2026-07-14

2026-07-14 — ad-hoc daily log

[2026-07-14] ingest + verify | engineered protein-bound CML deglycation (user-prompted)

Trigger: user supplied Trabosh et al. 2026, Nature Communications, doi:10.1038/s41467-026-75141-2.

Added and independently verified:

• studies/trabosh-2026-cmlase-deglycation.md — full extraction of the main paper, supplement, reporting summary, and deposited source data.
• molecules/compounds/cmlase.md — CMLase/CrGO-897 biologic page with age-adduct-deglycation mechanism, preclinical status, no active ClinicalTrials.gov record, and explicit delivery/safety/function gaps.
• age-adduct-deglycation in frameworks/intervention-classes.md — direct repair of a stable monovalent AGE adduct, kept distinct from age-crosslink-cleavage.

Scientific adjudication: CMLase is a real mature-damage repair enzyme: it oxidatively removes protein-bound Nε-carboxymethyl-lysine (CML) and regenerates native lysine. It is not an AGE crosslink breaker. CML modifies one lysine and does not bridge two residues; the paper did not test glucosepane, pentosidine, or another bifunctional crosslink.

Evidence: model-protein ELISA fell approximately 53–97%; site proteomics found lower CML at 30/33 mapped sites. A single 64-year-old human lens showed 45% lower total CML by hydrolysis/LC-MS/MS (78% by ELISA). The paper states >70% lower aortic staining, but deposited means imply approximately 66% versus inactive enzyme; skin staining fell >55%. Seven human donors appear across all figures, but each headline older lens/aorta/skin result uses one donor, and image regions/assay replicates are not independent subjects. No living mammalian efficacy, tissue function, delivery, PK, immunogenicity, or safety study was performed.

Source inconsistencies retained: the paper states 15 substitutions although round-listed additions total 16; the aortic text percentage does not reproduce from deposited means; and Figure 3/Results call the site-proteomics enzyme CMLase/CrGO-897 while the corresponding Methods paragraph names CrGO-865.

Prior-literature correction: full-text review of Delanghe et al. 2024 (PMCID:PMC11083825) showed that its reported CML/CEL disappearance came from untargeted UHPLC-HRMS of free amino-acid/sugar mixtures, not intact protein-bound CML repair or native-lysine restoration. The result remains suggestive FAOD precedent, not independent CMLase replication.

Propagation: updated carboxymethyl-lysine, advanced-glycation-end-products, microbial-amadori-deglycation, glucosepane, age-crosslink-breakers, sens-damage-categories, causal-graph-data, and mass-spec-age-hydrolysates. The causal graph and simulator continue to treat CML signaling separately from the ECM-crosslink state; no numerical CMLase intervention was added.

Open gaps: intact-ECM penetration; in-vivo target engagement; functional rescue of RAGE signaling, arterial/skin mechanics, or lens physiology; bacterial-enzyme immunogenicity; H₂O₂/glyoxylate safety; durability/re-dosing; proteome-wide off-targets; independent replication; and a genuine glucosepane-cleaving enzyme.