# BLOC1S6 / pallidin — research and annotation notes

## 2026-10-01: initial manual review

Human pallidin is a small coiled-coil BLOC-1 subunit with a cytoplasmic/peripheral membrane distribution. The strongest mechanistic evidence is its contribution to the BLOC-1 assembly and to endosomal delivery of membrane cargo. Human pallidin forms a stable subassembly with Cappuccino and BLOS1; human PaHA complementation of pallid mouse melanocytes restores pigmentation and Tyrp1 sorting. These findings support an assembly-mediated cargo-trafficking role, not an autonomous motor, membrane-spanning transporter or catalytic enzyme.

All 177 imported source annotations, including qualifiers and supporting entities, are retained. The three `alternative_products` are also retained; `products` is absent, and Q9UL45-3 remains **Not described**. No isoform-specific function or new GO annotation is proposed. There are 26 ACCEPT, 101 KEEP_AS_NON_CORE and 50 UNDECIDED decisions. The one core represents the BLOC-1 structural adaptor and lysosome-related-organelle cargo pathway.

### Assembly, binding and localization

- [PMID:10610180](https://pubmed.ncbi.nlm.nih.gov/10610180/) describes the pallid mouse gene, platelet storage-pool deficiency, pigmentation and otolith phenotypes, and syntaxin-13 association by yeast two-hybrid and co-immunoprecipitation. The complete cached abstract was read; the complete article and every construct/species detail were not. This is positive target evidence for the existing syntaxin-binding and otolith annotations.
- [PMID:12019270](https://pubmed.ncbi.nlm.nih.gov/12019270/) establishes pallidin-containing BLOC-1, a soluble/peripheral membrane distribution and self-association. Besides its complete cached abstract, selected indexed original Methods, Results and Discussion were read, including human constructs, HeLa localization and F-actin sedimentation. Cytosolic pallidin-containing material associates with F-actin, but the authors leave direct binding versus mediation by other proteins unresolved. The tagged overexpression localization preferentially marks perinuclear actin and does not establish a general actin-crosslinking mechanism. Actin associations are retained non-core. The stress-fiber, focal-adhesion and cell-contact assignments remain U pending their exact mouse assays and ancestral context, not because an uninspected donor is assumed wrong.
- [PMID:12191018](https://pubmed.ncbi.nlm.nih.gov/12191018/) and [PMID:15102850](https://pubmed.ncbi.nlm.nih.gov/15102850/) independently establish human pallidin in a heterooligomer and the eight-component BLOC-1 assembly. Both canonical caches are abstract-only; pair-level supplementary matrices were not independently read.
- [PMID:22203680](https://pubmed.ncbi.nlm.nih.gov/22203680/) reconstitutes a human octamer with one copy of each component and identifies a stable pallidin–Cappuccino–BLOS1 trimer. Selected Methods, Results and Discussion of the authenticated full-text variant were read separately; the canonical abstract cache was preserved. Isolated pallidin–BLOS1 and pallidin–Cappuccino pairs were tested. Whole-octamer AP-3 pull-down is distinguished from an isolated pallidin–AP-3 interface. The one-copy octamer does not refute self-association in isolated-protein assays; those self-binding and homodimer observations are retained non-core.
- The normal [Reactome R-HSA-429815](https://reactome.org/content/detail/R-HSA-429815) summary describes BLOC-1 formation. The cytosolic annotation is also grounded in pallidin fractionation, not inferred from the event title alone.

Generic protein-binding rows retain their exact experimental source and partner identifiers. The 69 generic IPI annotations are non-core physical associations, with different evidence limits for small-scale assembly studies, binary screens and AP-MS. An interaction does not transfer the partner's function to pallidin, and AP-MS does not necessarily establish direct binary contact. The user-defined action rules take precedence over a blanket policy to remove generic binding. No experimental association is rejected merely because its cache is abstract-only.

### Cargo sorting, pigment and neuronal context

[PMID:17182842](https://pubmed.ncbi.nlm.nih.gov/17182842/) directly uses human PaHA to complement pallid mouse melanocytes. Selected authenticated full-text Methods, Results and Discussion establish restoration of pigmentation and Tyrp1 routing from early endosomes to melanosomes. The original canonical cache remains abstract-only; this distinction is retained in `reference_review`. The inspected body separately maps MuHA-reconstituted BLOC-1 to tubulovesicular endosomes. No pallidin-specific supplementary localization image was independently inspected. Thus the precise transport-vesicle localization remains U, despite strong evidence for membership of the functioning complex.

Whole-melanosome movement (GO:0032402) and melanosome positioning are different from Tyrp1 cargo delivery (GO:0035646). The former annotations remain U; the latter are accepted. Likewise, restoration of pigment in an established melanocyte line does not by itself resolve positive regulation of pigment-cell differentiation. These are evidence limits, not declarations that experimental annotations are erroneous.

The [2017 correction, PMID:28137951](https://pmc.ncbi.nlm.nih.gov/articles/PMC5341732/), invalidates the original transferrin-FITC control. Repeat experiments using corrected reagent show a mild early recycling defect in the mutant contexts. The review does not repeat the original claim of unchanged transferrin recycling or generalize the control failure to all Tyrp1 experiments. The notice was read through official primary content; figure images and supplementary datasets were not independently assessed.

[PMID:19546860](https://pubmed.ncbi.nlm.nih.gov/19546860/) directly tests primary pallid mouse hippocampal neurons, despite its title foregrounding dysbindin. Selected complete relevant Methods, Results and Discussion show modest neurite-number/length changes. Native brain BLOC-1 detected by anti-pallidin associates selectively with SNAP25 isoforms and open syntaxin conformations; those assays are not isolated-pallidin contact maps.

[PMID:21998198](https://pubmed.ncbi.nlm.nih.gov/21998198/) includes actual pallid fibroblasts and dentate-gyrus cargo measurements, plus human SH-SY5Y complexes containing endogenous pallidin. Selected Results and Methods were read. Other experiments also use dysbindin, muted or AP-3 mutants. The neuronal cargo delivery annotations are retained non-core without a motor claim. A preparation of small intracellular vesicles from homogenized PC12 cells does not establish localization to extracellular microvesicles (GO:1990742); that transferred location remains U.

For the fly-grounded IBA annotations, independently inspected primary [PMID:28317021](https://pmc.ncbi.nlm.nih.gov/articles/PMC5356223/) describes Pldn-mutant high-activity recycling-pool defects and endosomal changes, with relatively preserved basal transmission. The selected text also places the donor protein in postsynaptic muscle as well as neuronal structures. This supports non-core retention of synaptic-vesicle recycling and neuromuscular postsynapse context. It is not a claim of direct human localization, full figure coverage, or reconstruction of the complete PAINT tree. Target self-entries in other IBA donor sets are legitimate descendant experimental grounding, not circularity; donor count is not used to judge validity.

### Mouse metabolic, immune and unresolved donor effects

ROOT's bounded independent consultation identified and read the full official abstract and selected indexed Methods, Results and Table 1 of [PMID:28701731 / PMC5507893](https://pmc.ncbi.nlm.nih.gov/articles/PMC5507893/). It directly compares pallid and control postnatal day 1–2 mouse hippocampi and reports 28 changed metabolites, including amino acids, adenosine and phospholipid derivatives. Rows 87–91 and 150 are retained non-core as downstream consequences. This does not identify pallidin as an amino-acid, nucleoside or lipid enzyme. The exact current donor GO-row/reference edges were not recovered, so this study is explicitly corroborating evidence, not an asserted reconstruction of every transfer.

Science's separate seven-row consultation read direct pallid-mouse neutrophil experiments in [PMID:17785837](https://pubmed.ncbi.nlm.nih.gov/17785837/), including selected Methods, Results and Figure 3 legend. Deficient inflammatory neutrophils release less elastase and provide less support for parasite control. Rows 92/93 are retained non-core. ROOT's independent whole-review check identified that the measured elastase release also supplies positive evidence for row 101, protein secretion; this existing assertion is likewise retained non-core. Pallidin is not the elastase, TLR4 or oxidant-generating enzyme used elsewhere in the paper. The exact current mouse annotation-source edges remain unresolved. No full figure-image or supplement claim is made.

Positive redox evidence is acknowledged: selected indexed Results and Figure 7 caption in [PMID:22423091 / PMC3313842](https://pmc.ncbi.nlm.nih.gov/articles/PMC3313842/) describe increased basal oxidant signal and decreased peroxiredoxins after pallidin shRNA in human SH-SY5Y cells. This is not automatically a demonstrated *response to oxidative stress*, so row 94 remains U rather than denying the phenotype or assigning peroxidase activity.

A [2005 primary Regensburg thesis](https://epub.uni-regensburg.de/10341/1/Salim%20Maa%20Bared_Dr_Arbeit.pdf), selected Results pages 87–88 and Figure 40 caption, reports pooled sera from four pallidin-knockout and four control mice with altered HDL lipids and VLDL triglycerides. Lipid composition, homeostatic maintenance and lipoprotein-particle numbers are distinct claims. Rows 162/166/171 remain U because the exact current donor sources and adequate process-level interpretation remain unresolved, not because there is no positive lipid evidence. Proposed LCAT or secretion explanations are not pallidin catalytic results. The consultant's attempted screenshot failed; no figure-image reading is claimed.

The recovered [MGI Bloc1s6 graph](https://www.informatics.jax.org/marker/gograph/MGI%3A1927580) is explicitly generated in 2023. It and the current summary identify the legitimate mouse Bloc1s6/Pldn donor. Its missing rows do not prove current annotation absence, a wrong donor, or confusion with another protein. Spatial/developmental/behavioral assignments whose exact assay remains inaccessible are left U, including ER lumen. Cytosolic and peripheral membrane localization do not independently establish an intraluminal pool or rule out every additional pool.

### Human disease and citation quality

The UniProt warning was investigated independently. [PMID:21665000](https://pubmed.ncbi.nlm.nih.gov/21665000/) is the retracted 2011 human HPS9 paper; the [retraction PMID:28475864](https://pmc.ncbi.nlm.nih.gov/articles/PMC5985258/) concerns duplicated Figure 3C material. This is not generalized to every BLOC1S6 disease association. The separate [2012 report PMID:22461475](https://pubmed.ncbi.nlm.nih.gov/22461475/) has no PubMed abstract. Indexed selected original published text reports homozygous p.Q78X, deficient pallidin, abnormal patient NK-cell CD107a behavior and rescue after PLDN replacement, while platelet aggregation was normal. Direct full-paper opening was unsuccessful, and the review does not claim complete article, image or supplement inspection. These uncached sources remain readable primary links in notes rather than fabricated cached references or quotation-bearing formal entries.

The official [correction PMID:41039152](https://pubmed.ncbi.nlm.nih.gov/41039152/) to PMID:40205054 corrects a duplicated loss-function equation; the notice does not name a change to the pallidin–dysbindin interaction. This is a bounded correction assessment, not a claim to reanalysis of that dataset.

### Research provenance and validation boundary

The standard Falcon command with `perplexity-lite` fallback was attempted once after the seed import, using writable UV directories. Both provider startup paths failed during offline dependency resolution for `deep-research-client[cyberian]==0.2.7rc1`; neither provider ran and no research output was created. No retry or manually authored provider-named file was made. The notes above are manual research, using the normal immutable caches, separately identified archived variants and primary web evidence.

All 15 cached PMID abstracts and the Reactome summary were read; the larger primary-paper reading scope and limits are stated per reference above. The 10610180 and 12019270 caches were imported unchanged before final reference validation. The 14 existing auxiliary sources were preserved and both family files stayed absent. No family summary is treated as a PAINT reconstruction.

Only four short exact anchors were added to the YAML: 5 words from PMID10610180, 12 from PMID12019270, 5 from PMID22203680 and 5 from UniProt Q9UL45. No publication quote is repeated in these notes. Existing BLOC1S1/BLOC1S3 quotations from PMID22203680 total 18 words; this additional 5-word anchor brings the shared allocation to 23 words. All other consulted sources are paraphrased with their reading limits. No new publication cache or formal uncached PMID entry was manufactured.

At the initial proposal boundary on 2026-10-01, this review was prepared in TMP for independent scientific review; canonical curated content and history were unchanged. Application and subsequent validation are recorded as later journal entries rather than inferred from this preparation statement.

### Focused validation completed

The normal validator passed with reference-title, exact-quotation, schema and GOA checks enabled. It reports 74 warnings: 69 reflect the repository's generic-binding policy versus the user-directed non-core retention, and five request propagation metadata for undecided IBA rows. No propagation failure class was invented where the actual ancestral placement remains unresolved. An initial TMP-only pass failed because the companion GOA file had not yet been copied; that result was preserved, the exact immutable GOA and UniProt companions were copied into TMP, and the corrected focused run passed. No curated canonical file or source cache was changed.

ROOT's independent source-consistency check led to the one annotation change described above: row 101 was moved from UNDECIDED to KEEP_AS_NON_CORE using the measured neutrophil elastase-release evidence. The prior sealed proposal was preserved. Revalidation of the revised proposal passed with the same 74 warnings; all other annotation decisions, core, references, source objects, products and quotations remained unchanged.


2026-10-01 — Canonical review completed. Distinct scientific peer d49857 preceded application 697396. Focused schema, GOA, reference and term validation passed (8bd926) with 74 explained warnings: 69 generic-binding non-core decisions under the supplied ActionEnum and five unresolved IBA metadata requests. Rendering 603c3b and history validation 0f71f0 passed. The page embeds the exact reviewed YAML. All 177 source objects, three alternative products and 18 protected source files are preserved. No successful external research-provider report, complete figure/supplement reading or global validation is claimed.


## Functional reassessment after publication, 2026-10-01 UTC

This entry supersedes the earlier interpretations specifically identified below. The preceding research journal remains an unchanged historical record.

**Specific binding activities.** Q86Y82 is human STX12, not a BLOC-1 co-subunit. The existing associations with syntaxins are refined to syntaxin binding (or syntaxin-1 binding for STX1A); source-listed SNAP25/SNAP47 associations are refined to SNARE binding. Official GO definitions distinguish these protein classes. [PMID:19546860](https://pubmed.ncbi.nlm.nih.gov/19546860/) tests native brain BLOC-1 in recombinant SNARE pull-downs with anti-pallidin detection, including SNAP25 isoforms and open versus closed syntaxin constructs. These experiments support selective complex association, not a direct contact map of isolated pallidin. Other curated physical interactions remain non-core where their specific function has not been resolved. Missing supplementary pair-level records are not negative evidence, and the partner’s function is not transferred to pallidin. No binding interaction is removed solely because it has a broad label.

**Core function.** The accepted syntaxin-binding activity is now explicit in the same core that records contribution to BLOC-1 molecular adaptor activity. The two slots distinguish the subunit’s demonstrated interaction from the assembled complex’s broader trafficking function. This adds no NEW annotation, invented structural activity, or autonomous membrane-fusion activity. Existing core locations and process terms are retained.

**Metabolomics versus secretion.** The six metabolic annotations are marked over-annotated because altered hippocampal metabolite concentrations in [PMID:28701731](https://pmc.ncbi.nlm.nih.gov/articles/PMC5507893/) do not identify a chemical reaction, metabolic scaffold or cofactor role performed by pallidin. This supersedes the earlier blanket non-core retention of those six process assertions. Their exact current mouse donor-reference edges remain unresolved. In contrast, [PMID:17785837](https://pubmed.ncbi.nlm.nih.gov/17785837/) directly measures reduced elastase release from pallid neutrophils, in the context of defective organelle trafficking. Pallidin is a component of that machinery, not the released protein. The existing protein-secretion assertion remains non-core; its retention does not assign elastase activity or claim a resolved fusion step. The independent six-row metabolic reassessment agrees with this distinction; the recovered normal full text confirms the experimental scope.

**Human patient evidence and retraction.** [PMID:22461475](https://pubmed.ncbi.nlm.nih.gov/22461475/), DOI10.1182/blood-2012-01-404350, is a separate 2012 report. The [author-uploaded original Blood letter](https://www.researchgate.net/publication/223990220_Exome_Sequencing_Reveals_a_Pallidin_Mutation_in_a_Hermansky-Pudlak-Like_Primary_Immunodeficiency_Syndrome) was independently read through its patient/functional-results text and Figure1 caption. The patient’s NK cells show abnormal CD107a/CD63 distribution, reduced degranulation and cytolysis; PLDN replacement restores CD107a distribution. Platelet aggregation was normal in this patient. The 2011 report [PMID:21665000](https://pubmed.ncbi.nlm.nih.gov/21665000/) is retracted by [PMID:28475864](https://pmc.ncbi.nlm.nih.gov/articles/PMC5985258/) and is not used as positive evidence. The 2012 paper cites that earlier case but reports its own patient and experiments. No figure-image or supplemental-data inspection is claimed, and no new quotation from either report is introduced here.

**Localization and unresolved source evidence.** ER lumen means the compartment inside the ER membrane. A cytosolic/peripheral-membrane protein lacking a listed signal peptide presents a real topology concern, but that observation alone does not prove that no minor imported pool can exist. The exact donor assay remains uninspected, so the lumen claim stays undecided. A cytokinetic contractile ring is cytoplasmic and cannot be rejected on that lumen argument; its precise localization remains unresolved. Likewise, whole-melanosome movement differs from cargo delivery into melanosomes. The inspected Tyrp1-rescue experiment supports the latter, already accepted separately; incomplete figure/supplement coverage leaves the former experimental annotation undecided. No PAINT tree reconstruction, wrong donor, or absence of all supporting evidence is asserted.

**Evidence provenance.** All 177 source objects, three alternative products and previous quotations are preserved. The products field remains absent, and the third alternative sequence is still Not described. The original local eight-reference fetch failed with DNS errors and created no caches. A separately verified normal recovery subsequently supplied the eight unchanged records. Five contain recovered full text; the 2012 patient letter and 2017 retraction notice contain metadata-only Abstract sections, and the neutrophil paper supplies a narrative abstract only. The correction notice also has a metadata-only Abstract section but contains the complete correction body. External selected primary reading is explicitly distinguished from these normal-cache limits. No manual provider output or publication cache was authored.


**Additional normal-source reassessment.** The pallid P1–P2 hippocampal metabolomics Results and Methods report metabolite pools and transporter abundance, not a pallidin-catalysed pathway. The human SH-SY5Y pallidin-shRNA Results in [PMID:22423091](https://pubmed.ncbi.nlm.nih.gov/22423091/) support a basal redox phenotype, while the exact mouse oxidative-stress donor assertion stays unresolved. In [PMID:28317021](https://pubmed.ncbi.nlm.nih.gov/28317021/), Drosophila pallidin nulls retain normal basal synaptic development and transmission but fail to sustain vesicle recycling under high activity; neuronal rescue, endosome-marker changes and shibire experiments support trafficking. These fly results corroborate the existing synthesis without adding a NEW human annotation or rejecting other species-specific developmental evidence. No complete figure-image or supplement audit is claimed.

**Quotation accounting.** Existing source quotations are unchanged. The two new exact normal-cache anchors are limited to seven words each from PMID:17785837 and PMID:28701731; no new quotations from shared BLOC-1 architecture/cargo papers or either human case report are added.

**Validation, 2026-10-01 UTC.** The final candidate passed normal schema, cached-reference/quotation and ontology-term validation (start1843de, completion1289ab; exit0). There were 63 nonblocking advisories: 58 broad binding rows retained as non-core under the supplied action rules and five undecided IBA rows without an independently reconstructed phylogenetic placement. Those limits are retained explicitly; no unsupported propagation metadata was invented. Complete source-object equality separately verifies all177 imported annotations and allthree alternative products. The provisional draft’s unsupported knowledge-gap description key was moved into the existing boundary slot before validation, retaining the scientific content.


## Canonical follow-up verification — 2026-10-01

The distinct final science peer approved the exact held proposal. Canonical schema, GOA, cached-reference/quotation and ontology-term validation passed with the same 63 disclosed advisories: 58 generic-binding policy warnings and five unresolved IBA metadata requests. No unsupported REMOVE or phylogenetic metadata was introduced. Rendering and the new history record passed; the rendered embedded YAML equals the review. All 177 source assertions, three alternative products, prior history and source-cache bytes are preserved. The reviewed follow-up is ready for publication on the existing gene PR.


## Second PR follow-up — 2026-10-01

The [second review of PR #3770](https://github.com/ai4curation/ai-gene-review/pull/3770#issuecomment-5936932124) correctly challenges the earlier demand to exclude every hypothetical ER-lumen import route. That is too strong a requirement for judging an electronic transfer. The ER-lumen row is now MARK_AS_OVER_ANNOTATED: pallidin's established cytoplasmic/peripheral-membrane architecture does not support this compartment. This is a judgment of over-specific localization, not a claim that the unidentified mouse experiment was erroneous. The contractile-ring assertion remains unresolved because a cytoplasmic ring does not create the same topological mismatch.

The gene-expression row remains UNDECIDED. Its rationale now distinguishes the hypothetical insufficiency of downstream abundance changes from an actual examination of the donor experiment. The six metabolic decisions are based on an inspected perturbation/metabolomics study; no equivalent source-to-assertion assessment has been established for this row. A targeted literature search located the expression-profiling study [PMID:32436302](https://pubmed.ncbi.nlm.nih.gov/32436302/), and selected indexed Table 6/Results describe changes in pallid mouse hippocampi. This is a lead, not a verified join to the current donor annotation; it has not been added as a curated reference or used to claim a new function. The earlier lipid and redox consultation remains applicable.

The requested project policy retains supported generic physical interactions as non-core when no more informative molecular function has been established. No binding assertion is removed merely because its term is broad. Pairwise interaction evidence and membership in the BLOC-1 complex describe different claims; the presence of a complex-membership annotation alone does not establish that a binding assertion is false or circular. The eleven supported syntaxin/SNARE refinements remain in place.

Likewise, the three identical-protein-binding records are not narrowed to homodimerization simply because separate homodimerization experiments exist on the same gene. Self-association in an interaction assay does not by itself establish its stoichiometry. One pallidin copy in the assembled BLOC-1 octamer does not invalidate isolated self-association observations.

This follow-up preserves the biological summary, single core, three alternative products, all 177 source objects and 32 references. It changes one action and clarifies one unresolved rationale. Earlier journal entries remain as historical records; this entry supersedes the earlier ER-lumen decision.


## Second follow-up validation — 2026-10-01

The distinct science peer, focused schema/term/reference/GOA checks, history validation and rendering passed. The 63 existing advisories remain: 58 generic-binding policy advisories and five unresolved IBA provenance advisories. No uninspected PAINT history was invented to silence those warnings. The rendered YAML equals the reviewed file. Fetched source bytes and prior history remain unchanged. Remote publication and approval are subsequent steps.


## Scope of the binding decisions — 2026-10-01

The [third PR review](https://github.com/ai4curation/ai-gene-review/pull/3770#issuecomment-5938278897) identifies a documentation gap: the general annotation-reviewer guidance favors removing uninformative generic binding, whereas this review retains 58 such observations as non-core. The [BLOC1S6 binding rationale](../../../projects/CLINGEN_MENDELIAN/BLOC1S6-binding-rationale.md) now records the scope and evidence limits of that exception. Earlier references here to a “requested project policy” should be read as this task-specific BLOC1S6 decision, not as a new global user policy or an instruction to migrate other reviews.

The retained rows preserve existing curated physical associations, with their partner identifiers and sources, outside the core functional synthesis. Selected human reconstitution experiments distinguish isolated pallidin–BLOS1 and pallidin–Cappuccino pairs from whole-octamer association. Binary-screen records and AP-MS associations have different limits: the supplementary pairs were not all independently inspected, and co-purification is not proof of direct contact. The accepted complex-membership rows do not erase this source-specific information or make it circular. The eleven supported syntaxin/SNARE refinements remain the more informative decisions where partner class and evidence support them.

This exception concerns the low information content of a supported association, not unresolved evidence. `UNDECIDED` remains necessary when the relevant assertion cannot be judged from accessible evidence; retaining a curated association does not certify every primary assay or establish a new mechanism. Self-association likewise does not establish homodimer stoichiometry. The core synthesis and all current annotation actions are unchanged by this clarification.