Tardigrade Stress Response Protein Curation

The Problem: Surviving the Impossible

• Ramazzottius varieornatus (RAMVA) is an extremotolerant tardigrade
• It survives near-complete desiccation by entering a dormant tun state
• The same machinery confers resistance to ionizing radiation, extreme temperatures, vacuum, and high pressure
• During desiccation, massive ROS accumulation damages biomolecules in every cellular compartment
• Radiation tolerance is likely a byproduct of desiccation tolerance — both stresses generate hydroxyl radicals

Goal: curate the complete reviewed (Swiss-Prot) stress-response proteome of RAMVA

Key Biology: A Compartment-Partitioned Defense

Tardigrade-unique intrinsically disordered proteins (TDPs) each protect a different compartment:

Plus conventional ROS-scavenging enzymes (Cu/Zn-SODs, Mn-SOD).

Two Unifying Themes

Intrinsic disorder

• Most TDPs are intrinsically disordered proteins (IDPs)
• Disorder enables flexible, multivalent interactions (chromatin for Dsup, cytoskeleton for CAHS)
• Enables reversible phase transitions (CAHS hydrogel formation) that physically stabilize structures during water loss

Evolutionary uniqueness

• CAHS, SAHS, MAHS, Dsup are tardigrade-specific — no homologs outside the phylum
• RvLEAM is the exception: LEA proteins occur in plants/nematodes — convergent evolution of desiccation tolerance

The Approach: AI-Assisted Gene Review

• Curate each Swiss-Prot protein against existing GOA annotations using GO guidelines
• For each annotation: ACCEPT / MODIFY / REMOVE / NEW with literature + bioinformatic support
• Deep research (falcon) synthesized per gene; reviewer adjudication on top
• Where annotations were missing or wrong, propose NEW terms or corrections
• Custom bioinformatics pipeline (sequence conservation + PROSITE motifs + Pfam) to test catalytic claims

Finding 1: A Systematic Annotation Gap

• The most consistent gap across all TDPs: absence of GO:0009269 (response to desiccation)
• This is the core biological process for the entire family
• Proposed as NEW for every gene in the project
• Response to osmotic stress proposed for several genes with gain-of-function data (MAHS, RvLEAM expressed in human cells)

The defining function of the whole protein family was simply not in the database.

Finding 2: The Nuclear Shield — Dsup

• Dsup (P0DOW4): nucleosome-binding chromatin shield
• Protects against both X-ray and H2O2-induced DNA damage via the same nucleosome-shielding mechanism
• Curation nuance: GO:0003677 (DNA binding) first proposed for MODIFY → nucleosome binding
• Corrected to ACCEPT: GO:0031491 (nucleosome binding) is NOT a child of DNA binding (it sits under chromatin binding) — both are valid independent MF annotations
• Key refs: Hashimoto et al. 2016 (PMID:27649274); Chavez et al. 2019 (PMID:31571581)

Finding 3: The SOD Paralog Family Is Half Broken

RAMVA has ~10 Cu/Zn-SOD paralogs. Bioinformatic analysis (sequence + PROSITE PS00087 + Pfam):

At least 4 of 9 Cu/Zn-SOD-family paralogs appear to have lost/impaired canonical SOD activity.

Finding 3 (cont.): RvSOD15 the Pseudoenzyme

• Crystal structure (PMID:37358501) shows Val87 replaces the catalytic His copper ligand
• Confirmed structurally → its 4 SOD-activity annotations are OVER-ANNOTATED
• Three more paralogs keep all four catalytic Cu histidines at the residue level, yet fail PROSITE PS00087 (N-terminal Cu coordination signature)
• By analogy with the failed V87H rescue in RvSOD15 (loop dynamics), these are likely impaired too
• Validates Sim & Inoue (2023): "some other RvSODs are also unusual SODs" — now with a precise count

Finding 4: Annotation Propagation Errors Are Systematic

• Automated pipelines (InterPro2GO, EC2GO, UniRule, ARBA) assigned GO:0004784 (SOD activity) to ALL Cu/Zn-SOD-family proteins by Pfam membership alone
• No check of catalytic-residue conservation or motif integrity
• Canonical "annotation propagation by family membership" failure — documented for ≥4 of 9 paralogs
• RvY_01767 (Mn-SOD) also got a spurious "respiratory chain complex" annotation from an ARBA rule (it is a soluble matrix protein) → 1 REMOVE

Finding 5: SAHS Proteins Have FABP-Like Folds

• SAHS1 and SAHS2 adopt beta-barrel folds homologous to fatty acid-binding proteins
• Crystal structures available (SAHS1: Fukuda et al. 2017, PMID:28703282)
• Lipid binding was annotated for SAHS1 but not SAHS2 → proposed to add for SAHS2
• Illustrates how structural homology informs missing molecular-function annotations

What Makes Curation Hard Here

• Tardigrade-unique proteins have no homologs → little transferable annotation; the core process (desiccation response) was simply absent
• Pseudoenzymes hide in plain sight: full catalytic residues present at sequence level can still be non-functional (loop/motif context matters)
• Family-membership propagation systematically over-annotates enzymatic activity
• GO graph subtleties: nucleosome binding vs DNA binding placement changes the right action
• Disorder + phase behavior are biologically central but awkward to capture in current GO terms

Conclusions & Status

• Full reviewed RAMVA stress proteome curated: Dsup, 3 CAHS, 2 SAHS, MAHS, RvLEAM, 10 Cu/Zn-SODs, 1 Mn-SOD
• GO:0009269 (response to desiccation) proposed as the unifying NEW term across all TDPs
• A reusable SOD bioinformatics pipeline (analyze_sods.py, check_prosite.py) revealed ~half the SOD expansion may be non-catalytic
• "Gene duplication = more antioxidant capacity" is only partially correct

Future: capture intrinsic-disorder/phase-transition function in GO; extend pseudoenzyme screening to the broader paralog set; experimental validation of impaired-SOD predictions.