Human AP2M1 is UniProt Q96CW1, HGNC:564, the medium μ2 subunit of the AP-2 adaptor complex. The imported review contains 184 distinct source annotations and two alternative products. The original UniProt record and GOA bytes are preserved; the original annotation source objects remain in the review. All 38 referenced PMID caches and 53 Reactome records are present. This inventory does not mean every paper or supplementary interaction table has been read. The reviewed source objects, isoforms and qualifiers will be preserved.
The configured deep-research wrapper materialized a substantive cached Falcon/Edison report and two artifacts. A network-resolution warning did not prevent this cached output. Its citations are research leads requiring primary-source verification; the report is not independent experimental evidence. The reported outcome was use of cached automated synthesis, not a successful new online provider search.
AP2M1 supplies cargo-recognition structure within AP-2. It must be distinguished from the kinases that phosphorylate it, the other adaptor subunits' binding sites, and the cargo being internalized. The official definition of GO:0035615 clathrin-cargo adaptor activity concerns bringing cargo together with clathrin; it is a child of cargo adaptor activity and includes clathrin binding. A μ2 contribution to this complex activity does not imply that isolated μ2 supplies every contact. Existing contributes_to qualifiers must remain intact. No NEW biological-process assertion is currently proposed.
PMID:20603002, A large-scale conformational change couples membrane recruitment to cargo binding in the AP2 clathrin adaptor complex, DOI 10.1016/j.cell.2010.05.006. Official PubMed identity/abstract and indexed original introductory mechanism were inspected. The μ2 C-terminal domain recognizes YxxΦ signals, whereas σ2 recognizes acidic dileucine signals. Membrane-dependent opening exposes cargo sites. Detailed recombinant construct species and full Methods remain unverified; do not describe every structural reagent as human. PubMed, original article.
PMID:17035630, Functional analysis of AP-2 alpha and mu2 subunits, DOI 10.1091/mbc.e06-05-0452. Official abstract and indexed original plasmid/cell Methods, selected Results and Discussion were read. Stable HeLa knockdown/rescue assays distinguish μ2 sorting-signal and phosphorylation mutants from its phosphoinositide-site mutant. The Methods name mouse α constructs; they do not independently establish the species of the inherited μ2 construct. Human host cells therefore should not be conflated with human recombinant μ2. The study also shows why a tagged adaptor's presence at a membrane need not establish normal vesicle dynamics. PubMed, original article.
PMID:31430451, Temporal Ordering in Endocytic Clathrin-Coated Vesicle Formation via AP2 Phosphorylation, DOI 10.1016/j.devcel.2019.07.017. Original indexed summary, targeted RPE-cell phosphorylation Results, NECAP recruitment Results/Discussion and retroviral construct Methods were inspected. Phosphorylated μ2 provides a NECAP-binding platform during coat maturation. The study distinguishes this effect from a large change in cargo affinity in vitro. Its inherited μ2 construct sequence/species has not yet been traced; explicitly human NECAP constructs do not resolve that question. Original article.
PMID:32155440, DOI 10.1016/j.devcel.2020.02.010, is the published erratum to PMID:31430451. The complete short correction was inspected: it restores an omitted author, Susanne Salomon. It reports no corrected experiment and is not additional mechanistic evidence. Correction.
PMID:39383236, SLC26A4-AP-2 mu2 interaction regulates SLC26A4 plasma membrane abundance in the endolymphatic sac, DOI 10.1126/sciadv.adm8663. Official identity/abstract, Fig. 1 caption and indexed study-design Methods were inspected. The work combines cargo-tail interaction assays with mouse endolymphatic-sac localization and uptake context; modeling of a human complex is not a human binding experiment. The authors explicitly did not perform endogenous tissue co-immunoprecipitation because of transient binding and limited tissue. Further construct and Results reading is pending the normal cache. PubMed, original article.
One ordinary five-ID cache batch for these references timed out after 45 seconds with no output records. This is one terminated batch, not five individually completed failures. The original log and receipt are retained; no retry or canonical cache replacement was performed.
PMID:31104773, DOI 10.1016/j.ajhg.2019.04.001, reports recurrent human p.Arg170Trp and impaired clathrin-mediated uptake. Earlier targeted original Methods and Results reading identified the AP2u2-mCherry reagent as Addgene #27672, whose repository record specifies a rat insert. The separate IRES-mRFP construct remains unresolved. Human patient genetics, human host cells and construct species are separate evidence scopes. The normal cache is now present; its complete relevant body still needs reconciliation with these preliminary notes. Original article, reagent.
PMID:34315807, DOI 10.1126/scisignal.abg3555, is already cached. The indexed plasmid Methods identify human AP2M1 cDNA DNASU HsCD00041654. Targeted kinase Results establish LRRK2-dependent phosphorylation of AP2M1 Thr156. AP2M1 is the regulated substrate, not the kinase; the entire trafficking/neurodegeneration experiment series has not yet been reviewed. Original article.
PMID:19419997, DOI 10.1152/ajpcell.00630.2008. Actual cached targeted Methods and KIAA0319-binding Results identify μ2 clones with accession NM_004068, map the KIAA0319 tail interaction and show that Y995A disrupts binding and internalization. GST pull-down from HeLa lysates and yeast interaction tests are distinct assays; neither should be silently described as purified binary binding. This provides source-specific cargo-recognition evidence beyond an undifferentiated interaction claim.
PMID:23676497, DOI 10.1172/JCI65401. The normal record is abstract-only. The publisher's actual human hippocampal co-immunoprecipitation Results and selected ADAM10-tail internalization Results/Fig. 3 were additionally read. They support AP-2 association with human ADAM10 and a functional tail motif in COS7 trafficking assays. Do not infer a human recombinant μ2 sequence or purified interface from human tissue coassociation. Publisher.
Reactome summaries are being assessed as individual pathway statements. Cargo delivery to lysosomes or endolysosomes does not by itself establish continued localization of the AP-2 coat there, particularly when the cited event explicitly includes uncoating. The source assertions remain preserved while their component-level scope is examined.
The complete cached abstracts of PMID8918456 and PMID9171339 directly describe mu2 recognition of Yxx-hydrophobic sorting peptides, measured by selection/surface plasmon resonance and photoreactive peptide approaches. They support the conserved signal-recognition role but do not identify reagent species in their abstracts. PMID12032142 maps the V-ATPase H-subunit interaction to mu2 residues 1–145 and H-subunit residues 133–363; this is a regulatory H-subunit interaction, not evidence that H itself is an ATPase catalytic subunit. Its abstract supports a specific adaptor context; source Methods still require review before a narrow domain-specific refinement.
PMID14530274’s abstract explicitly images alpha/beta2 adaptins at the plasma membrane and distinguishes static AP2-associated clathrin from a dynamic population. This supports localization of the complex, without establishing AP2M1-specific trafficking kinetics. PMID23529131 describes human HT29cl.19A epithelial KCNQ1 endocytosis and AP2 association in its complete abstract, but the subunit-specific IP Methods must be inspected before independently adjudicating the mu2 transporter-binding row. PMID23864651’s complete abstract identifies human GLP1R bait/human fetal-brain prey, CHO coIP and AP2M1 among three focused interactors; it supplies a human interaction provenance rather than establishing an intrinsic signal-inhibitory enzyme.
PMID34964704’s complete abstract explicitly connects AP2M1 to CLDN2 YxxΦ motifs 67–70 and 148–151, with coIP and mutation evidence across cell/mouse/human explant settings. This makes the source-specific generic binding row a candidate for sorting-signal recognition refinement; full target Results/Methods are pending. PMID7989329’s complete abstract is an in-vitro AP50/V-ATPase reassembly experiment: removal impairs V1/V0 reassembly and purified AP2 restores it. It is evidence for a structural assembly contribution, not AP2M1 ATPase chemistry. Source species is not assigned by that abstract. The lengthy PMID31104773 abstract output ended before the final sentence, so no new full-abstract-read claim is made from that display; its complete target Methods/Results remain required.
The independently checked RCSB 2XA7 structure record identifies the structural μ2 chain as rat P84092, with human beta2 and mouse sigma2. That chain-specific result must not be generalized to all constructs in PMID20603002 or the later functional papers. No canonical annotation decisions have yet been authored.
PMID:31104773 was subsequently read at the complete cell/astrocyte, molecular-biology, transfection, imaging, co-IP/fractionation and transferrin-uptake Methods, plus the two functional Results sections and opening mechanistic Discussion. Human HeLa AP2M1 knockdown/rescue and conditional mouse astrocytes show impaired transferrin uptake with Arg170Trp, despite preserved expression, membrane recruitment and clathrin-pit localization. These are functional uptake and localization experiments, distinct from model-based predictions about cargo-recognition energetics. The rescue constructs derive from Addgene27672 and a prior AAV vector; the article does not by itself establish every construct’s species. The primary functional evidence supports existing CME and localization assertions, without adding a disease-phenotype BP or claiming that a cargo-affinity mechanism was directly measured.
The official Addgene27672 record was then inspected at Gene/Insert and depositor comments: the AP2u2-mCherry insert is rat Ap2m1, derived from a David Owen construct, with mCherry inserted at residue 236. This resolves the named mCherry reagent in PMID:31104773, without assigning a species to the separately inherited AAV construct or replacing the source’s curated human annotation. Human-cell depletion/rescue and the rodent reagent must be described separately.
The official indexed GO:0005048 page defines signal sequence receptor activity as recognition of a short peptide that directs cellular localization; it is not limited to cleavable ER signal peptides. Its parent is molecular function. The YxxΦ cargo-motif experiments therefore fit this established source term. Initial direct requests timed out; the indexed official definition and parent table were read at that stage. A later direct official-page check succeeded, as documented below.
The five normal records were subsequently recovered once and imported unchanged by the project owner. The recovered record identities, exact titles and full-text flags were checked against the two earlier independent citation assessments; the source-specific boundaries are summarized here. PMID:17035630 is HTML-derived; PMID:20603002, PMID:31430451 and PMID:39383236 contain XML-derived bodies. PMID:32155440 contains bibliographic erratum metadata only; the authorship correction was read earlier from the external original, not inferred from that short cache. The recovered records supersede the pending-cache wording above without changing their immutable content.
In the actual PMID:39383236 Methods, the prey library is mouse kidney, SLC26A4 is mouse NP_035997.1 residues 512–780, and μ2 is mouse NP_033809.1 residues 124–435. Human HeLa cells are assay hosts. Mouse tissue imaging uses alpha-adaptin as an AP-2 proxy, rather than a μ2-specific tissue antibody; the human molecular model is separate. PMID:20603002's RCSB 2XA7 μ2 chain is rat P84092. PMID:31430451's μ2 construct remains of unresolved species, while its phosphorylation-dependent NECAP recruitment supplies a mechanistic platform role. None of these experiments makes μ2 a kinase, ATPase or autonomous clathrin polymerase.
PMID:22916011: the actual cached core-binding Results, motif-mutation experiments, Fig. 2 and plasmid Methods were read. The AP2M1 ORFs came from the Human ORFeome library. Microfluidic binding, complementation and endogenous co-IP in infected human Huh-7.5 cells support recognition of the HCV core sorting motif; the paper reports similar binding for both tested AP2M1 isoforms. This supports refining the source-specific generic interaction to signal-sequence recognition. It does not justify a new viral-assembly process assertion or collapsing the two recorded products.
PMID:23864651: the complete available cache includes the abstract and Discussion, but omits Methods and the primary Results. Its human fetal-brain prey library and human GLP1R bait, together with the reported CHO co-IP validation and AP2M1 trafficking experiments, support the named receptor interaction. The proposed GPCR-binding refinement describes an intracellular receptor interaction, not an extracellular agonist or antagonist. Endogenous interaction validation was limited by antibody performance, and overexpression/knockdown effects on insulin secretion do not establish intrinsic enzymatic signaling activity.
PMID:34964704: targeted actual Fig. 7 Results and cell/construct Methods were read. Endogenous human AP2M1 co-precipitates with human CLDN2 in HEK293 cells; mutation of the CLDN2 YxxΦ sequences reduces association. Human Caco-2 depletion experiments connect AP2M1 to CLDN2 internalization and degradation. This is evidence for sorting-motif recognition in its trafficking context, without claiming purified binary binding or giving μ2 LC3's activity.
PMID:33436498: the normal cache remains abstract-only. The indexed original Results, Figs. 3–4, Tables 2 and 4, and expression/fluorescence-polarization Methods were additionally inspected. Table 4 explicitly identifies human Q96CW1 residues 160–435. The purified domain binds an ATG9A 4–14 probe and ACE2 tail peptides; ACE2 Tyr781 phosphorylation inhibits binding. These two exact source pairs can be refined to signal-sequence recognition. The low-affinity ACE2 result is biochemical evidence, not proof of a necessary viral-entry route. The other ATG9A screen rows retain their own unresolved source boundaries.
PMID:24603492: the indexed original AP2-interaction Results and Fig. 3 were read. Purified AP2M1 pulls down TBC1D5 from HeLa lysate; the text separately reports a purified-protein supplementary experiment that was not independently inspected. ATG9A co-precipitates the assembled AP2/clathrin machinery in U2OS cells. These support the two positive associations, but not a single mapped μ2–ATG9A interface or an inferred GTPase-regulatory activity. No autophagy BP is proposed merely because AP2 depletion disrupts delivery.
PMID:23529131: the indexed original Fig. 3 and associated Results explicitly assay endogenous μ2 and KCNQ1 in human HT29cl.19A cells. Estrogen increases their co-IP and colocalization during internalization, with antibody/bead/IgG controls. This supports the existing transmembrane-transporter-binding assertion; candidate KCNQ1 motifs were not directly mapped by that experiment, so no more specific motif claim is inferred.
The complete cached abstracts of all 38 seeded PMID records have now been read. For large interaction screens, a paper's general assay description and an interaction list in UniProt do not establish that the exact pair's supplementary evidence has been inspected. Those original source-specific pair records remain uncertain where uninspected, including pairs independently plausible from AP-2 assembly. P63010 is AP2B1 (beta-2), not AP2S1 (sigma-2). Independent structural or targeted evidence does not silently become the original screen's experimental evidence for a new mechanism.
All 53 cached Reactome summaries were inspected as pathway statements. The cytosolic and plasma-membrane coat pools agree with actual μ2 localization and AP2 structural evidence. Their individual event descriptions do not make μ2 the kinase, lipid enzyme, GTPase or motor acting elsewhere in the reaction. Seven lysosome/endolysosome annotations remain uncertain: the short records describe cargo delivery/degradation, and one explicitly includes uncoating, without resolving whether μ2 itself remains on the destination membrane. This is an unresolved source-specific localization, not evidence of a false curated pathway.
PMID:31671891's AP-2 sections explicitly discuss postsynaptic receptor surface regulation and presynaptic cargo sorting. Its mouse models, human disease observations and mechanistic hypotheses remain distinct. PMID:11102472's bovine brain assembly assays and squid synapse intervention support conserved complex-level participation in synaptic vesicle retrieval; they are not direct human isolated-μ2 catalysis. The V-ATPase assembly observation in PMID:7989329 and LDLR peptide binding in PMID:12121421 are retained as secondary contexts, not a second autonomous AP2M1 enzyme function.
The actual PTHR10529 export was inspected, including the GO:0006896 IBD at PTN000055849 and a same-term IRD at PTN002575694. Full target-to-node topology/MSA are unavailable, so row order or taxon alone cannot assign the loss to AP2M1. GO:0006896 includes Golgi-to-lysosome transport and is not a yeast-only process. The Golgi-route assertion remains UNDECIDED; no donor-count or self-source objection is made. Other actual AP2-complex, cytosol, endocytosis and synaptic node assertions agree with independent biology, without a claim to have reconstructed the whole tree. The independent check reached the same limited conclusion from the preserved family assertion export; no target-branch loss was inferred from row order.
The proposed single core places μ2 signal-sequence recognition alongside its contribution to the assembled AP2 clathrin-cargo adaptor. It connects this work to established endocytosis and coat formation at the cytoplasmic plasma-membrane surface. It does not transfer the sigma subunit's dileucine-binding site or the beta subunit's clathrin contacts to isolated μ2. Existing qualifiers remain unchanged. No NEW assertion is proposed, and no GO-CAM target activity was found in the local index. The current official GO:0005048 definition and parent table were subsequently retrieved directly; an unverified more-specific targeting-term label is not used.
The independent all-184 prospective consultation passed the source-specific map. The initial authored review had 110 ACCEPT, six MODIFY, nine KEEP_AS_NON_CORE and 59 UNDECIDED, with no NEW annotation. The six refinements use experiments in their original cited papers: KIAA0319, HCV core, ATG9A, ACE2 and CLDN2 sorting-signal recognition, plus GLP1R binding. Separate papers do not turn unread screen pairs into new source-specific mechanisms. At that initial review stage, fifty generic screen-pair assertions remained explicitly uncertain; four directly inspected but incompletely characterized accessory/coassembly interactions remain non-core.
One core combines μ2 sorting-signal recognition with its contribution to the assembled clathrin-cargo adaptor. Receptor uptake and synaptic cargo retrieval are established contexts. LDLR peptide recognition and V-ATPase reassembly remain secondary; neither is used to invent an intrinsic μ2 enzyme activity. The Golgi-to-vacuole PAINT assertion retains uncertainty despite the inspected ancestral IBD and IRD because the target-to-node topology remains unresolved. Seven lysosome/endolysosome assertions and the exosome proteomic assertion likewise retain source-specific limits.
Every one of the 184 original source objects, ordinary qualifiers, partners and both products is preserved. No isoform collapse, source substitution or newly asserted process is introduced. All 96 required normal publication/Reactome records are present and immutable. The five actual Source36 records have been reconciled with their earlier external read scopes; the erratum is bibliographic in the normal cache and is not counted as a new biological experiment. Validation and final independent consultation follow the authored draft.
Initial review validation passed before the three HuRI pair records were inspected. At that stage, four bare-binding advisories concern the directly supported V1H, multipartner ADAM10/AP2 and ATG9A/TBC1D5 associations retained as non-core under the explicit user action definitions; genericity alone does not establish an erroneous interaction. The local term validator expects the label “signal sequence binding” for GO:0005048, whereas the current official AmiGO record still displays “signal sequence receptor activity” and lists signal-sequence binding as a synonym. Its definition covers recognition of a protein localization signal, not only an ER leader. The seeded labels and identifier are preserved, and the reviewed official label remains on the authored core/refinements. This is a label-version advisory, not a wrong-branch or invented identifier.
The automated Falcon research file was read as a source of leads; annotation support is attached to inspected original references instead. Its unused-reference advisory does not require treating automated synthesis as primary evidence. At that initial stage, COMPLETE denoted that all source assertions had been adjudicated, including 59 explicit uncertainties; it does not claim access to all pair-level experiments or complete PAINT ancestry. The final source-object, product, exact-quotation and cache-preservation checks accompany this review.
The final independent consultation identified quotations that were literal but insufficiently specific for location and postsynaptic assertions. The attachments now quote actual clathrin-pit imaging for membrane/pit localization, the locked cytosolic state for cytosol, cytoplasmic cargo recognition and coat assembly for forming carrier membranes, and the review’s postsynaptic receptor/AMPA-receptor statements for neuronal annotations. The rat rescue-reagent and complex-level mechanism limits remain explicit. This changes supporting passages only: every source object, action, rationale, product and core is unchanged. No additional paper or source-cache change was needed.
The PR review prompted a new direct check of the official GO:0005048 record. The page was successfully opened on 2026-09-28 and displays “signal sequence receptor activity”; “signal sequence binding” is a synonym. Its definition covers a short peptide directing a protein to a cellular location, without an ER-leader restriction. The page's ontology version is 2026-08-06. The earlier shell/page timeout is a historical access outcome, not the basis of the final label decision. The original source labels and current verified authored label therefore remain intact. A suggested question records whether a narrower endocytic-motif term would improve the distinction.
The seven lysosomal/endolysosomal uncertainties were rechecked against their individual Reactome summaries. Gap-junction, MHC-II, LDLR/PCSK9 and VLDLR/PCSK9 cargo destinations do not by themselves locate μ2; conversely, an event title or a general coat-removal model cannot rule out every transient destination pool. The MHC-II uncoating/fusion event does not specify exact μ2 release timing. The coat-loss event explicitly describes the timing and mechanism as incompletely understood. These are source-specific evidence gaps, not a claim that located_in requires permanent residence or a universal negative about AP2 at lysosomes. The original urinary-exosome target table also remains unadjudicated; abundance alone is not proof of contamination.
The HuRI main article is available, but the inspected cache contains neither an AP2M1/Q96CW1 target record nor the individual pair tables. Its full-text flag cannot substitute for an inspected target experiment. Other unresolved screen records retain their own specific access/assay limits. The four positive V1H, ADAM10/AP2 and ATG9A/TBC1D5 associations remain non-core: V1H is a regulatory, noncatalytic subunit; the ADAM10 row combines multiple partners; and the ATG9A/TBC1D5 source distinguishes complex co-precipitation, lysate pull-down and the separately reported supplement. No isolated μ2 interface or other paper's motif assay is silently assigned to these rows. Under the explicit user ActionEnum, genericity alone does not demonstrate an incorrect experimental annotation.
Cytosolic Reactome reasons now identify the particular cargo, recruitment, phosphorylation, actin-coat or uncoating event. This first prose revision changes no action, source object, qualifier, product, core function, supporting quotation or biological assertion. The separately checked HuRI update below subsequently changes three actions. Public notes provide the read conclusions and primary links directly; private execution receipts are not required to understand those conclusions.
The primary HuRI portal tables subsequently supplied three previously uninspected AP2M1 interactions. Independent searches exposed actual indexed interaction rows: MAB21L3–AP2M1 (HI-III, score 0.823), PRPF38A–AP2M1 (HI-III, score 0.888), and EAF1–AP2M1 (HI-III, score 0.809). The last pair appears among the neighboring interactions in the MAB21L3 network. The immutable UniProt record maps these partners to Q8N8X9, Q8NAV1 and Q96JC9, respectively.
The original HuRI article describes a human ORFeome screen with three Y2H versions and pairwise retesting/sequence confirmation. The observed HI-III pair records support retaining these three existing interactions as non-core. They do not establish an endocytic sorting motif, a cargo-to-coat bridge, endogenous colocalization or an isoform-specific interface. The portal scores are recorded as metadata, not interpreted as physiological evidence. Direct page opens returned an empty JavaScript shell; the positive evidence came from the indexed primary tables, whose crawl metadata was three months old. Raw PSI-MI/transcript-level construct and individual retest records were not retrieved, and no claim is made that these exact pairs were each confirmed in every orthogonal assay.
There are 25 original PMID32296183 rows in this review. Three are now KEEP_AS_NON_CORE; the other 22 retain their source-specific uncertainty. Overall, the 184 original assertions now have 110 ACCEPT, six MODIFY, 12 KEEP_AS_NON_CORE and 56 UNDECIDED; no NEW assertion or core change is added. All original evidence codes, supporting accessions, qualifiers and two products remain intact.
The EAF1–AP2M1 rationale now explicitly identifies the observed row as a neighboring-network interaction on the MAB21L3 portal page. An EAF1-specific page was not inspected. The pair identifier, HI-III score and limitations are unchanged; neither the network context nor the score establishes a physiological or cargo-recognition mechanism.
The current review retains 110 ACCEPT, six MODIFY, 12 KEEP_AS_NON_CORE and 56 UNDECIDED across all 184 original assertions. Seven positive generic-binding assertions remain non-core. The validation rule recommends different actions for generic binding, but the explicit user ActionEnum defines REMOVE as unlikely correct on combined evidence and permits KEEP_AS_NON_CORE. Here, inspected positive interactions do not become incorrect because they lack an established specific molecular function; returning them to uncertainty solely to satisfy that advisory would also misrepresent the inspected evidence. The remaining unadjudicated source-pair assertions retain their actual evidence gaps. This is an explicit policy disagreement, not an unaddressed validator result.
Earlier count and advisory paragraphs are now marked as historical. The previously verified current GO:0005048 label remains unchanged; a broad ontology-cache refresh is separate from this source-provenance clarification. No action, source object, supporting quotation, core function, product or source cache changes in this follow-up.