RAG1 S382* pathway hypothesis — what is observed, inferred, and testable

Short source note: Sam's RAG1 pathway note

Full supplied analysis: YugaBio RAG1 / AMPK / mTOR / BCL-2 conversation

PARP/RAG follow-up: Krita PARP/RAG synthesis and reconstructed source map

Variant source: May 12 preliminary Altera DNA/RNA report

Bottom line

The preliminary Altera table lists RAG1 S382* at 14% variant allele fraction and BCL2L1 E179Q at 7%. Those calls justify a validation question, but they do not yet establish that either mutant transcript or protein exists, that either change is functionally important in breast-tumor cells, or that the tumor is sensitive to AMPK activation, mTOR inhibition, a BH3 mimetic, DNA-damaging therapy, or radiation.

Three corrections are load-bearing:

RAG1 is not a general DNA-repair protein. RAG1/RAG2 is a lymphoid-specific endonuclease that creates programmed DNA breaks during V(D)J recombination; non-homologous end joining repairs those breaks. A breast-tumor DNA call does not establish that epithelial tumor cells express or use RAG1.
The direction of the AMPK result is reversed in the short note and parts of the Yuga export. The primary 2013 study reports that AMPK phosphorylation of RAG1 at S528 enhanced RAG catalytic activity and increased recombination in the tested substrates. It did not show downregulation, degradation, or relocalization of RAG1. Um et al., PNAS 2013
S382* is a nonsense stop, not a frameshift. Under standard translation it ends the protein without encoding a novel amino-acid tail. The truncation therefore is not automatically a useful neoepitope. A mutation-specific peptide would require evidence for an actually translated novel sequence, such as an alternate transcript, splice junction, readthrough product, or a corrected variant annotation.

Evidence ladder

Level	What the current record supports	What it does not support
Observed in the preliminary report	`RAG1 S382*` at 14% VAF; `BCL2L1 E179Q` at 7% VAF	Matched-normal status, clonality, biallelic loss, RNA expression, protein expression, or functional effect
Protein-sequence inference	If the annotated RAG1 transcript is expressed and translated, a stop at residue 382 would remove the canonical catalytic core and S528	Stable truncated protein, dominant-negative behavior, or a general DNA-repair defect
Primary RAG1 mechanism	AMPK can phosphorylate RAG1 S528 and enhance RAG activity in lymphoid recombination assays	That AMPK activation selectively kills `RAG1 S382*` breast-cancer cells
Breast-cancer combination evidence	PI3K/mTOR plus BCL-2/BCL-XL inhibition has shown synergy in breast-cancer cell models; everolimus can block adaptive MCL-1 translation in ER-positive models	A validated TNBC regimen, a clinical benefit claim, or an `RAG1 S382*` / `BCL2L1 E179Q` biomarker. Hamunyela et al. 2017 · Williams et al. 2019
Exact BCL2L1 allele	E179Q is a charge-neutralizing substitution in BCL-XL	The supplied export's claim that it likely weakens BCL-XL or increases drug sensitivity. No exact `BCL2L1 E179Q` functional paper was found in a PubMed title/abstract search on August 30, 2026

The Yuga export is useful as a hypothesis trail, but its numbered citations and referenced “supporting materials” are not included in the exported file. Treat claims that cannot be traced to a linked primary source as unverified.

PARP–RAG cross-check

The September 1 synthesis is directionally correct that its cited literature does not show a direct PARP1/2/3–RAG1/2 protein interaction or a RAG1-specific PARP-inhibitor vulnerability in epithelial breast cancer. Two case statements need correction:

“The tumor has no functional RAG1” is not established. A preliminary RAG1 S382* call at 14% VAF does not by itself prove biallelic loss, malignant-cell expression, stable truncated protein, or absence of all RAG-generated breaks. It may ultimately prove biologically silent in the breast tumor, but that conclusion needs the existing DNA, RNA, cell-compartment, and protein gates.
The BRCA1 rationale is independent of RAG1, but it still needs clinical validation and setting-specific review. The tumor/research record supports an HRD/BRCA1 confirmation question; it should not be shortened to an already-established clinical PARP-inhibitor indication. Eligibility and treatment choice depend on a validated biomarker, disease setting, prior therapy, current approvals/guidelines, and the oncology team.

The mechanistic lane reconstructed from the ten cited primary papers is narrower:

With intact classical NHEJ, PARP1 is not required for ordinary V(D)J recombination in the reported mouse systems. In Ku70-deficient lymphoid models, PARP1–XRCC1/LigIII alternative end joining can repair a subset of RAG-induced breaks.
RAG post-cleavage-complex stability, RSS sequence, and RAG2 mutations can change whether broken ends become accessible to alternative repair machinery.
The class-switch, gene-conversion, B/T-cell development, and general DSB-pathway studies describe lymphoid diversification or DNA-repair context. They do not validate RAG1 S382* as a PARP biomarker in breast cancer.
Constitutive single- or double-gene knockout is not equivalent to clinical PARP-inhibitor target coverage, PARP trapping, exposure, timing, or systemic effects. The synthesis's explanation for relatively modest clinical immunosuppression is therefore an inference, not a direct result of the cited mouse studies.

See the PARP–RAG primary-paper collection for full source status and boundaries.

What is worth verifying

1. Establish the variants before interpreting pathways

Reconcile genome build, transcript, codon, strand, read support, mapping quality, and tumor purity for both calls.
Check matched normal and an orthogonal tumor assay before calling either alteration tumor-only.
Estimate cancer-cell fraction and copy state; low VAF alone cannot distinguish subclonality from purity or local copy-number effects.
Confirm whether RAG1 S382* is truly a single-nucleotide stop-gain rather than a frameshift, splice event, or annotation artifact.

2. Ask whether the genes are active in malignant cells

Review gene-level RNA abundance and allele-specific reads for RAG1 and BCL2L1, with tumor-cell versus lymphocyte composition in mind.
Inspect splice junctions and transcript coverage around RAG1 exon/codon 382 for nonsense-mediated decay or alternate isoforms.
Use protein testing only with validated reagents and controls. RAG1 IHC in a lymphocyte-containing breast specimen could report immune cells rather than malignant-cell expression.

3. Keep the neoepitope gate sequence-aware

Do not rank S382* as a vaccine peptide merely because it truncates a protein; the ordinary upstream peptide sequence remains wild type.
Require a defined mutant amino-acid sequence, patient HLA binding, mutant RNA expression, and preferably immunopeptidomics or a mutation-specific T-cell assay.
If no novel translated sequence exists, treat loss of downstream RAG1 peptides as loss of antigen, not creation of a neoepitope.

4. Measure pathway state before considering a combination

IHC or phosphoprotein measurements could support a pathway-state screen, but static staining cannot establish drug dependence:

Question	Candidate readouts	Boundary
Is AMPK engaged?	p-AMPK and the downstream substrate p-ACC	Activation state, not proof of `RAG1` coupling or selective killing
Is mTORC1 active?	p-S6 and p-4EBP1, interpreted with total protein and controls	Pathway activity, not proof that an mTOR inhibitor will produce apoptosis
Which anti-apoptotic proteins are present?	BCL-XL (`BCL2L1`), MCL-1, BCL-2, BIM and related family members	Abundance is not mitochondrial dependence and does not establish the effect of E179Q
What is the autophagy state?	LC3B and p62/SQSTM1 with morphology and compartment annotation	Static IHC cannot distinguish increased flux from blocked degradation

The higher-value functional bridge is dynamic BH3 profiling plus an ex-vivo dose matrix of the specific mTOR/PI3K and BH3-mimetic agents, with single agents, combination controls, normal-cell controls, and predefined synergy/toxicity rules. If viable material is unavailable, the hypothesis remains untested; protein IHC alone should not be promoted into a treatment recommendation.

Clinical boundary

The breast-cancer evidence cited here is preclinical and not specific to either reported allele. The direct ALL trial mentioned in the Yuga export, NCT07522801, is a not-yet-recruiting relapsed/refractory T-ALL study and is not evidence for breast-cancer eligibility or efficacy. Adding DNA-damaging drugs or radiation to this proposed combination is another independent hypothesis with substantial normal-tissue and marrow-toxicity implications; it should not be inferred from the two preliminary variant calls.

For Diana, the immediate value is a compact validation packet, not a drug combination: confirm the calls, malignant-cell expression, protein/pathway state, and functional dependence in that order.