AIGR Gene Hypothesis Deep Research — Final Report

Target: rat Prkaa2 (Q09137), AMPK α2 catalytic subunit

Hypothesis: prediction-adaptor-activity — GO:0030674 protein-macromolecule adaptor activity

Focus type: computational_prediction

Reference context: doi:10.64898/2026.03.19.712954

Summary

The GO-GPT/BioReason-Pro prediction of protein-macromolecule adaptor activity (GO:0030674) for rat Prkaa2 (Q09137), the AMP-activated protein kinase (AMPK) α2 catalytic subunit, is refuted. It is best classified as a subunit-level mis-assignment / over-annotation. The genuine molecular function of the AMPK α subunit is protein serine/threonine kinase activity; the adaptor/scaffolding role in the AMPK heterotrimer belongs to the β subunit.

Three independent lines of evidence converge. First, the Q09137 domain architecture is a canonical protein kinase (kinase domain res 16–268, catalytic proton-acceptor res 139, ATP-binding pocket res 22–30/45) plus an autoinhibitory sequence — with no carbohydrate-binding module, no glycogen-binding domain, and no dedicated scaffold module. Its high-confidence molecular-function annotations are all catalytic (GO:0004674, GO:0004672, GO:0004679), each supported by direct-assay (IDA) evidence. Second, structural and biochemical studies show the β subunit physically bridges α and γ, and that β's carbohydrate-binding module forms the ADaM interface with the α kinase domain; reviews explicitly name β "the scaffolding subunit."

Third and decisively, the IDA source paper anchoring the GO:0030674 annotation on Q09137 — PMID:15695819 — is a β-subunit paper. Its title and results describe the AMPK β subunit tethering α and γ via its C-terminal sequence (β1 residues 186–270). The experimentally demonstrated adaptor/tethering activity is a property of β, not of the α2 catalytic subunit that carries the annotation. The recommended curation action is to remove GO:0030674 from Prkaa2 and, if preserved, transfer it to the AMPK β subunits (Prkab1/Prkab2). The α subunit's legitimate participation in heterotrimer assembly is already captured by GO:0044877 (protein-containing complex binding, IDA), which should be retained instead of adaptor activity.

Key Findings

Finding 1 — The core molecular function of Q09137 is protein Ser/Thr kinase activity, not adaptor activity

The domain architecture of UniProt Q09137 is that of a classical protein kinase: a protein kinase domain spanning residues 16–268, a catalytic active site (proton acceptor) at residue 139, an ATP-binding region (residues 22–30 and 45), and a C-terminal autoinhibitory sequence (AIS, residues ~291–376). There is no carbohydrate-binding module (CBM), no glycogen-binding domain, and no dedicated scaffold/adaptor domain anywhere in the α2 sequence. This is the signature of an enzyme, not a scaffold.

Consistent with this, the existing high-quality molecular-function GO annotations for Q09137 are all catalytic and supported by direct-assay evidence: GO:0004674 protein serine/threonine kinase activity (IDA:RGD), GO:0004672 protein kinase activity (IDA:RGD), and GO:0004679 AMP-activated protein kinase activity (IDA:UniProtKB). Multiple primary and structural sources describe AMPKα as the catalytic subunit and AMPKβ as the scaffold.

Supporting quotations from the literature:

• PMID:23721051(https://pubmed.ncbi.nlm.nih.gov/23721051/): "AMP-activated protein kinase (AMPK) is a sensor of energy status composed of a catalytic subunit (AMPKα), a scaffolding subunit (AMPKβ) and a regulatory subunit involved in nucleotide binding (AMPKγ)." — Explicitly assigns the catalytic role to α and the scaffolding role to β, directly contradicting adaptor activity for the α subunit.
• PMID:20529674(https://pubmed.ncbi.nlm.nih.gov/20529674/): "The catalytic (alpha) subunit of AMPK/SNF1, Snf1 in yeast, contains a protein Ser/Thr kinase domain (KD), an auto-inhibitory domain (AID), and a region that mediates interactions with the two regulatory (beta and gamma) subunits." — Describes the α subunit as a Ser/Thr kinase with an autoinhibitory domain, consistent with kinase (not adaptor) as its molecular function.
• PMID:26635351(https://pubmed.ncbi.nlm.nih.gov/26635351/): "AMP-activated protein kinase (AMPK) is a serine/threonine protein kinase that serves as a pleotropic regulator of whole body energy homoeostasis." — Confirms the enzyme's molecular function is serine/threonine protein kinase activity, carried by the catalytic α subunit.

Finding 2 — The scaffold/adaptor role in the AMPK heterotrimer belongs to the β subunit

Structural studies establish that within the AMPK αβγ core, the β subunit binds both α and γ, physically bridging them. The β subunit's carbohydrate-binding module forms the ADaM (Allosteric Drug and Metabolite) site — the interaction surface between the α kinase domain and the β CBM. Reviews of AMPK structure identify β as "the scaffolding subunit." The α subunit contributes its kinase domain to that interface but does not itself act as the bridge.

Because the α subunit's contact with β during subunit assembly is a subunit-of-a-complex interaction rather than a bridging function between two distinct macromolecules, the appropriate molecular-function term for the α side of this interaction is protein-containing complex binding (GO:0044877) — which is already annotated to Q09137 by IDA:RGD — not GO:0030674 adaptor activity.

Supporting quotations:

• PMID:18079111(https://pubmed.ncbi.nlm.nih.gov/18079111/): "The previously determined crystal structure of the core of mammalian alphabetagamma complex shows that beta binds alpha and gamma." — Demonstrates that the β subunit is the physical bridge between α and γ, i.e. the adaptor/scaffold, not α.
• PMID:30478170(https://pubmed.ncbi.nlm.nih.gov/30478170/): "...binds the AMPK ADaM site, which forms the interaction surface between the kinase domain (KD) of the α-subunit and the carbohydrate-binding module (CBM) of the β-subunit." — Shows the α subunit contributes its kinase domain to the interface while the β CBM is the scaffolding module.

Finding 3 — The IDA source (PMID:15695819) demonstrates adaptor activity of the BETA subunit, not α2

This is the decisive finding. QuickGO shows that the GO:0030674 (protein-macromolecule adaptor activity) annotation on Q09137 is a single, rat-specific IDA assigned by RGD, referenced to PMID:15695819. The annotation is not present on human PRKAA2 (P54646), on PRKAA1 orthologs, or on any AMPK β subunit (PRKAB1 Q9Y478, PRKAB2 Q9UGI9). This pattern — one isolated rat annotation, absent from orthologs and paralogs — is itself a red flag for a curation error.

Examination of the cited paper resolves the issue conclusively. PMID:15695819 (Iseli et al., 2005, J. Biol. Chem.) is titled "AMP-activated protein kinase beta subunit tethers alpha and gamma subunits via its C-terminal sequence (186-270)." The paper reports that the β subunit's conserved C-terminal 85-residue sequence (β1 186–270) "is the primary αγ binding sequence responsible for the formation of the AMPK αβγ heterotrimer." In other words, the experimentally demonstrated adaptor/tethering activity is a property of the β subunit. The α subunit is described as the catalytic subunit and merely provides a docking region (α1 313–473) that β binds.

Supporting quotations:

• PMID:15695819(https://pubmed.ncbi.nlm.nih.gov/15695819/): "The conserved C-terminal 85-residue sequence of the beta subunit (90% between beta1 and beta2) is the primary alphagamma binding sequence responsible for the formation of the AMPK alphabetagamma heterotrimer." — The cited IDA source demonstrates that the adaptor/tethering activity bridging α and γ is a function of the β subunit, not the α2 catalytic subunit that carries the GO:0030674 annotation.
• PMID:15695819(https://pubmed.ncbi.nlm.nih.gov/15695819/): "Truncation of the alpha subunit reveals that beta1 binding requires the alpha1-(313-473) sequence." — Shows the α subunit is a binding partner (provides a docking region) rather than the bridging adaptor, which is β.

Mechanistic Model / Interpretation

The AMPK holoenzyme is an obligate αβγ heterotrimer with a clean division of molecular labor. The following schematic captures the assignment of function to subunit:

    AMP / ADP / ATP
          │
          ▼
   ┌──────────────┐
   │   γ subunit  │   (nucleotide sensor; CBS/Bateman domains)
   │  (Prkag1/2/3)│
   └──────┬───────┘
          │  bound by β C-terminus (186–270)
          ▼
   ┌───────────────────────────────────────────────┐
   │              β subunit  (Prkab1/2)             │  ◄── ADAPTOR / SCAFFOLD
   │   CBM (glycogen binding) + C-terminal tether   │      GO:0030674 belongs HERE
   │   "tethers alpha and gamma"  (PMID:15695819)   │
   └───────────────────────────────────────────────┘
          │  β binds α1 (313–473)
          ▼
   ┌───────────────────────────────────────────────┐
   │           α subunit  (Prkaa2 = Q09137)         │  ◄── CATALYTIC
   │  Kinase domain (16–268) + AIS (291–376)        │      GO:0004674 kinase activity
   │  phosphorylates downstream substrates (pThr172)│      GO:0044877 complex binding
   └───────────────────────────────────────────────┘
          │
          ▼
      Ser/Thr phosphorylation of
ACC, ULK1, TSC2, HMGCR, etc. (catabolic switch)

Function-to-subunit mapping:

The prediction under review takes a function that genuinely exists within the AMPK complex — an adaptor/scaffolding activity — and assigns it to the wrong subunit. The mechanistic error is category-level: the α subunit is the enzyme that acts on substrates, whereas the β subunit is the structural glue that holds the complex together. Being bound by a scaffold (as α is by β) does not confer adaptor activity on the bound partner. The distinction between GO:0044877 (this protein is part of / binds a complex) and GO:0030674 (this protein bridges two other macromolecules) is exactly the distinction that the annotation conflates.

Evidence Base / Evidence Matrix

GO Curation Implications

Lead requiring curator verification: Remove GO:0030674 (protein-macromolecule adaptor activity) from rat Prkaa2 (Q09137).

Rationale for the recommended action:

1. The MF term is mis-assigned at the subunit level. GO:0030674 describes a bridging activity between two macromolecules. In the AMPK heterotrimer, this bridging is performed by the β subunit (PMID:15695819, PMID:18079111), not the α catalytic subunit.
2. The supporting IDA does not support the annotation as placed. PMID:15695819 experimentally demonstrates the β subunit's tethering function; it does not demonstrate adaptor activity for α2. The annotation therefore fails the "evidence supports the term for this gene product" test.
3. The retained/strengthened MF terms should be the catalytic ones already present: GO:0004674 (protein serine/threonine kinase activity, IDA), GO:0004672 (protein kinase activity, IDA), and GO:0004679 (AMP-activated protein kinase activity, IDA). These represent the α2 subunit's genuine, directly-assayed molecular function.
4. The α-side of subunit assembly is already correctly captured by GO:0044877 (protein-containing complex binding, IDA:RGD). This is the informative, non-"protein binding" term that describes the α subunit being incorporated into the heterotrimer.
5. Optional transfer lead: GO:0030674 is appropriately annotated to the AMPK β subunits (Prkab1/Prkab2), where the tethering activity was actually demonstrated. This is a separate curation action on a different gene product and should be verified independently.

GO decision table

Mechanistic Scope

The molecular function under test is adaptor/scaffolding activity — the physical bridging of two distinct macromolecules to bring them into proximity. This is an immediate, direct molecular property of a protein.

• Direct α2 gene-product activity: protein Ser/Thr kinase catalysis (phosphotransfer to substrate Ser/Thr residues), plus being incorporated into the αβγ complex via its α1(313–473)-equivalent docking region.
• Not α2's direct activity (the seed's proposed adaptor role): tethering α to γ — this is performed by the β subunit's C-terminal sequence.
• Downstream / not molecular-function-level: AMPK's roles in energy homeostasis, GLUT4 expression, mitochondrial biogenesis, autophagy regulation, and neuronal development (e.g., PMID:21233771, PMID:30359677, PMID:24295634) are biological processes driven by the kinase activity, not evidence for an adaptor molecular function.

The seed hypothesis correctly frames the discriminating question — is adaptor activity a property of the catalytic α subunit, or of the β scaffold? The evidence answers unambiguously: it is the β subunit's.

Conflicts and Alternatives

• Subunit confusion (the actual error): The annotation conflates the α (catalytic) and β (scaffold) subunits. The source paper (PMID:15695819) is a β-subunit study; the annotation sits on the α2 gene product. This is the single most important conflict, and it resolves against the prediction.
• "Complex binding" vs "adaptor activity": A defensible-but-weaker alternative reading is that the α subunit's participation in the heterotrimer justifies some interaction term. This is true — but the correct term is GO:0044877 (already present), not GO:0030674. Adaptor activity requires bridging two other macromolecules, which α does not do.
• Organism specificity / database carry-over: The annotation is rat-specific (RGD) and does not propagate to the human ortholog (P54646) or to PRKAA1, indicating an isolated curation event rather than a robustly conserved, orthology-supported function.
• No competing primary evidence found: Across 33 papers reviewed, no primary study demonstrates that the AMPK α2 catalytic subunit acts as a macromolecular adaptor/scaffold bridging two other proteins. The consistent picture is α = kinase, β = scaffold.

Limitations and Knowledge Gaps

1. QuickGO/UniProt provenance was read from database records, not re-derived programmatically here. The conclusion that the annotation is a single rat-specific RGD IDA to PMID:15695819 relies on database inspection; a curator should confirm the current annotation state directly in QuickGO/RGD, as annotations can be updated.
2. Full text of PMID:15695819 was assessed via title and abstract-level evidence. The subunit assignment is unambiguous from the title and reported results, but a curator verifying the removal should confirm the exact experimental claims in the full text.
3. GO:0030674 definition edge cases. Under the strict GO definition (bridging two distinct macromolecules), α does not qualify. If a curator interprets "adaptor activity" loosely to include any subunit that contributes an assembly interface, the interpretation should be confirmed against current GO documentation.
4. β-subunit transfer not independently curated here. The recommendation to add GO:0030674 to Prkab1/Prkab2 is a lead, not a completed curation; it needs its own evidence review on those gene products.

Proposed Follow-up Experiments / Actions (Curation Leads)

All items below are leads requiring curator verification.

Action change (primary lead):
- Remove GO:0030674 (protein-macromolecule adaptor activity) from Prkaa2 / Q09137 (currently RGD IDA, PMID:15695819).

Candidate reference with exact snippet to verify:
- PMID:15695819 — verify: "The conserved C-terminal 85-residue sequence of the beta subunit (90% between beta1 and beta2) is the primary alphagamma binding sequence responsible for the formation of the AMPK alphabetagamma heterotrimer." and "Truncation of the alpha subunit reveals that beta1 binding requires the alpha1-(313-473) sequence." These confirm the adaptor function belongs to β, and that α is the bound partner.

Retained / strengthened terms (verify present and well-evidenced):
- GO:0004674 protein serine/threonine kinase activity (IDA) — core MF.
- GO:0004679 AMP-activated protein kinase activity (IDA).
- GO:0044877 protein-containing complex binding (IDA) — use instead of adaptor activity.

Candidate new/transfer term (separate gene products, verify independently):
- Consider GO:0030674 on Prkab1 / Prkab2, where PMID:15695819 demonstrates the tethering activity.

Discriminating tests to resolve any residual doubt:
1. Subunit-swap / truncation binding assays (as in PMID:15695819): deleting β C-terminal 186–270 abolishes αγ tethering; α alone cannot bridge β and γ.
2. Structural interface mapping in existing full-length crystal structures (PMID:30478170, PMID:29480466): tabulate direct α–γ contacts; if α–γ contacts are β-mediated, α cannot be the adaptor.
3. In vitro reconstitution: test whether purified α2 can co-recruit two macromolecules without β (predicted: it cannot).
4. Orthology/annotation-conservation check: confirm GO:0030674 is absent on all α orthologs and addable on β orthologs.

Suggested questions for the curator:
- Does current QuickGO still show GO:0030674 on Q09137 as a single RGD IDA to PMID:15695819?
- Under the strict GO:0030674 definition, does contributing an assembly interface count as adaptor activity, or is bridging of two distinct macromolecules required? (This report assumes the latter.)

Conclusion

The prediction that rat Prkaa2 (AMPK α2 catalytic subunit, Q09137) has protein-macromolecule adaptor activity (GO:0030674) is refuted. The α2 subunit's genuine molecular function is protein serine/threonine kinase activity; the adaptor/scaffolding role in the AMPK heterotrimer is carried by the β subunit, and the very paper cited to support the annotation (PMID:15695819) is a demonstration of the β subunit's tethering activity. The GO:0030674 annotation should be removed from Prkaa2 and, if preserved at all, transferred to the AMPK β subunits. This is a textbook example of a computational prediction producing a subunit-level mis-assignment by attaching a real complex-level function (scaffolding) to the wrong member of the complex.

Artifacts

• OpenScientist final report
• OpenScientist final report