in this row has its own positive support, and the donor check prescribed by the organelle policy was run. Both PAINT donors carry GO:0005743. Yeast Nuc1 does, and the purification of the major yeast mitochondrial nuclease places that activity in the inner membrane (PMID:3286639). Human EXOG carries it by IDA, from PMID:18187503, and that paper subfractionated human mitochondria expressing tagged EndoG and EXOG and found both in membrane fractions while excluding them from the soluble intermembrane space and the soluble matrix. That is evidence for membrane association of an EndoG protein rather than only of the PAINT donors, so the row is kept rather than marked over-annotated. Note which EndoG, though. The construct in that paper is the bovine ortholog, expressed in human cells alongside human EXOG, so this is not a measurement on the human protein; it is still independent of the donor annotations, because bovine EndoG is not one of them. The same paper also speaks to inner versus outer. Subfractionation alone cannot separate them, because those membrane fractions co-stained for COX-IV, TOM20 and cytochrome c together; but the immunogold electron microscopy in the next panel can, and its legend reports both tagged proteins delineating the cristae, which are inner membrane. One honest discrepancy inside that paper, recorded rather than smoothed over. The running text describes the EM only for EXOG and cites earlier work for mammalian EndoG, so the legend's claim about both proteins is the stronger of the two statements; the whole experiment used tagged, expressed protein; and, as above, the EndoG construct is bovine. That is enough to keep an inner-membrane row with its own support, and not enough to make the location core, so the row stays non-core. One further limit, recorded as a source defect rather than argued here. This evidence places the protein at the inner membrane; it does not show the nuclease acting there, and the same row argues below that mtDNA cleavage needs matrix access, so located_in fits the support better than the inherited is_active_in. Sub-compartment check under the organelle policy, which is a separate question from whether this row stands. Fractionation places mammalian EndoG in the intermembrane space and not the matrix (PMID:12444964), consistent with its release on outer-membrane permeabilization. In human HeLa cells it instead colocalizes with the matrix nucleoid marker TFAM (PMID:36394256), and its cleavage of mtDNA (PMID:29719607) and, in mouse, of mitochondrial transcripts (PMID:39752519) requires matrix access. The worm ortholog CPS-6 moves from the intermembrane space to the matrix after fertilization (PMID:27338704), so both may hold under different conditions. A sequence analysis narrows the import route without settling the compartment. ENDOG has a cleavable arginine-rich presequence and only three cysteines with no twin CX9C or CX3C motif, so it enters by the TIM23 presequence route and is not a MIA40 disulfide-relay intermembrane-space protein. That route delivers proteins to both compartments, the same signature is shared with the EXOG paralog, and the classifier scored only 60 percent on reference proteins of known sub-compartment, so it does not discriminate. Because those lines disagree, no soluble intermembrane-space or matrix row is added; the membrane-fraction result above is consistent with that, since it finds ENDOG in neither soluble compartment. The root mitochondrion rows carry the overall location.

Summary: ENDOG degrades DNA in two compartments, so the broad BP is refined to both supported descendants.

Reason: ARBA assigns a broad DNA catabolic process row. Direct human evidence shows ENDOG nuclease-dependent removal of mtDNA coupled to compensatory mitochondrial genome replication (GO:0032043). That term is logically defined as DNA catabolic process occurring in the mitochondrion, so it does not cover ENDOG's cleavage of nuclear chromatin after release (PMID:11452314, PMID:11606588), which apoptotic DNA fragmentation (GO:0006309) carries. Both specific descendants are listed rather than narrowing to the mitochondrial one alone.

Proposed replacements: mitochondrial DNA catabolic process apoptotic DNA fragmentation

Summary: ENDOG-family nucleases can contribute to apoptotic DNA fragmentation, but this is not human ENDOGs principal function.

Reason: ENDOG is catalytically capable of fragmenting chromosomal DNA after mitochondrial release, so the PAINT process transfer is plausible. Recent direct human data instead center ENDOG on 5hmC-modified DNA cleavage, mitochondrial genome maintenance, and autophagy, making apoptotic DNA fragmentation a retained non-core branch rather than the defining function. Checked against the APOPTOSIS project's assay-readout rule. This is not an endpoint-readout inference, because ENDOG is itself the nuclease that performs the step. Recombinant human EndoG reproduces the defining two-stage pattern on isolated nuclei, large fragments then nucleosomal cleavage (PMID:11606588). EndoG fragments chromatin in DFF-deficient mouse cells (PMID:11452314). In worms, cps-6 loss impairs apoptotic DNA degradation, and the nuclease activity of CPS-6 mutants tracks their killing activity (PMID:11452313, PMID:22223640). Necessity in mammals is weak. A null mouse allele that spares the overlapping gene shows no obvious apoptotic DNA-degradation defect (PMID:15601850), so the role is redundant with CAD/DFFB. Non-core is the right weight. The apparent conflict between the two mouse lines does not change that. The PMID:14663139 allele was embryonic lethal, but the PMID:15601850 authors built a null that spares the gene lying tail to tail with Endog and attribute the lethality to disrupting that neighbour, which is why PMID:14663139 is marked DISPUTED in the reference list. The cleaner allele is therefore the better test of necessity. The activity itself does not rest on either mouse, but on the purified human enzyme cleaving isolated nuclei (PMID:11606588) and on fragmentation in CAD-deficient cells (PMID:11452314), so the activity is real and the requirement is not.

Summary: ENDOG-dependent 5hmC cleavage can promote conservative DNA recombination.

Reason: The seeded TAS reference is a broad apoptosis-effector review, but later direct work showed EndoG-dependent cleavage of 5hmC-modified DNA and increased conservative recombination, supporting the existing DNA recombination process term.

Summary: ENDOG-generated DNA breaks can activate a DNA damage response during autophagy induction.

Reason: The 2021 study linked ENDOG endonuclease activity to DNA damage-response signaling that helps trigger autophagy, but ENDOG is the nuclease generating DNA lesions rather than a canonical repair or checkpoint factor; keep this as secondary to nuclease and autophagy annotations.

Summary: This rat-derived stress or stimulus row is too context-specific for human ENDOG core curation.

Reason: The Compara source reflects a rat experimental context in which ENDOG was perturbed or mobilized under a stress condition. Human ENDOG has conserved nuclease and autophagy activities, but the specific calcium, glucose, hypoxia, oxidative, estradiol, mechanical, or hydrogen-peroxide cell-death contexts should not be propagated as reusable human ENDOG biological processes without direct proximal evidence.

Summary: ENDOG promotes starvation-induced autophagy.

Reason: Human ENDOG promotes autophagy through 14-3-3-dependent mTOR suppression and through endonuclease-dependent DNA damage-response signaling; the function is conserved in mouse, Drosophila, and C. elegans starvation assays.

Summary: ENDOG promotes starvation-induced autophagy.

Reason: Human ENDOG promotes autophagy through 14-3-3-dependent mTOR suppression and through endonuclease-dependent DNA damage-response signaling; the function is conserved in mouse, Drosophila, and C. elegans starvation assays.

Summary: The InterPro hydrolase row is a broad proxy for ENDOG endonuclease activity.

Reason: ENDOG hydrolyzes nucleic-acid phosphodiester bonds as a DNA/RNA endonuclease, so the generic hydrolase ancestor should be replaced with catalytic endonuclease terms.

Proposed replacements: DNA endonuclease activity RNA endonuclease activity

Summary: ENDOG suppresses mTORC1 signaling during autophagy induction.

Reason: Upon starvation, phosphorylated ENDOG binds YWHAG/14-3-3gamma, releases TSC2 and VPS34 from YWHAG/14-3-3gamma, and thereby suppresses mTORC1 to initiate autophagy; the standing TOR-signaling term should be narrowed to the TORC1-specific child.

Proposed replacements: negative regulation of TORC1 signaling

Summary: ENDOG suppresses mTORC1 signaling during autophagy induction.

Reason: Upon starvation, phosphorylated ENDOG binds YWHAG/14-3-3gamma, releases TSC2 and VPS34 from YWHAG/14-3-3gamma, and thereby suppresses mTORC1 to initiate autophagy; the standing TOR-signaling term should be narrowed to the TORC1-specific child.

Proposed replacements: negative regulation of TORC1 signaling

Summary: ENDOG cleaves damaged mitochondrial DNA.

Reason: Knockdown, knockout, re-expression, and nuclease-mutant experiments in human cells support ENDOG-dependent mtDNA cleavage and removal when oxidative lesions are not repaired efficiently.

Summary: This rat-derived stress or stimulus row is too context-specific for human ENDOG core curation.

Reason: The Compara source reflects a rat experimental context in which ENDOG was perturbed or mobilized under a stress condition. Human ENDOG has conserved nuclease and autophagy activities, but the specific calcium, glucose, hypoxia, oxidative, estradiol, mechanical, or hydrogen-peroxide cell-death contexts should not be propagated as reusable human ENDOG biological processes without direct proximal evidence.

Summary: This rat-derived stress or stimulus row is too context-specific for human ENDOG core curation.

Reason: The Compara source reflects a rat experimental context in which ENDOG was perturbed or mobilized under a stress condition. Human ENDOG has conserved nuclease and autophagy activities, but the specific calcium, glucose, hypoxia, oxidative, estradiol, mechanical, or hydrogen-peroxide cell-death contexts should not be propagated as reusable human ENDOG biological processes without direct proximal evidence.

Summary: ENDOG homodimerization is conserved and required for nuclease activity.

Reason: Mouse-derived transfer of ENDOG self-association to human is plausible for the conserved nuclease fold, but the self-association row is a structural prerequisite rather than a separate core molecular function. Ortholog structural and biochemical work makes the prerequisite concrete. The dimer binds two DNA strands at separate metal-finger motifs and is needed for full endonuclease activity (PMID:27738134), and oxidation dissociates it into monomers with reduced activity, which is how reactive oxygen species gate the enzyme (PMID:27346342). This supports keeping the row while leaving it non-core.

Summary: ENDOG homodimerization is conserved and required for nuclease activity.

Reason: Mouse-derived transfer of ENDOG self-association to human is plausible for the conserved nuclease fold, but the self-association row is a structural prerequisite rather than a separate core molecular function. Ortholog structural and biochemical work makes the prerequisite concrete. The dimer binds two DNA strands at separate metal-finger motifs and is needed for full endonuclease activity (PMID:27738134), and oxidation dissociates it into monomers with reduced activity, which is how reactive oxygen species gate the enzyme (PMID:27346342). This supports keeping the row while leaving it non-core.

Summary: Perikaryon localization is a rat neuron-specific transfer and is not a core human ENDOG location.

Reason: Human ENDOG is broadly characterized as a mitochondrial nuclease with nuclear activity after mobilization. Orthology transfer of a rat neuronal cell-body localization is overly context-specific compared with mitochondrion and nucleus annotations.

Summary: The InterPro metal-binding row should be narrowed to catalytic magnesium binding.

Reason: ENDOG-family nucleases use Mg(2+) at the catalytic center; the generic metal ion binding parent should be replaced by magnesium ion binding.

Proposed replacements: magnesium ion binding

Summary: Perinuclear-region localization is over-specific relative to ENDOGs core mitochondrial and nuclear pools.

Reason: ENDOG can move from mitochondria toward nuclear DNA, but this rat-derived perinuclear-region transfer is not needed for human ENDOG and should be superseded by mitochondrion or nucleus depending on the assay.

Summary: This rat-derived stress or stimulus row is too context-specific for human ENDOG core curation.

Reason: The Compara source reflects a rat experimental context in which ENDOG was perturbed or mobilized under a stress condition. Human ENDOG has conserved nuclease and autophagy activities, but the specific calcium, glucose, hypoxia, oxidative, estradiol, mechanical, or hydrogen-peroxide cell-death contexts should not be propagated as reusable human ENDOG biological processes without direct proximal evidence.

Summary: This rat-derived stress or stimulus row is too context-specific for human ENDOG core curation.

Reason: The Compara source reflects a rat experimental context in which ENDOG was perturbed or mobilized under a stress condition. Human ENDOG has conserved nuclease and autophagy activities, but the specific calcium, glucose, hypoxia, oxidative, estradiol, mechanical, or hydrogen-peroxide cell-death contexts should not be propagated as reusable human ENDOG biological processes without direct proximal evidence.

Summary: This rat-derived stress or stimulus row is too context-specific for human ENDOG core curation.

Reason: The Compara source reflects a rat experimental context in which ENDOG was perturbed or mobilized under a stress condition. Human ENDOG has conserved nuclease and autophagy activities, but the specific calcium, glucose, hypoxia, oxidative, estradiol, mechanical, or hydrogen-peroxide cell-death contexts should not be propagated as reusable human ENDOG biological processes without direct proximal evidence.

Summary: ENDOG mtDNA cleavage stimulates compensatory mitochondrial DNA replication.

Reason: Human cell knockdown, knockout, and rescue experiments showed that ENDOG nuclease activity promotes both mtDNA depletion and compensatory mitochondrial replication under oxidative stress.

Summary: This rat-derived stress or stimulus row is too context-specific for human ENDOG core curation.

Reason: The Compara source reflects a rat experimental context in which ENDOG was perturbed or mobilized under a stress condition. Human ENDOG has conserved nuclease and autophagy activities, but the specific calcium, glucose, hypoxia, oxidative, estradiol, mechanical, or hydrogen-peroxide cell-death contexts should not be propagated as reusable human ENDOG biological processes without direct proximal evidence.

Summary: The rat-derived apoptotic DNA-fragmentation regulation row should be represented as direct apoptotic DNA fragmentation.

Reason: When ENDOG contributes to apoptotic DNA fragmentation, it does so as the nuclease rather than merely as an upstream positive regulator. The standing PAINT row to apoptotic DNA fragmentation captures this non-core branch at the better level.

Proposed replacements: apoptotic DNA fragmentation

Summary: Phosphorylated ENDOG binds YWHAG/14-3-3gamma to suppress mTOR signaling.

Reason: Existing GOA captures the downstream negative regulation of TOR signaling and positive regulation of autophagy but not the specific phosphorylation-dependent 14-3-3 binding activity that releases TSC2 and VPS34 from YWHAG/14-3-3gamma during starvation. A mouse liver study extends the same binding mode to a second client, Rictor, which ENDOG displaces from 14-3-3gamma to activate mTORC2 (PMID:37794041). That is corroboration for the binding activity, not a separate human annotation.

Summary: Human ENDOG binds G-quadruplex DNA in mitochondria, selectively over the rest of the mitochondrial genome.

Reason: The GO:0003676 nucleic acid binding MODIFY names this term among its replacements; this row is what establishes it, by IDA on the primary assays. Binding is an activity ENDOG performs itself, and is a step distinct from the cleavage the endonuclease rows record. In human cells, mitochondrial immunoprecipitation shows ENDOG occupying the G-quadruplex-forming regions and not other parts of mtDNA, and ENDOG colocalizes with the G4-specific BG4 antibody; the same paper characterizes the structures in vitro and shows cleavage at G4 single-to-double-stranded junctions. It does not rest on the breakpoint-motif hypothesis run, which was negative and which argued only that these motifs do not explain where population mtDNA breakpoints fall. Remaining caveat for a reviewer, recorded rather than hidden. That paper shows selectivity by immunoprecipitation and imaging, not by a measured affinity comparison against unstructured duplex DNA, so the specificity is qualitative.