DLG4 (human P78352; HGNC:2903) has cooperating synaptic receptor-adaptor and structural postsynaptic-density roles. PDZ receptor recognition, SH3–GK recruitment and postsynaptic structural support are cooperating mechanisms. Binding a catalytic partner does not give PSD-95 that partner's activity. The official products are P78352-1 (724 residues), P78352-2 (767) and P78352-3 (721); historical constructs and domain fragments are not assumed to test all three products.
The fragmentomics results and executable analysis expose the original measurement values, exact source partner identities, peptide modifications and sequence-selection rules. The reproduction instructions use the published PMID:36115835 workbook and the saved official sequence inputs.
The workbook has 1,344 DLG4 measurements. Accession matching yields 1,197 candidate crosswalk records: 1,179 natural-sequence matches and 18 unmatched natural-sequence candidates. Three explicit AXIN1 internal peptides are additionally sequence-supported, giving 1,182 selected records. Of 388 source pairs, 248 have a nonzero selected composite and 140 have only zero selected composites. Zero is neither zero affinity nor proof of noninteraction. Unmatched candidates, including particular viral phosphopeptides, are not silently used as identity support.
Most positive isolated-domain interfaces remain non-core binding observations because their physiological role is not resolved. The exact project curation instruction supports retaining positive generic binding when an informative narrower function is not justified. Peptide reagents do not by themselves establish a physiological peptide-binding function. Supported specificity uses MODIFY, with the imported term and supporting entities retained.
Repeated partners were compared across sources. DLGAP1 scaffold recognition and its targeted phosphoprotein assay describe compatible aspects; SPAR's GAP class and phosphorylation-dependent recognition are likewise compatible. ERBB4 has independent receptor-tyrosine-kinase binding evidence despite the expression of concern on one later paper. APBA1 has independent BioPlex co-complex evidence, while its isolated PDZ-tail composites remain zero. DLGAP4 has independent GK-fragment recognition at residues 546–724; that does not convert its zero PDZ-tail measurements into positive results. Source-specific uncertainty is retained where no independent positive evidence resolves a zero-only fragment assertion.
| Source | Actual evidence used | Access and scope |
|---|---|---|
| PMID:22117215(https://pmc.ncbi.nlm.nih.gov/articles/PMC3243629/) | Human PSD-95 SH3–GK 426–724 and GK 531–724 recognize phosphorylated DLGAP1, SPAR and LGN ligands; Results and Figure 5/Methods | Full original read. The crystal is SAP97/phospho-LGN; it is not a PSD-95 crystal. Domain assays are not tests of every natural full-length product. |
| PMID:12151521(https://pmc.ncbi.nlm.nih.gov/articles/PMC6758133/) | Alpha/beta targeting, human beta N-terminal cDNA insert, rat neuronal cultures, COS-cell channel-clustering comparison | Primary indexed Methods/Results were separately recovered; the normal cache remains abstract-only. The species of the complete pre-existing alpha backbone was not established. |
| PMID:7477295(https://pubmed.ncbi.nlm.nih.gov/7477295/) | Shaker-family channel-tail recognition and clustering | Abstract and subscription preview only; original Methods and every partner panel were not recovered. Curated pairs and independent targeted evidence remain distinct. |
| PMID:17046693(https://pubmed.ncbi.nlm.nih.gov/17046693/) | PSD-95/PSD-93 participation in AMPA-receptor targeting at mature synapses | Complete abstract read; no full-original Methods claim. |
| PMID:21525273(https://pubmed.ncbi.nlm.nih.gov/21525273/) | Rat-neuron tomography and knockdown establish PSD-95-containing structural elements and PSD maintenance | Selected original Results/Methods read separately from the abstract-only normal cache. |
| PMID:10725395(https://pubmed.ncbi.nlm.nih.gov/10725395/) | Controlled ERBB4/PSD-95 association and PDZ dependence | Selected original read. The initial owner cache was abstract-only; the published file reuses the independently reconciled, unchanged genuine full-text DLG3 cache. |
| PMID:21653829(https://pubmed.ncbi.nlm.nih.gov/21653829/) | Defined neuronal protein-fragment interactions, including the DLG4 GK interface | Selected Methods and literal target records read. Validation of a study subset is not asserted for every pair. |
| PMID:30126976(https://pubmed.ncbi.nlm.nih.gov/30126976/) | Short human tails, phosphomimetic libraries and isolated PDZ reagents | Original design and target records distinguish reagent recognition from full-length physiology. |
| PMID:33961781(https://pubmed.ncbi.nlm.nih.gov/33961781/) | APBA1-associated AP-MS complex | Original AP-MS Methods and literal target record read; co-complex association is not purified binary proof. |
| PMID:20531396(https://pmc.ncbi.nlm.nih.gov/articles/PMC2927054/) | Receptor-specific calcium-response sign, Figure 3, shRNA Results and Fura-2 Methods | Primary indexed excerpts support negative regulation in the 5-HT2A assay. The normal cache remains abstract-only; no neuronal culture species or global calcium effect is inferred. |
| PMID:26609151(https://pubmed.ncbi.nlm.nih.gov/26609151/) | KCl-associated PSD-95 degradation | Selected original experiments show a regulated abundance change without resolving a DLG4 contribution to the potassium-response process. This is not a universal rule against response-term inference from expression evidence. |
| PMID:21920314(https://pubmed.ncbi.nlm.nih.gov/21920314/) and PMID:23397623(https://pubmed.ncbi.nlm.nih.gov/23397623/) | Positive mouse epithelial Dlg4 junctional localization versus a review's Dlg1/SCRIB polarity mechanisms | Cell-junction location is supported; the specific DLG4 epithelial-polarity assertion remains uncertain. There is no blanket exclusion of epithelial function. |
Additional selected originals support KIF1Balpha tail binding (12097473), receptor-associated complexes (17360663, 19243221, 22593058), BCR/ABR recognition (20962234), MTMR2 lipid-phosphatase recognition (20410104), MDM2 association (23260144), ADAM22 tail dependence (27066583), P2Y1 and D1 receptor contexts (20847060, 23041629), and presynaptic/juxtaparanodal pools (8922396, 19109503). Each reference assessment records its actual scope; no abstract-only entry is silently relabeled as a full-text cache. The DGK-iota paper (21119615) was accessed as primary indexed excerpts and supports lipid-kinase binding, not protein-kinase-specific recognition.
PMID:38114668(https://doi.org/10.1038/s41591-023-02757-y) is an expression of concern for PMID:16767099(https://pubmed.ncbi.nlm.nih.gov/16767099/). It identifies immunoblot-image concerns and unavailable original gels. The affected reference is DISPUTED, not retracted or marked invalid. Independent ERBB4 association evidence is assessed on its own merits.
The two cached GO-CAM models already place DLG4 in adaptor roles: 62900b6400001630 uses synaptic receptor adaptor activity in the intracellular postsynaptic density with rat orthology support, and 67b1629100002554 uses macromolecule adaptor activity in the LGI1–ADAM22 context with mouse support. They do not make DLG4 a receptor, channel or enzyme.
The exact human Reactome paths distinguish target location from the activity of an entire complex. Public entity records are accessible through the Content Service:
| Event | Target-resolved path and interpretation |
|---|---|
| AMPA trafficking R-HSA-416320 | 202135 is P78352 at plasma membrane; 416847 is the endocytic-vesicle membrane pool. |
| Calcium influx R-HSA-432164 | Catalyst/complex traversal reaches P78352 entity 202135; PSD-95 is not the channel pore. |
| Kainate binding R-HSA-451283, activation R-HSA-451311 | Receptor complex components reach 202135; target complex membership is distinguished from receptor chemistry. |
| CIT-3 R-HSA-5671993, RAS exchange R-HSA-5672965 | Input/output or catalyst-complex paths reach 202135. No kinase or exchange-factor activity is transferred to PSD-95. |
| LGI1 R-HSA-5682794 | ADAM22/PSD-95/stargazin complex 5682831 contains the P78352 membrane entity. |
| ABP1/SHANK R-HSA-6797554 | DLG2/3/4 set 6794325 includes P78352 cytosolic entity 6794267. |
| PSD-95/NMDA R-HSA-8849878, SALM/PSD-95 R-HSA-8849891, SALM1/NMDA R-HSA-8849906 | Direct or complex paths reach 202135, through the PSD-95-family set 376003 or complex 8849869 where applicable. |
Thirteen PAINT assertions were traced to literal IBD nodes on the human target's ancestral path, including PTN000563569, PTN001136892, PTN004609043, PTN002750666 and PTN008381197. A DLG1/SCRIB source can support an ancestral function; it is not automatically a paralog error. A short donor list and target self-evidence are not reasons to reject inheritance. Donor function, target biology and term scope are evaluated separately. Six ARBA conditions were matched to actual target CATH FunFam/InterPro assignments rather than inferred from a rule name.
The specific endoplasmic-reticulum, synaptic-vesicle, cortical-cytoskeleton, mossy-fiber-rosette, dendritic-arborization and epithelial-polarity assertions remain uncertain where the available target-resolved evidence does not settle them. These are source/term-specific limits, not denials of established synaptic biology. No NEW annotation is proposed. The follow-up explicitly represents the complementary structural molecular role alongside receptor-adaptor activity.
The fragmentomics Methods distinguish DAPF (double affinity-purified PDZ protein with fluorescence readout) normalized binding-intensity threshold 0.05 from LYSC/SAPF threshold 0.1. Effective Kd limits vary with peptide concentrations (30–790 µM, mean 320 µM); the smallest nonzero workbook composite is not a universal biological cutoff. The composite chooses a holdup method and does not replace independent fluorescence polarization. Of 729 zero composites across 1344 measurements, 380 have negative raw DAPF intensities. Four zero composites have positive FP measurements elsewhere in the grid, but none belongs to the selected measurements of the 138 uncertain source rows. All 414 selected measurements for those 138 rows use DAPF and remain below 0.05. Negative intensity is an assay estimate relative to a control; it is not negative affinity or a declaration of nonbinding. See PMID:36115835(https://pmc.ncbi.nlm.nih.gov/articles/PMC9482650/), Methods “Estimation of binding intensities”/holdup affinity calculations and Discussion on below-quantification pairs. The unchanged public workbook artifact exposes the values and exact sequence joins.
The structural core is grounded in PMID:21525273(https://pubmed.ncbi.nlm.nih.gov/21525273/), Materials and Methods (rat hippocampal cultures, PSD95-EYFP constructs, immunogold/electron tomography), Results on PSD95-containing vertical filaments and their loss after RNAi, and Discussion on structural maintenance. These are E20 rat neurons; only a subset of filaments is immunolabeled, and receptor-type classifications are morphological/putative. The source does not imply universal loss of NMDA-receptor clusters. It supports physical structural participation, not simply a gene-knockdown phenotype. Current GO:0098919 has a part_of edge to GO:0099562. The latter's current definition contains a presynaptic/postsynaptic wording discrepancy; the proposal uses its postsynaptic label, linked structural activity and actual source, not a presynaptic claim.
The PMID:26609151 original, Results “Enhanced activity-dependent degradation” and Figure 7A–C, reports: “Our results confirmed that PSD-95 is degraded in an activity-dependent manner”. This eleven-word original-only anchor is distinct from the abstract-only normal cache. The actual experiment uses mouse hippocampal neurons, 50 mM KCl for 15 min and proteasome inhibition; the significant additional genotype effect is on PSD-93. Accordingly the two potassium-response judgments retain the real abundance observation but do not infer a performed response step from being degraded. No original-only quotation is placed in cache-validated YAML.