PRKAA2 Research Notes

Key Research Findings

Core Catalytic Function

PRKAA2 encodes the α2 catalytic subunit of AMPK, which has intrinsic serine/threonine kinase activity. The α2 isoform differs from α1 in tissue distribution and substrate specificity PMID:8955377.

Direct Substrates with Strong Evidence

1. ACC (Acetyl-CoA Carboxylase): Direct phosphorylation inhibits fatty acid synthesis PMID:12065578.

2. PFK-2 (6-Phosphofructo-2-kinase): Direct phosphorylation at Ser466 activates the enzyme, promoting glycolysis during energy stress PMID:11069105.

3. ChREBP: Direct phosphorylation at Ser568 inhibits DNA binding, preventing lipogenic gene expression PMID:11724780.

Tissue-Specific Expression Patterns

• Liver: Highly expressed, critical for hepatic metabolism
• Skeletal Muscle: Essential for exercise-induced metabolic changes
• Heart: Critical during ischemic stress
• Adipose Tissue: Responds to adrenergic stimulation PMID:17253964
• Brain: Specialized functions in neurons, recently discovered role in photoreceptors

Recent Paradigm Shifts (2024)

Autophagy Regulation Controversy

Traditional view: AMPK → inhibits mTOR → activates autophagy
New evidence: AMPK may actually suppress autophagy under certain conditions, particularly amino acid starvation. This represents a major shift in understanding [Web search findings 2024].

Neuronal Specialization

2024 research identified PRKAA2-specific functions in photoreceptor neurons involving IMPDH (inosine monophosphate dehydrogenase) regulation, representing novel therapeutic targets.

Energy Sensing Mechanisms

AMPK α2 responds to:
1. AMP:ATP ratio changes: Primary activation mechanism
2. Upstream kinases: LKB1 complex PMID:14511394
3. Calcium signaling: [PMID:25788287 studies show calcium-dependent activation]
4. Pharmacological activators: Caffeine PMID:19608206, AICAR, metformin

Complex Assembly

AMPK functions as a heterotrimeric complex:
- α subunit (catalytic): PRKAA1 or PRKAA2
- β subunit (regulatory): bridges α and γ subunits PMID:15695819
- γ subunit (regulatory): contains CBS domains for AMP/ATP binding

Metabolic Integration Points

1. Fatty Acid Homeostasis: Inhibits synthesis (via ACC), promotes oxidation
2. Glucose Homeostasis: Context-dependent effects on glycolysis vs gluconeogenesis
3. Cholesterol Homeostasis: Inhibits HMGCR, reducing cholesterol synthesis
4. Energy Homeostasis: Master coordinator of anabolic vs catabolic balance

Annotations Requiring Careful Review

Strong Evidence (Accept):
- Kinase activities (AMP-activated, serine/threonine)
- ATP/nucleotide binding
- Fatty acid metabolic processes
- Energy homeostasis
- Glucose homeostasis

Context-Dependent (Mark as Non-Core):
- Autophagy regulation (complex, bidirectional)
- Circadian rhythm regulation
- Cellular responses to specific stimuli

Likely Over-Annotations (Consider Removal):
- Chromatin remodeling (likely indirect)
- Wnt signaling (likely indirect)
- Steroid biosynthesis (should be "negative regulation")
- Histone H2BS36 kinase activity (overly specific without strong rat evidence)

Future Research Directions

1. Isoform-Specific Functions: Better understanding α1 vs α2 specialization
2. Tissue-Specific Mechanisms: Detailed mechanisms in brain, heart, liver
3. Therapeutic Targets: IMPDH inhibition for photoreceptor disorders
4. Autophagy Paradox: Resolving conflicting evidence about AMPK's role in autophagy

Methodology Notes

This analysis synthesized:
- Experimental papers cited in existing GO annotations
- Recent literature (2020-2024) from web searches
- Established reviews and comprehensive studies
- Tissue-specific functional studies

Evidence Quality Assessment:
- Direct biochemical evidence > Genetic evidence > Computational predictions
- In vivo studies > In vitro studies > Cell culture
- Multiple independent studies > Single study findings

2025-01-14 - Annotation Retirement Fix

Issue: Validation failure due to 25 annotations with GO_REF:0000096 reference that no longer exist in the current GOA file.

Root Cause: The reference GO_REF:0000096 ("Automated transfer of experimentally-verified manual GO annotation data to mouse-rat orthologs") has been retired from the GOA annotation pipeline and replaced with other reference systems like GO_REF:0000121.

Action Taken: Marked all 27 annotations with original_reference_id: GO_REF:0000096 as retired: true to exclude them from GOA validation while preserving the annotation review work.

Annotations Affected: All annotations with ISO evidence type using GO_REF:0000096, including:
- GO:0031669 (cellular response to nutrient levels)
- GO:0004679 (AMP-activated protein kinase activity)
- GO:0004674 (protein serine/threonine kinase activity)
- GO:0140823 (histone H2BS36 kinase activity)
- GO:0042149 (cellular response to glucose starvation)
- Multiple other metabolic and regulatory terms

Validation Status: After marking these annotations as retired, the gene should pass GOA validation checks since retired annotations are now excluded from validation.

Note: These annotations represent legitimate functional information that was previously transferred from experimentally verified mouse/human data. The retirement only reflects changes in the GOA annotation pipeline, not changes in the underlying biology.

Re-review 2026-10-10

Re-review after the GOA/UniProt refresh (commit a3cf70b6d). 193 rows in total.

What GOA changed

• Large turnover. 25 rows were newly seeded and 36 rows were newly retired; 25 rows had
  already been retired before this refresh, so 61 rows are now retired in total (review text kept).
• Newly retired: the whole GO_REF:0000043 (UniProt keyword) IEA block (nucleotide binding, kinase
  activity, transferase activity, lipid/fatty acid/steroid/cholesterol metabolic and biosynthetic
  terms, autophagy, Wnt signaling, rhythmic process, metal ion binding, chromatin organization);
  the GO_REF:0000120 ATP binding, protein serine/threonine kinase activity and nucleus rows; the
  GO_REF:0000116 protein serine kinase activity row; the protein binding IPI from PMID:16648175;
  and 15 ISO GO_REF:0000121 rows (Golgi apparatus, ciliary basal body, cytoplasmic
  translation, cellular response to starvation and to amino acid starvation, TORC1 signaling,
  positive and negative regulation of translational initiation, protein localization to lysosome,
  protein K6-linked ubiquitination, positive regulation of TORC1 signaling, phosphatidylethanolamine
  and phosphatidylcholine biosynthesis, positive regulation of cytochrome c release, hepatocyte
  apoptotic process).
• Newly seeded (all resolved): donor-split ISO rows from human PRKAA2 (UniProtKB:P54646) and mouse
  Prkaa2 (MGI:MGI:1336173), ISS rows from P54646 and mouse UniProtKB:Q8BRK8, re-sourced IEA rows
  (EC:2.7.11.1, InterPro, UniProtKB-SubCell nucleus and late endosome), adiponectin-activated
  signaling pathway (ISO, P54646), and four EXP rows from UniProt catalytic-activity curation:
  GO:0047322 and GO:0106310 from PMID:2369897 (HMG-CoA reductase) and PMID:9029219 (acetyl-CoA
  carboxylase).

Judgments

• EXP GO:0047322 from PMID:9029219 (an ACC paper) is ACCEPT: GO:0050405 [acetyl-CoA carboxylase]
  kinase activity is obsolete and replaced_by GO:0047322, whose xrefs include RHEA:20333 (the ACC
  reaction) PMID:9029219.
• Protein binding (GO:0005515): all three IPI rows are REMOVE (none left as MARK_AS_OVER_ANNOTATED).
  PFKFB2 and ChREBP are kinase substrates [PMID:11069105 "Heart PFK-2 was phosphorylated on Ser466
  and activated by AMPK"; PMID:11724780 "AMPK specifically phosphorylated Ser(568) of ChREBP."], so
  no binding-type MF is supported beyond the catalytic activity already annotated; the PMID:16648175
  partners are other AMPK subunits, captured by the complex row. Removal does not mean the
  interactions are false.
• Action changes on previously reviewed rows: protein kinase activity (IEA InterPro and IDA
  PMID:12065578) REMOVE -> MODIFY to AMP-activated protein kinase activity (too general, not wrong);
  protein-macromolecule adaptor activity (IDA PMID:15695819) REMOVE -> UNDECIDED on this branch
  (the earlier removal rested on the paper being about the beta subunit); superseded on merge with
  main by the later OpenScientist follow-up below, which restores REMOVE; regulation of stress granule assembly (ISO mouse) MARK_AS_OVER_ANNOTATED ->
  KEEP_AS_NON_CORE PMID:27430620; cellular response to prostaglandin E stimulus (ISO mouse)
  MARK_AS_OVER_ANNOTATED -> UNDECIDED (cached abstract of PMID:23479225 does not describe the AMPK
  experiment).
• Remaining MARK_AS_OVER_ANNOTATED active rows (regulation of microtubule cytoskeleton organization;
  positive regulation of protein localization) carry an explicit overshoot statement.
• Donor named in reason on every new ISO/ISS row; missing reasons added to three legacy rows
  (negative regulation of TOR signaling ISS; negative regulation of TORC1 signaling ISS and ISO).
• No stale UniProtKB: quotes remain (checker reports 0).
• Description: rat symbol used, activation mechanism (gamma-subunit nucleotide binding, Thr-172
  phosphorylation by LKB1/STK11) and mTORC1 substrates added.
• core_functions: added in_complex nucleotide-activated protein kinase complex; replaced negative
  regulation of TOR signaling with the more specific negative regulation of TORC1 signaling; the
  HMGCR/ACC kinase function is now linked to regulation of lipid metabolic process instead of
  cholesterol metabolic process (AMPK regulates, rather than executes, lipid metabolism).

Open questions

• PMID:15695819 protein-macromolecule adaptor activity (IDA) on the alpha subunit needs a full-text
  check.
• PMID:23479225 (cellular response to prostaglandin E stimulus, mouse donor IGI) needs a full-text
  check of the AMPK experiment.
• Two advisory validator warnings remain because retired GO_REF:0000096 rows keep their legacy
  MARK_AS_OVER_ANNOTATED actions while the active GO_REF:0000121 rows were re-judged.

Review follow-up 2026-10-17

In response to the PR #4625 review, the GO:0047322 EXP (PMID:9029219) ACCEPT no longer depends
on a remembered obsoletion. It now rests on checkable mappings: the Prkaa2 GOA IEA GO_REF:0000120
row maps ARBA:ARBA00098394|RHEA:20333|RHEA:23172|EC:2.7.11.31 onto GO:0047322; the UniProt
CATALYTIC ACTIVITY line for RHEA:20333 is the ACC reaction, with evidence PubMed:9029219; and
QuickGO (checked 2026-10-17) lists RHEA:20333 among the GO:0047322 xrefs and shows GO:0050405
as obsolete ("this term was based on EC:2.7.11.27 which has been transferred to EC:2.7.11.31").
OLS gives term_replaced_by GO:0047322, so ACCEPT stands and there is no live ACC-specific term
to MODIFY to. The core_functions wording "merged into it" was replaced by the xref statement.
The boilerplate "Generated using OpenAI Deep Research API" quote on the IBA protein
serine/threonine kinase activity row was replaced by a real quote, plus PMID:9029219, and the
row now has a reason. The earlier Judgments bullet stated the obsoletion without a source; it
is now confirmed against QuickGO/OLS.

2026-10-10 - OpenScientist adaptor-activity follow-up

The focused OpenScientist report for prediction-adaptor-activity resolves the
GO:0030674 protein-macromolecule adaptor activity row as a subunit-level
misassignment. PMID:15695819 shows that the AMPK beta subunit's conserved
C-terminal 85 residues tether alpha and gamma into the heterotrimer; Prkaa2 is
the alpha-2 catalytic subunit being tethered, not the adaptor. The existing
GO:0044877 protein-containing complex binding row remains the correct MF
shape for alpha-subunit incorporation into AMPK.

Merge note (2026-10-17): this section arrived from main (PR #4647) while the re-review branch
had independently moved the same row to UNDECIDED. On merge the main-side judgment (REMOVE,
with PMID:15695819, PMID:18079111 and the OpenScientist report as support) was kept as the more
recent and more informative one; the rest of the row ordering and formatting follows the
re-review branch. Both cached sources are abstract-only, so the full-text check listed under
open questions still applies.

2026-10-28 - adaptor-activity follow-up corrected to UNDECIDED

The singleton sidecar audit supersedes both the 2026-10-10 conclusion and the
2026-10-17 merge note above. The GO:0030674 protein-macromolecule adaptor activity row is now UNDECIDED, and the OpenScientist sidecar reference_review
is HIGH/DISPUTED. The beta-scaffold misassignment case is plausible, but the
PMID:15695819 cache is abstract-only while the row is an RGD IDA whose curator
read the full paper. The cached abstract itself reports that deleting beta
Ile-270 blocks beta-gamma association only without alpha, and that alpha can
restore beta-gamma binding, so the full text and curator rationale are needed
before the Prkaa2 alpha-subunit row can be removed confidently.