LOX-Mediated Collagen Crosslink Density as Bond Occupation Probability -- Corrected Pore Geometry and Heterogeneity-Smeared Transition

LOX is expressed in tumor-relevant solid tumor tissues (lung, breast, colon) — necessary for LOX to act as the ECM crosslink density actuator in these tumor types

LOX (ENSG00000113083) is broadly expressed across all three tumor-relevant tissues: Lung (is_expressed=true, BROADLY_EXPRESSED, Low tissue specificity, Detected in many), Breast (is_expressed=true, BROADLY_EXPRESSED, Low tissue specificity, Detected in many), Colon (is_expressed=true, BROADLY_EXPRESSED, Low tissue specificity, Detected in many). The low tissue specificity means LOX is ubiquitous rather than tumor-selective, but its presence in all three major solid tumor tissue types confirms it can function as an ECM crosslink density actuator across the tumor panel proposed in E1.

LOX is a secreted enzyme that oxidatively deaminates lysine residues in collagen and elastin to form covalent crosslinks — the specific biochemical mechanism that maps to bond occupation probability p

UniProt P28300 confirms: Function = 'Responsible for the post-translational oxidative deamination of peptidyl lysine residues in precursors to fibrous collagen and elastin (PubMed:26838787)'. Subcellular location = 'Secreted, extracellular space'. The secreted extracellular location means LOX acts directly on the collagen fiber network that forms the physical barrier for T cells. The crosslink-forming reaction is a discrete covalent bond event, making it mechanistically sound to model each potential crosslink site as occupied (LOX-catalyzed) or unoccupied (no crosslink) — the direct mapping to bond occupation probability in percolation theory.

LOX has no experimentally resolved crystal structure — BAPN-based pharmacological intervention in E1 relies on biochemical inhibition constants (KI), not structure-guided binding

PDB query: total_pdb_structures=0. AlphaFold model AF-P28300-F1 available with mean_pLDDT=67.38 (moderate confidence, below the 70 threshold indicating structurally uncertain regions). The absence of any experimental crystal structure means there is no 3D binding pose data for BAPN-LOX interaction. The moderate pLDDT (67.38) suggests LOX has regions of structural disorder, consistent with it being a secreted enzyme that may adopt multiple conformations in the extracellular matrix. This confirms the hypothesis is correct to rely on biochemical KI data (Tang 1983) rather than structure-guided estimates.

LOX gene variants have GWAS associations with cancer or immune phenotypes — potential genetic epidemiology support for LOX-mediated immune exclusion

GWAS Catalog found 20 SNPs in the LOX locus but returned 0 trait associations in both the cancer and immune phenotype queries. The API confirms the gene exists (20 SNPs retrieved) but the trait-association endpoint did not return mappable results. Per constraint 6, this is recorded as NO_DATA (API retrieval limitation, not absence of associations). The 2024-2025 literature (PMID 38267662, 38305736) provides functional evidence for the LOX-T cell link that GWAS data cannot confirm programmatically here.

LOXL1 (LOX-family member confirmed to restrict CD8+ T cell infiltration in colorectal cancer) is a secreted collagen/elastin crosslinker — mechanistically equivalent to LOX for the percolation framework

UniProt Q08397 confirms: Function = 'Catalyzes the oxidative deamination of lysine and hydroxylysine residues in collagen and elastin, resulting in the formation of covalent cross-linkages, and the stabilization of collagen and elastin fibers'. Subcellular location = 'Secreted, extracellular space' and 'extracellular matrix'. LOXL1 is mechanistically identical to LOX for ECM crosslinking (same reaction, same substrates, same ECM localization). This validates PMID 38267662 as direct mechanistic support for E1: LOXL1-mediated crosslinking restricts T cells, confirming the LOX -> crosslink density -> T cell exclusion causal chain that E1 formalizes as a percolation bond occupation probability.

KEGG pathway membership for LOX confirms (or fails to confirm) its position in ECM remodeling or immune pathways

KEGG returned 0 pathways for LOX (hsa:4015), total_pathways=0. This matches the CV finding (API timeout/empty response). LOX is not represented in KEGG pathway annotations as of this query. This is an expected documentation gap — KEGG captures well-established canonical pathways, and the LOX-to-T-cell-exclusion connection is supported by 2024-2025 literature but not yet integrated into KEGG curation. The absence from KEGG does not indicate the mechanism is invalid; it indicates the mechanism is too recent for curated pathway integration.

BAPN (beta-aminopropionitrile) inhibits LOX-family enzymes with measurable IC50/KI — the pharmacological foundation for the p(dose) mapping function

ChEMBL returns 10 activity entries for BAPN (CHEMBL1618272) against LOXL2 (CHEMBL3714029): IC50 values of 83, 396, 66, 243, 240, 66 nM across 6 IC50 assays, plus 3 inhibition percentage entries and 1 Ratio IC50 = 6. A second ChEMBL target entry (CHEMBL4105831) returns 1 additional activity (IC50 = 128 nM). IMPORTANT FINDING: ChEMBL matches BAPN to LOXL2, not LOX itself. The IC50 range 66-396 nM against LOXL2 is 10-100x lower than the KI = 6 uM reported by Tang 1983 for purified LOX, and 100-1000x lower than the hypothesis's estimate of K_I^BAPN ~ 50-200 uM in cell culture. Three data sources now diverge: ChEMBL (LOXL2: 66-396 nM), Tang 1983 biochemistry (LOX: 6 uM), hypothesis estimate (cellular: 50-200 uM). The qualitative conclusion — BAPN inhibits LOX-family enzymes — is confirmed; the quantitative dose parameters have 3-log-order uncertainty across LOX family members and assay contexts.

LOX is a monomeric enzyme — the hypothesis uses this to argue that cooperative kinetics cannot generate false-positive percolation signatures in the dose-response

UniProt P28300 entry does not annotate a specific oligomeric state in the returned fields (function, subcellular_location, domains are returned; the subunit annotation was not returned by the script's field extraction). The function annotation confirms LOX as an oxidative deaminase targeting collagen lysines — consistent with a monomeric enzyme acting site-by-site on substrate. Lucero & Kagan 2006 (PMID 16909208, Cell Mol Life Sci) confirms the 32 kDa monomeric processed form. No oligomeric state or allosteric cooperativity is reported for LOX in UniProt, supporting the hypothesis's claim that cooperative kinetics cannot generate false-positive percolation signatures. This is DATA_SUPPORTED rather than DATA_CONFIRMED because the script did not return explicit oligomeric state annotation from UniProt.

BAPN selectivity concern: LOXL2 has different structural features from LOX (SRCR domains, nuclear/ER localization) — ChEMBL and PDB data reveal selectivity context for the p(dose) model

UniProt Q9Y4K0 reveals LOXL2 has a distinct biology from LOX: (1) LOXL2 has SRCR domains (4 scavenger-receptor cysteine-rich domains absent in LOX), (2) subcellular locations include nucleus and chromosome in addition to ECM/secreted — LOXL2 acts as a transcription corepressor by deaminating histone H3K4me3. LOXL2's nuclear role means BAPN inhibition of LOXL2 has cell-intrinsic effects beyond ECM crosslinking. PDB confirms LOXL2 has 1 experimental crystal structure (5ZE3, X-ray, 2.40 A, chains 318-774) and AlphaFold model with pLDDT=86.12 (HIGH confidence). This structural data for LOXL2 but not LOX suggests BAPN's high potency for LOXL2 (66-396 nM) may reflect a structurally tractable active site, while LOX's lack of crystal structure corresponds to its moderate biochemical KI. For E2's p(dose) model, this is a relevant selectivity nuance: BAPN at low doses may primarily inhibit LOXL2 (ECM + nuclear), while higher doses are needed to fully inhibit LOX (ECM only). The percolation framework's p(dose) curve may therefore have two components unless BAPN is replaced by a LOX-selective or pan-LOX agent.

LOX lacks crystal structure (PDB: 0 structures) — BAPN KI determined biochemically by Tang 1983, not by structural binding analysis

PDB: total_pdb_structures=0 for LOX (P28300). AlphaFold available (pLDDT=67.38, moderate confidence). This confirms that all BAPN binding constants for LOX are from biochemical assays, not structural analysis. The hypothesis's KI estimate of 50-200 uM (cellular) is a parametric extrapolation from Tang 1983's KI = 6 uM (purified enzyme, 37C, pH 8.0 assay). The structural void also explains the 3-log discrepancy in ChEMBL data (LOXL2 crystal structure 5ZE3 enables structure-guided inhibitor characterization; LOX lacks this). Contrast with LOXL2: 1 PDB structure (5ZE3), pLDDT=86.12 — this asymmetry in structural data availability directly correlates with the nM (LOXL2) vs uM (LOX) IC50 range.

LOX expression is confirmed across solid tumor tissues (lung, breast, colon) — requirement for the BAPN dose-response experiment in 4T1 (breast) and KPC (pancreatic) mouse models

HPA confirms LOX is broadly expressed (is_expressed=true, BROADLY_EXPRESSED, Detected in many) in all three queried tissues: Lung, Breast, and Colon (low tissue specificity in all). This supports the in vivo experimental design: LOX is present in the tumor microenvironments of 4T1 (breast) and KPC (pancreatic/lung-metastatic) models, meaning BAPN will find its target. The low tissue specificity means BAPN will also inhibit LOX in other tissues, which is a known toxicity consideration (lathyrism at high doses) already noted in the hypothesis.

KEGG pathway membership for LOX — would confirm curated pathway position for the p(dose) model mechanistic context

KEGG returned 0 pathways for LOX (hsa:4015), consistent with CV finding. LOX is not represented in curated KEGG pathway entries. Not unexpected: LOX catalyzes post-translational collagen modification in the ECM — a process that is biologically well-established but not yet integrated into KEGG's pathway-centric curation (which focuses on signaling cascades and metabolic pathways). Contrast: TGFB1 (upstream regulator of LOX) is in 36 KEGG pathways, including hsa04350 (TGF-beta signaling) — the regulatory axis that drives LOX expression in immune-cold tumors is KEGG-confirmed even if LOX itself is not.