DSE directly changes uronic-acid stereochemistry within chondroitin chains, generating iduronic-acid-containing dermatan sulfate. One catalytic core combines chondroitin-glucuronate 5-epimerase activity with dermatan-sulfate proteoglycan biosynthesis at the Golgi membrane. The positive ER-associated pool remains explicit in the biological description and annotations. The source guide links the original evidence and distinguishes repository cache availability from separate original access.
PMID:16505484(https://pubmed.ncbi.nlm.nih.gov/16505484/) identifies the bovine spleen enzyme and separately assays sequence-verified human full-length SART2/DSE with a C-terminal His tag. Tritium release and carbon-14 product mapping establish the GlcA-to-IdoA reaction on chondroitin. Metabolic labeling additionally shows increased iduronic acid and iduronate-rich blocks in cellular CS/DS chains. DSE therefore performs the chemical modification itself; its biosynthetic role does not rest only on necessity or a disease phenotype.
PMID:19004833(https://pubmed.ncbi.nlm.nih.gov/19004833/) tests tagged human wild-type and mutant proteins and shows that N-glycosylation supports activity. Its proposed 2009 active-site mechanism is a model, not the same evidential claim as the measured net epimerization. The later human structure in PMID:33815739(https://pmc.ncbi.nlm.nih.gov/articles/PMC8006597/) revises that model and identifies a metal-binding environment. The structure is not a substrate cocrystal. DSE's own glycosylation does not make it a glycosyltransferase, and a proposed transient bond-cleavage step does not establish net glycan-degrading lyase activity. These features belong to the same catalytic mechanism rather than separate core functions.
The electronic heparan-sulfate pathway assignment is rejected on a positive substrate distinction. The original PMID:7092807(https://pmc.ncbi.nlm.nih.gov/articles/PMC1163673/) compares the dermatan/chondroitin and heparan epimerase systems under their respective substrate, pH and cofactor conditions. EC 5.1.3.19 and RHEA:21084 describe the DSE reaction; GLCE/O94923 instead acts on N-sulfated heparosan under EC 5.1.3.17. The current DSE UniProt pathway statements do contain the suspect heparan assignment, so it is documented as a target-source mapping error rather than silently ignored. The heparan pathway itself is valid. The Rhea chondroitin 4′-sulfate equation does not imply an exclusive sulfation requirement, because primary DSE assays also use unsulfated chondroitin.
The original PMID:10679095(https://pubmed.ncbi.nlm.nih.gov/10679095/) studies human SART2/DSE fused C-terminally to GFP. COS7 microscopy and sucrose-gradient fractionation support a predominant ER and partial Golgi distribution. Human MCF-7 transfectants show a similar organelle pattern but lack the vesicle pattern seen in COS7. These are positive tagged-human-protein observations in specified hosts, not direct measurements of untagged native occupancy. Broad ER and Golgi assertions are refined to their supported membrane compartments while contextual vesicle annotations are retained as non-core. Hydrophobicity predictions are not experimentally resolved membrane topology.
All five scanned pages of PMID:7092807 were inspected. Its enzyme preparation is from cultured human skin fibroblasts, whereas bovine, chick and rat materials have other substrate or reagent roles. More than 90% of activity is recovered in membrane fractions; the highest specific activity is in the microsomal pellet, and detergent plus salt solubilizes a substantial fraction. This supports membrane association but does not separate ER from Golgi. The study predates molecular DSE/DSEL identification, so later recombinant human DSE work supplies the gene-level connection. The HPA XML inspected for DSE contains no subcellular assay section and supplies no additional localization claim.
The official Q9UL01 record contains one displayed 958-residue sequence and no alternative-products block. Signal peptide 1–22 is sequence-predicted; the annotated processed chain spans 23–958. The soluble 23–775 and 23–894 proteins used in the structural study are experimental constructs, not natural splice products. Only the 23–775 construct yielded the catalytic-domain crystal; the extension toward the membrane used cross-link-constrained modeling. No distinct natural-product function or obligatory oligomer is inferred.
The PTHR15532 target alignment matches all 958 residues of Q9UL01. Its human leaf PTN002496188 descends from the recorded IBD node PTN001036588 along a 17-node path. Three literal IBDs cover epimerase activity, dermatan-sulfate metabolism and chondroitin-sulfate metabolism. Their source identities are preserved: bovine P0C2H4 and human Q9UL01 provide target-family activity grounding, while MGI:2442948/Q0VBN2 is mouse Dsel, not mouse Dse. A shared ancestral function can be supported across those paralogs; neither donor count nor legitimate target self-inclusion is a failure.
The complete abstract of PMID:22496542(https://pubmed.ncbi.nlm.nih.gov/22496542/) reports reduced brain CS/DS iduronic acid and altered sulfation after mouse Dsel loss. Full original access was unavailable, so no unseen assay or supplementary detail is attributed to that source. The direct human DSE reaction and cellular chain modifications independently support the target's biosynthetic function. The broad dermatan metabolism IBA is refined to biosynthesis; chondroitin-chain metabolism remains valid because DSE chemically modifies that chain rather than merely consuming an unrelated substrate.
The actual ARBA00095326 rule matches both required FunFam identifiers and the Primates lineage in the official target record. Reactome's DS-GAG pathway contains the epimerization reaction, whose catalyst links through the DSE/DSEL alternative-enzyme set to Q9UL01 at the Golgi membrane. That set does not establish a required DSE–DSEL complex. The inspected GO-CAM index has no DSE activity; absence alone is not an annotation gap.
All 21 imported assertions retain their complete source dictionaries and qualifiers: 15 are accepted, three are refined, two contextual locations remain non-core and one electronic pathway assignment is rejected. No new annotation is proposed. Existing publication caches and access flags remain unchanged. The 1982 scanned original, the 2000 indexed localization sections and the independently inspected enzymology originals are distinguished from their abstract-only normal caches. The 2021 additional cache contains genuine full XML; the 2012 donor cache remains abstract-only.
Genuine Falcon research failed authentication, and its configured PerplexityLite fallback failed with HTTP 401 and an insufficient-quota response. No provider report was produced or substituted with authored text. The review is based on the primary evidence, official records and documented independent source consultation. One seven-word publication excerpt appears once in the core; these notes add no quotations.
The same epimerization mechanism directly modifies chondroitin chains and generates dermatan regions. The core now includes both GO:0050654 chondroitin sulfate proteoglycan metabolism and GO:0050651 dermatan sulfate proteoglycan biosynthesis. Current typed is_a/part_of ancestry places GO:0050651 below GO:0050655, but neither GO:0050651 nor GO:0050654 is an ancestor of the other. The two process descriptions therefore cover different aspects of the enzyme's demonstrated chemical work, without adding a second activity or a new annotation. The cached human-expression result includes a “22-fold increase in epimerase activity in 293HEK cell lysate.” PMID:16505484(https://pubmed.ncbi.nlm.nih.gov/16505484/) The original Methods and product analysis distinguish the human construct from bovine enzyme discovery.
Both ER-membrane assertions are now retained as non-core. This preserves the positive tagged-human-protein ER pattern and ER-marker cofractionation, including the similar ER/Golgi distribution in MCF-7 cells. It does not recast that pool as false, transient or background. The original localization experiment does not measure epimerization in the ER; the historical human-fibroblast microsomal assay also does not separate ER from Golgi. The compact core uses the target-resolved Golgi catalytic compartment. The ER-to-ER-membrane refinement remains valid as a location statement without asserting ER-localized catalytic work. This is a source-specific core distinction, not a general requirement for organelle-resolved catalytic assays. PMID:10679095(https://pubmed.ncbi.nlm.nih.gov/10679095/), PMID:7092807(https://pmc.ncbi.nlm.nih.gov/articles/PMC1163673/), Reactome reaction.
Manual reference availability is recorded separately from immutable repository-cache coverage. Genuine scanned/manuscript/XML originals were recovered for PMID:7092807, PMID:16505484 and PMID:33815739; indexed original text was accessible for the documented PMID:10679095 and PMID:19004833 sections. Those original reads do not change abstract-only cache flags or claim unexamined PDFs/supplements. PMID:22496542 remains full-text unavailable, while its authentic abstract supports the limited mouse Dsel finding and verified citation identity. Separate full-text support fields identify the few original-only fragments; their public section locators are in the source guide.
The earlier review/access paragraph is preserved as history. The effective dispositions are now 13 ACCEPT, three MODIFY, four KEEP_AS_NON_CORE and one REMOVE across the same 21 source assertions. This append adds one nine-word cached quotation; all other new excerpts occur only in the YAML. Including the inherited seven-word core excerpt, aggregate quoted words are 20 for PMID:16505484, 22 for PMID:10679095, 20 for PMID:19004833, 18 for PMID:7092807, 19 for PMID:33815739 and 10 for PMID:22496542. No source field, qualifier, product identity, source cache or earlier note was changed.