Proteostasis Review Batch 5 — Gene Selection

Date: 2026-06-07
Branch: claude/proteostasis-50-genes-xFOZK

Context

The four prior batches are complete:
- proteostasis-pr-1217 (50 human genes, merged 2026-06-02)
- proteostasis-batch-2026-06-03 (50 human genes, alphabetical sweep AAAS..ATP6V0D1)
- proteostasis-batch-2026-06-06 (20 human genes; V-ATPase core, ER folding/QC,
autophagy/mitophagy receptors, co-chaperone/UPS regulation)
- proteostasis-batch-2026-06-07 (30 human genes; V-ATPase tissue isoforms + ClC-7,
mito/ER chaperones, collagen biogenesis, histone chaperones, CRL/UPS adaptors)

This batch selects 50 more genes from the PN projected candidate-additions
report (reports/pn_projection/pn_projected_candidate_additions.tsv), all PN
candidates with ok_for_propagation_to_go scope that had no local
*-ai-review.yaml before this batch.

Theme: the protein-folding chaperone & co-chaperone network

Rather than continue the alphabetical sweep into low-value family/domain UPS
inclusions, batch 5 takes a biologically coherent slice through the cytosolic
and ER protein-folding machinery — the mechanistic heart of proteostasis. It
is dominated by the J-domain (HSP40) co-chaperone family, which had a large run
of unreviewed PN candidates, plus their HSP70/HSP90 partners, the small heat
shock proteins, the FKBP immunophilin co-chaperones, and the ER oxidative-folding
and peptidyl-prolyl isomerase enzymes.

Deliberate exclusion of the mega-genes

HSPA5 (BiP), HSP90AA1, and HSP90AB1 were considered but deferred.
Each carries 50+ unique GOA GO terms (HSPA5 alone has 58 terms / 214 annotation
rows) and a very large literature; folding them into a 50-gene batch would force
shallow review. They are flagged for dedicated single-gene PRs and replaced here
by lighter but high-value hub co-chaperones (STIP1/Hop, SGTA, SERPINH1/HSP47).

Selected genes (50)

J-domain (HSP40 / DNAJ) co-chaperones — 29

Cytosolic, ER, and mitochondrial HSP70 J-domain co-chaperones that set HSP70
substrate specificity and stimulate its ATPase.

1. DNAJB4 — cytosolic HSP40 (DNAJB1 paralog)
2. DNAJB5 — cytosolic HSP40
3. DNAJB11 (ERdj3) — ER-lumenal BiP J-protein co-chaperone
4. DNAJB13 — flagellar/axonemal HSP40
5. DNAJC1 (ERdj1/MTJ1) — ER membrane J-protein
6. DNAJC3 (p58IPK/ERdj6) — ER J-protein, PERK/eIF2α regulation
7. DNAJC4 — J-domain protein
8. DNAJC5 (CSPα) — synaptic vesicle co-chaperone (ANCL/DNAJC5)
9. DNAJC5B — CSP paralog
10. DNAJC5G — CSP paralog
11. DNAJC6 (auxilin/PARK19) — clathrin-uncoating HSC70 co-chaperone
12. DNAJC7 (Tpr2) — TPR-containing HSP70/HSP90 shuttling co-chaperone
13. DNAJC9 — histone H3-H4 co-chaperone / HSP70 J-protein
14. DNAJC10 (ERdj5) — ER reductase / PDI-family J-protein, ERAD
15. DNAJC11 — mitochondrial MIB/MICOS-associated J-protein
16. DNAJC12 (JDP1) — cytosolic HSP70 co-chaperone (phenylalanine metabolism)
17. DNAJC13 (RME-8/PARK21) — endosomal HSC70 co-chaperone
18. DNAJC14 (DRIP78) — ER J-protein, receptor trafficking
19. DNAJC15 (MCJ) — mitochondrial inner-membrane J-protein
20. DNAJC16 — J-domain protein
21. DNAJC17 — J-domain protein
22. DNAJC19 (TIM14/TIMM14) — mitochondrial import motor J-protein
23. DNAJC21 (DNAJA5) — ribosome biogenesis HSP70 co-chaperone
24. DNAJC22 — J-domain protein
25. DNAJC24 (DPH4) — diphthamide biosynthesis / J-domain
26. DNAJC25 — J-domain protein
27. DNAJC27 (RBJ) — Rab-domain J-protein
28. DNAJC28 — J-domain protein
29. DNAJC30 — mitochondrial J-protein (LHON-related)

Small heat shock proteins (sHSP / HSPB) — 5

ATP-independent holdase chaperones.

1. HSPB2 (MKBP) — small HSP, muscle
2. HSPB3 — small HSP, muscle
3. HSPB7 (cvHSP) — cardiovascular small HSP, aggregate handling
4. HSPB8 (HSP22) — CASA-pathway holdase (BAG3 partner)
5. HSPB9 — testis small HSP

HSP70 / HSP90 hub co-chaperones — 6

1. HSPA13 (STCH) — microsomal/ER atypical HSP70
2. HSPA14 (HSP70L1) — ribosome-associated complex (RAC) HSP70
3. STUB1 (CHIP) — HSP70/HSP90 co-chaperone & U-box E3 (triage/degradation)
4. STIP1 (HOP) — HSP70-HSP90 organizing protein / adaptor
5. SGTA — small glutamine-rich TPR co-chaperone (BAG6/GET tail-anchor QC)
6. SERPINH1 (HSP47) — ER collagen-specific molecular chaperone

FKBP immunophilin co-chaperones — 4

PPIase-domain HSP90 co-chaperones.

1. FKBP4 (FKBP52) — HSP90 co-chaperone, steroid receptor maturation
2. FKBP5 (FKBP51) — HSP90 co-chaperone, stress/GR signaling
3. FKBP8 (FKBP38) — membrane FKBP, Bcl-2/mTOR, BNIP3 mitophagy adapter
4. FKBPL (WISP39) — FKBP-like HSP90 co-chaperone

ER oxidative folding & peptidyl-prolyl isomerization — 6

1. P4HB (PDI / PDIA1) — protein disulfide isomerase, prototypical ER folding enzyme
2. ERO1A (ERO1L) — ER oxidoreductin 1 alpha (reoxidizes PDI)
3. ERO1B (ERO1LB) — ER oxidoreductin 1 beta
4. ERP27 — PDI-family non-catalytic chaperone (ERp27)
5. ERP29 — PDI-family ER chaperone (ERp29)
6. PPIB (cyclophilin B) — ER peptidyl-prolyl isomerase, collagen folding

Method

For each gene:
- uv run ai-gene-review fetch-gene human <GENE> (UniProt, GOA, cached
publications, seeded -ai-review.yaml).
- Research from cached publications + UniProt + literature; notes recorded in
<GENE>-notes.md with provenance.
- Each seeded GOA annotation reviewed per GO guidelines (ACCEPT /
KEEP_AS_NON_CORE / MODIFY / MARK_AS_OVER_ANNOTATED / REMOVE / UNDECIDED) with
supported_by evidence.
- description, core_functions, suggested_questions, suggested_experiments
populated.
- Validated with uv run ai-gene-review validate human <GENE>.

Curation watch-points for this batch

• Avoid protein binding (GO:0005515) as a core function; for J-proteins
  prefer GO:0001671 ATPase activator activity / GO:0030544 Hsp70 protein binding / GO:0051087 protein-folding chaperone binding as appropriate.
• Holdase vs foldase: small HSPs and many J-proteins are holdases/co-chaperones,
  not autonomous foldases — keep protein folding (GO:0006457) as non-core where
  the gene only assists HSP70/HSP90.
• Family/paralog over-annotation: poorly characterized J-proteins (e.g.
  DNAJC4, DNAJC16, DNAJC22, DNAJC25, DNAJC28, DNAJC5B/G, HSPB9) may carry IBA
  terms transferred from better-studied paralogs; mark over-annotations.
• Compartment specificity: assign ER vs mitochondrial vs cytosolic
  localization carefully (e.g. DNAJC11/15/19/30 mitochondrial; DNAJB11/C1/C3/C10
  ER-lumenal/membrane).

Feedback into the PN mapping sets (2026-06-07)

The batch-5 gene reviews were fed back into the curated PN source-code -> GO
mapping YAMLs (projects/PROTEOSTASIS/mappings/). Changes:

• Correction (er_proteostasis): the Protein disulfide isomerase reoxidation
  type node was retargeted from GO:0003756 (PDI activity) to GO:0016971
  (flavin-dependent sulfhydryl oxidase activity); ERO1A/ERO1B are PDI-reoxidizing
  oxidases, not isomerases. ERO1A/ERO1B were ALSO added to the parent
  Protein disulfide isomerases group excluded_subjects so the parent
  PDI-activity term no longer double-propagates to them.
• Exclusions: ERP27, ERP29 (non-catalytic, no CXXC) from the PDI group;
  DNAJC27 (Rab/J MEK-ERK scaffold) from the cytonuclear J-domain cochaperone
  node; DNAJC11 (structural MIB/MICOS subunit) from the mitochondrial J-domain
  node; HSPA13/STCH (atypical, truncated SBD) from the ER HSP70 node; STIP1
  from the CMA Substrate selection node (no CMA evidence).
• Confirmations + provenance: added file: gene-review references to the
  HSP90-cochaperone (FKBP4/5/8/FKBPL), small-HSP (HSPB2/3/7/8/9), cytonuclear
  HSP70 (HSPA14), GET-pathway (SGTA), collagen processing (SERPINH1/PPIB/P4HB),
  CASA/aggrephagy (HSPB8/STUB1), and mitophagy (FKBP8) nodes, with curation
  notes (incl. that GO:0001671 / GO:0051082 are the more-specific J-protein /
  small-HSP MFs supported at the gene level).

Verified by re-running the PN projection against the affected genes: the
excluded gene-GO pairs are gone and ERO1A/B project only to GO:0016971.

Caveat: the canonical candidate-additions report
(reports/pn_projection/pn_projected_candidate_additions.tsv) was NOT
regenerated in this session because the DuckDB GOA source
(~/repos/go-db/db/goa_human.ddb) is unavailable here; regenerating against
only the local genes/human GOA folders would mark ~3000 workbook genes as
no_local_goa and degrade the committed report. The projection should be
regenerated against the DuckDB GOA source to propagate these mapping edits into
the candidate queue.