DISTILLATION №27 — DISCARDED

FOXO4-DRI · N-terminal Palmitoyl-Lys Conjugate · Pharmacokinetic Extension via Albumin Anchoring

Mechanism of Action (Background)

FOXO4-DRI is a D-retro-inverso senolytic peptide that mimics the CR3 domain of the transcription factor FOXO4. In senescent cells, FOXO4 forms a complex with p53 that sequesters p53 in the nucleus and suppresses p53-driven apoptosis — a key mechanism by which senescent cells resist clearance. FOXO4-DRI competes with endogenous FOXO4 for binding to the p53 transactivation domain 2 (p53TAD2), displacing FOXO4 and restoring p53-mediated apoptotic signaling. The downstream axis — nuclear exclusion of p53, BCL-2 downregulation, Caspase-3 activation — has been documented in senescent Leydig cells, endothelial cells, chondrocytes, keloid fibroblasts, and cancer-associated fibroblasts across multiple independent studies.

A critical structural insight from a 2025 NMR study (PMID:40593617, Bourgeois et al.) reframes the peptide's architecture: the C-terminal cationic CPP tail (RPPPRRRQRRKKRG) is not merely a cell-penetration appendage — it makes direct contact with p53TAD2 as part of the binding interface. The CPP tail is therefore bifunctional: it drives membrane translocation AND contributes to target engagement. This finding is central to evaluating any modification that might alter the CPP tail's conformation, charge, or accessibility. Fold #12 in this lab confirmed the structural indispensability of this region: truncating the CPP tail collapsed pLDDT to 0.56 and yielded a DISCARDED verdict.

Modification Hypothesis (What We Tested)

This fold attached a C16 palmitoyl fatty acid to the N-terminus of FOXO4-DRI via a Lys spacer (Pal-Lys-LTLRK...), mirroring the albumin-anchoring strategy used in liraglutide (C16 palmitoyl) and semaglutide (C18 with linker). The hypothesis:

• FOXO4-DRI's highly cationic structure and small size likely drive rapid renal clearance, limiting tissue exposure after bolus dosing
• N-terminal palmitoylation would enable reversible albumin binding (Kd ~1–10 µM for C16), converting the peptide from a flash-exposure agent to a sustained-exposure depot
• Attaching the lipid at the N-terminus — maximally distal from the C-terminal CPP tail — would minimize steric disruption to either functional domain
• The all-D-amino-acid backbone (proteolytic stability already ensured by the DRI design) would make albumin binding the primary PK extension mechanism

The design rationale was sound at the level of analogy: liraglutide/semaglutide demonstrate that N-terminal fatty acid conjugation can extend half-life from minutes to hours-to-days. The cyclic IP dosing used in all FOXO4-DRI in vivo studies implicitly acknowledges short persistence. Extended half-life would directly increase the time the peptide is available in senescent-cell-rich tissues where p53 phosphorylation (pS15) enhances binding affinity.

Why the Prediction Was Uninformative (Technical Analysis)

Structural confidence metrics:

• pLDDT: 0.58 — below the threshold for reliable secondary structure assignment in a peptide of this complexity
• ipTM: 0.13 — near the floor of meaningful interface scoring; essentially no confident peptide–protein docking information was generated
• Chai-1 cross-validation: not available — single-model output cannot be treated as a reliable structural prediction
• Boltz-2 affinity module: no values returned — no quantitative binding change predicted

The low confidence scores are not surprising given the target system. FOXO4-DRI is an all-D-amino-acid peptide — structure predictors trained on L-amino acid protein databases have fundamentally degraded performance on DRI peptides. The modified sequence also includes a non-standard palmitoyl-Lys N-terminal extension that is chemically outside the training distribution of most protein structure models. The disordered nature of the CPP tail further limits confidence globally, as unstructured polycationic regions consistently produce low pLDDT regardless of the helical core's actual folding.

What the model could and could not tell us: The prediction confirms that the CR3-mimetic helix is present as a short helical segment and that the palmitoyl-Lys extension appears as a flexible appendage without obvious clashes. However, at pLDDT 0.58 and ipTM 0.13, this is not a reliable structural prediction — it is a low-confidence sketch. Absence of visible aberrant lipid–helix contacts in the model does not constitute evidence of structural tolerance; the model lacks the resolution and the parameterization to detect subtle hydrophobic insertion or helix-destabilizing contacts in this sequence context.

Comparison with Fold #25 (MOTS-c myristoylation, PROMISING, pLDDT 0.63) is instructive: a slightly higher pLDDT and a shorter, simpler target peptide allowed that fold to clear the PROMISING threshold. FOXO4-DRI's greater length, disordered CPP tail, and all-D backbone make it a systematically harder prediction target, and the ipTM gap between the two (0.13 here vs. not reported but inferred higher for Fold #25) reflects this.

What This Tells Us (Negative Results Are Data)

1. The structure predictor limitation is specific, not general. The prediction failure here is not evidence that palmitoylation disrupts FOXO4-DRI's structure — it is evidence that current in silico tools cannot reliably model all-D, long, polycationic peptides with non-canonical N-terminal modifications. This is an important distinction: the fold is discarded for informational failure, not biological failure.

2. The pharmacokinetic hypothesis has a more fundamental problem than the prediction failure. Literature analysis surfaces two compounding mechanistic incompatibilities that the structural prediction cannot resolve but which should govern future design decisions:

• Albumin anchoring may impair CPP-mediated cell entry. Albumin-bound peptide complexes (~70 kDa) cannot cross cell membranes via the direct electrostatic/CPP mechanism that FOXO4-DRI relies on for nuclear translocation. If the palmitoyl–albumin Kd is not sufficiently weak to allow rapid dissociation at the cell surface, the sustained plasma half-life benefit would come at the cost of reduced intracellular bioavailability.

• Albumin orientation may sterically restrict the CPP tail's target engagement. The 2025 NMR finding (PMID:40593617) that the CPP tail directly contacts p53TAD2 means that albumin binding at the N-terminal palmitoyl anchor could geometrically restrict the CPP tail's freedom to engage p53 in the free-peptide state post-dissociation.

3. Context-dependent safety risk for sustained senolysis. PMID:36515093 documents that FOXO4-DRI worsened pulmonary hypertension by eliminating senescent cells performing a vascular-protective role. A sustained-exposure formulation amplifies this risk by extending the senolytic window beyond the intended dosing interval. This is not a prediction-derived caveat — it is a literature-derived contraindication for the sustained-exposure design strategy in vascular disease contexts.

4. The PK premise is unverified. No published data characterize FOXO4-DRI's actual plasma half-life, renal clearance rate, or dominant elimination pathway. The assumption that rapid renal clearance is the primary PK liability — rather than hepatic clearance, tissue sequestration, or receptor-mediated endocytosis — is biologically plausible but unsubstantiated. Palmitoylation is the right solution only if renal filtration is the rate-limiting clearance mechanism.

Alternative Hypotheses to Test (Avoiding the Failure Mode)

The pharmacokinetic extension goal remains valid. The following strategies avoid the specific failure modes identified here:

1. PEGylation at a non-CPP, non-helix site via Lys spacer in the linker region. The GGK linker between the CR3 helix and CPP tail (around residues 28–30) provides a candidate attachment point. PEG20–40k extends half-life without driving albumin-complex formation, preserving CPP-driven membrane translocation while reducing renal filtration. This is structurally testable with higher confidence than palmitoylation.

2. Cleavable albumin-binding linker (pH- or protease-sensitive release). A palmitoyl group attached via a protease-cleavable dipeptide (e.g., Val-Cit, cleaved by cathepsin B in endosomal/lysosomal compartments or at tumor/senescent-tissue sites) would provide the sustained plasma PK benefit while releasing free FOXO4-DRI intracellularly where it can engage the CPP-driven nuclear pathway. This design has precedent in ADC (antibody-drug conjugate) chemistry.

3. Nanoparticle encapsulation without covalent modification. Lipid nanoparticles or PLGA microparticles loaded with unmodified FOXO4-DRI would provide controlled release and extended local exposure without any risk of CPP-tail interference. This preserves the native peptide sequence entirely and avoids all the structural concerns raised here.

4. Half-life extension via D-amino acid stapling or macrolactamization at a non-functional site. If the pharmacokinetic goal is secondary to improving stability and structural rigidity, hydrocarbon stapling or macrolactamization within the helical CR3 region (tested with D-amino acid stapling chemistry compatible with the all-D backbone) could increase proteolytic half-life and receptor residence time without driving albumin sequestration. The Fold #22 approach on Humanin (disulfide bridge, PROMISING) provides a conceptual precedent in this lab.

5. Direct pharmacokinetic characterization before further modification. Before engineering solutions, measure FOXO4-DRI's actual plasma half-life, renal clearance fraction, and tissue distribution in a rodent model. If half-life is already >2 hours (as the all-D backbone and highly cationic character might produce via tissue sequestration), the PK problem may be smaller than assumed, and the modification priority should shift to affinity or selectivity.

In silico prediction only. All structural and property data are computational estimates. This report does not constitute medical advice. Requires wet-lab validation before any biological conclusions can be drawn.