The review covers 171 normalized annotations from 173 imported GOA records. Two differences in assignment/date metadata collapse during normal seeding; their raw records are retained. Decisions are 39 ACCEPT, 70 MODIFY, 61 KEEP_AS_NON_CORE and one UNDECIDED. No source assertion is removed, and no NEW annotation is proposed. All five official product identifiers and sequence notes are preserved.
The single core is a GTPase motor that couples membrane recruitment and oligomerization to remodeling and scission. Endocytic, recycling-endosome/autophagic and cortactin-dependent actin-bundle contexts share this machinery; each is not a separate enzyme. The two short primary excerpts are placed once in the core support. Detailed source locators and assay limits are in the source guide.
Human catalytic support includes purified human DNM2 in PMID:24970086 and human Dyn2 isoform 1 in PMID:36445308. In the former, the human association/GTPase experiments must be distinguished from the rat-brain dynamin fission preparation. In the latter, the human biochemical preparation must be distinguished from the rodent muscle-cell phosphorylation and transport experiments. PMID:34744632 and the purified actin reconstitution in PMID:19605363 use rat DNM2; human host cells do not establish the origin of a transgene.
PMID:32315611 supports LC3 recruitment and recycling-endosome scission, with positive human endogenous evidence. Its HA-DNM2 construct is rat; the current purified-protein catalog sequence matches human P50570-1, but the historical lot is not sequence-authenticated, and the gifted GFP construct remains unresolved. GABARAP-family and LC3B associations support ubiquitin-like protein binding without assigning conjugation or an adaptor activity to DNM2. The R465W phenotype involves altered recruitment and distribution; it is not evidence of universal catalytic loss.
Domain-resolved SH3/PRD assays support binding-class refinements for several repeatedly observed partners. A later AP-MS or co-IP result retains its own co-complex scope: an independently mapped interface does not make that later assay a direct-contact experiment. All nine HuRI target and partner sequence identifiers match the official products, including UBASH3A-2 and FLAD1-2. A partner class supports a binding refinement only; it does not transfer that partner's catalytic or signaling role. The portable interaction analysis reproduces literal raw-record selections, direction and accession groups.
Eight positive generic-binding annotations remain non-core where no supported, more informative class was established. This follows the explicit ClinGen project curation instruction, preserving positive evidence rather than deleting it for genericity. The resulting generic-binding advisories are expected.
All eight PAINT ancestral assertions are on the recovered path to the DNM2 leaf. Donor number and the target's presence among descendant evidence are not failure diagnoses. Model-specific neuronal, microtubule and adhesion evidence remains explicit. ARBA condition recovery is incomplete, so the rule-source assessment is unresolved even when independent biology supports a term refinement.
Broad cytoplasm/endosome replacements describe independently supported DNM2 pools, not exhaustive localization of every cell in an original experiment. HPA supports cytosol and Golgi. Specialized urinary-vesicle and adhesion detections are retained with unresolved target-table access; they are not dismissed as contamination. The sole UNDECIDED source is the historical NAS G2/M transition claim: predicted kinase motifs and later mitotic phosphorylation/cytokinesis evidence do not establish that particular transition.
The five products differ in the stalk-region alternative segment, the four-residue PH-region insertion/deletion and a proline-rich-region serine. No universal functional equivalence or inactivity is inferred. Historical aa/ab/ba/bb labels, rat constructs and disease mutants are not assigned to modern human product IDs without sequence evidence.
Normal publication-cache flags and machine source bytes remain unchanged. Separately recovered primary text supplies additional evidence where normal caches are abstract-only; those distinctions are recorded in the source guide and reference assessments. Ordinary provider research was attempted but failed authentication/quota checks; no provider report was fabricated. This manual review instead uses the cited primary sources and records.
The effective publication versions of PMID:15703209 and PMID:17257598 match the existing DNM1 source versions: full text and abstract-only, respectively. Genuine initial DNM2 fetch variants remain preserved separately. Detailed PMID:17257598 Methods/Results were read in the full variant and are not claimed to be present in the effective abstract cache. The existing full PMID:18425118 cache is preserved; only its abstract was read for this review. No biological decision or quotation changed during this reconciliation.
The eight imported GTPase assertions retain their motor-activity refinement. Current QuickGO is_a ancestry places GO:0061791 below GO:0003924. The seven human comparator queries (DNM1, DNM2, DNM3, DNM1L, OPA1, MFN1 and MFN2) contain no motor annotations, but the complete all-taxon motor query returns four PomBase records: msp1/fzo1 on GO:0061791 and dnm1/vps1 on its membrane-scission child GO:1990606. Thus human non-use is real, while a general prohibition on dynamin-family use is not supported. Direct human biochemical evidence remains the basis of the DNM2 refinement; the comparator query is a convention check, not a transfer of yeast physiology. Query links and counts are in the source guide.
The two broad autophagy rows are refined at different stages. PMID:32315611 supports macroautophagy through LC3-dependent scission and release from recycling endosomes. Its completion assay found unchanged open-phagophore counts, accumulated completed autophagosomes and fewer autolysosomes; this is not assigned as double-membrane formation or closure. The authors distinguish that route from the earlier SNX18-dependent ATG9A carrier pathway. PMID:29437695 supports the latter membrane-supply step in autophagosome assembly through WT/W38K recruitment and rescue experiments. The single motor core uses macroautophagy to encompass both supported stages without duplicate ancestor/descendant process slots. The current maturation term specifies PI3P/Atg8 removal, which these experiments do not establish, and release alone is not evidence of directed transport. PMID:32315611(https://pubmed.ncbi.nlm.nih.gov/32315611/), PMID:29437695(https://pmc.ncbi.nlm.nih.gov/articles/PMC5891424/)
Cytoplasm replacements now retain cytosol and recycling-endosome membrane; the cytoskeleton replacement retains microtubule. Clathrin-coated-pit and focal-adhesion positives remain in their own source annotations, and the pit remains in the core. These edits respect the defined compartments without claiming that missing ontology ancestry alone disproves a localization.
Three short assay anchors are added once each in the corresponding YAML reviews. The transferrin endpoint is measured in rodent C2C12 cells, with human Dyn2 isoform 1 biochemistry a separate part of the paper. The SNX18 anchor comes from the authors' trafficking model, supported by the original WT/W38K experiments. The polymer-disassembly anchor concerns rat disease constructs, supplemented by independent human polymer evidence. Neither source species nor all-product equivalence changes. PMID:36445308(https://pmc.ncbi.nlm.nih.gov/articles/PMC9712776/), PMID:34744632(https://pmc.ncbi.nlm.nih.gov/articles/PMC8563704/)
All 171 source objects, five products, existing quotations, source archives and interaction-artifact bytes remain unchanged. Effective actions are 37 ACCEPT, 72 MODIFY, 61 KEEP_AS_NON_CORE and one UNDECIDED; no NEW. The source guide records the literal enzyme-binding definition and explains why the QARS1/FLAD1-2 refinements do not transfer catalytic activity to DNM2. PIP2 binding remains a supported recruitment property represented within the integrated motor mechanism, without adding a separate core unit.