Automated deep research was attempted with just deep-research-falcon human MS4A6A --fallback perplexity-lite, but the run timed out before producing a deep-research file. This review therefore uses cached GOA publications, the UniProt record, and the PANTHER family fetch.
MS4A6A encodes membrane-spanning 4-domains subfamily A member 6A, a CD20-like four-pass transmembrane protein in the MS4A family. UniProt describes the protein as a membrane, multi-pass membrane protein with four predicted transmembrane helices, and the local PANTHER fetch places it in PTHR23320:SF135, within the MS4A/CD20-like family.
The full-text B-cell expression and localization paper provides the strongest direct evidence for compartment assignment. It states that the study cloned human MS4A6A as a GFP fusion for BJAB localization assays PMID:23874341. In those assays, GFP-MS4A6A did not behave like CD20 or MS4A4A at the plasma membrane; instead, "GFP-MS4A6A and GFP-MS4A7 were localized intracellularly in the perinuclear space" and "Both GFP-MS4A6A and GFP-MS4A7 co-localized with giantin, indicating that they were concentrated in the trans-Golgi complex" [PMID:23874341 "GFP-MS4A6A and GFP-MS4A7 were localized intracellularly in the perinuclear space"; PMID:23874341 "Both GFP-MS4A6A and GFP-MS4A7 co-localized with giantin, indicating that they were concentrated in the trans-Golgi complex"].
The same paper argues against treating MS4A6A as a normal plasma-membrane CD20 partner in B cells. It reports that "MS4A6A and MS4A7 transcripts were detected in most B-cell samples but at two to four orders of magnitude lower than MS4A1 (CD20)" and concludes that "MS4A6A and MS4A7 transcripts were detected in most normal B samples but at very low levels and the corresponding proteins are probably not normally expressed at the plasma membrane" [PMID:23874341 "MS4A6A and MS4A7 transcripts were detected in most B-cell samples but at two to four orders of magnitude lower than MS4A1 (CD20)"; PMID:23874341 "MS4A6A and MS4A7 transcripts were detected in most normal B samples but at very low levels and the corresponding proteins are probably not normally expressed at the plasma membrane"].
For curation, retain membrane and trans-Golgi network annotations as the best-supported functional surface. Treat the plasma-membrane IBA as a likely over-propagated family inference and propose replacement with trans-Golgi network, which is already directly supported by IDA evidence. Treat the broad cell surface receptor signaling pathway IBA as over-annotated because no resolved MS4A6A receptor/signaling mechanism is established in the cached evidence. The generic protein binding annotation from the HuRI binary interactome paper should also be marked over-annotated: PMID:32296183 is a valuable high-throughput interactome resource, but for MS4A6A it does not define an informative molecular function beyond a single binary interaction.
The second-pass audit confirmed the existing MS4A6A review and manual reference metadata. No annotation action changes were needed: MS4A6A remains curated as an intracellular/trans-Golgi multi-pass MS4A-family membrane protein, with plasma-membrane and receptor-signaling family transfers treated as over-propagated relative to the direct cached localization evidence.