LOXL1 review notes

Evidence blockers and provenance

• Automated Falcon deep research timed out. The Perplexity fallback failed with HTTP
• Per project policy, no LOXL1-deep-research-{provider}.md file was created; this
  manual journal records the fallback review.
• PMID:37602378 is a decisive citation blocker for the seeded IDA extracellular-region
  annotation: the verified record is a HELZ2 exoribonuclease paper and contains no LOXL1
  result PMID:37602378.
  This is flagged as WRONG_IDENTIFIER at reference level. Because extracellular
  localization is independently correct but the provenance of this experimental row
  cannot be reconstructed, its action is UNDECIDED under curator-deference rather than
  removal. The cause of the citation mismatch is unknown.
• Direct catalytic evidence is species-bounded. The strongest LOXL1-specific in vivo
  evidence is mouse, and the direct purified-enzyme activity evidence is bovine. The
  current human UniProt function statement is explicitly by similarity to mouse. Do not
  present the human protein-lysine 6-oxidase assignment as if a purified human LOXL1
  catalytic assay were available in this cache.
• Human LOXL1 processing is experimentally mapped, but processing is not equivalent to a
  direct activity assay. The 2022 human study mapped BMP1 and ADAMTS14 cleavage, while
  noting that the older bovine cleavage-site assignments disagree. The safest conclusion
  is that human LOXL1 is secreted and proteolytically diversified, not that every observed
  processed species is catalytically activated.
• Three HDA ECM citations (PMID:27068509, PMID:28327460, PMID:28675934) do not name LOXL1
  in the cached narrative text. Their annotation support may reside in supplements or
  deposited proteomics tables. Treat them as UNVERIFIED, not false.

Identity and molecular boundaries

LOXL1 (Q08397) is the short-subfamily paralog most closely related to canonical LOX, not
an alternate name for LOX/P28300 and not one of the SRCR-domain LOXL2/3/4 proteins. The
original human cDNA paper established homology only to the carboxyl end of lysyl oxidase
PMID:7689553.

The reviewed human record defines a signal peptide (residues 1-25), propeptide (26-95),
and mature chain (96-574), but newer proteomics demonstrates a more complex processing
landscape. In human cellular models, BMP1 cleavage was mapped to 151-152, whereas
ADAMTS14 sites were mapped to 216-217, 292-293, and 375-376
PMID:35328709.
Therefore, avoid treating a single N-terminal fragment as a uniquely defined LOXL1
propeptide with an independently established function, and avoid transferring canonical
LOX's Gly168-Asp169 processing scheme to LOXL1.

Prioritized functional synthesis

1. Elastic-fiber homeostasis and elastin cross-linking are the best-supported core.
   Loxl1-null mice fail to deposit normal postpartum uterine elastic fibers and show lung,
   skin, and vascular abnormalities with tropoelastin accumulation
   PMID:14745449.
   The same abstract distinguishes LOXL1 from LOX and assigns spatially targeted elastin
   deposition to LOXL1 PMID:14745449.
   This supports elastic-fiber assembly/homeostasis more strongly than a broad claim that
   LOXL1 is a general collagen-fibril organizer.

2. LOXL1 is targeted to an elastogenic scaffold through fibulin-5. Human recombinant
   binding experiments detected LOXL1-fibulin-5 interaction and localized the major
   binding contribution to the fibulin-5 C-terminal domain
   PMID:17371835.
   The interaction is mechanistically informative but not unique to LOXL1, because LOXL2
   and LOXL4 also bound, and it does not establish one stable macromolecular complex.
   Fibulin-4 binding is also reported PMID:27339457,
   but that paper is centered on fibulin-4 variants and should be treated as supporting,
   not defining, evidence.

3. The mature catalytic chain is extracellular and generated through regulated
   proteolysis. Human Tenon's-capsule fibroblasts and engineered human LOXL1 expression
   models produced multiple extracellular LOXL1 species
   PMID:35328709.
   Direct activation evidence comes from bovine LOXL1: the purified precursor was largely
   inactive and BMP1 processing yielded activity on elastin and collagen
   PMID:11684696.
   This supports a conserved processing-to-activity model with an explicit bovine-to-human
   inference boundary.

4. Human tissue evidence supports extracellular-matrix/elastic-tissue deployment, not
   direct catalytic specificity. LOXL1 was detected in an extracellular-space-enriched
   guanidine fraction from human aorta PMID:20551380.
   In human ocular tissues, LOXL1 RNA was present broadly except in retina, and protein was
   detected in relevant tissues PMID:18037624.
   These localization data are compatible with an extracellular elastogenic role, but the
   ocular paper is a disease-association study and does not assay catalysis.

Annotation-facing cautions

• Favor the specific molecular function protein-lysine 6-oxidase activity over the broad
  parent oxidoreductase term, while documenting that human assignment is substantially
  orthology-supported.
• Elastic-fiber assembly/homeostasis is the strongest biological-process axis. Collagen
  oxidation is biochemically plausible and directly demonstrated for processed bovine
  LOXL1, but human LOXL1-specific collagen-fibril organization evidence is weaker than the
  elastin evidence.
• Copper binding and the lysine-tyrosylquinone cofactor are conserved-family catalytic
  requirements in the reviewed UniProt record; they are not independently established by
  the LOXL1 primary papers reviewed here.
• Acrosomal-vesicle, basement-membrane, aorta-development, LPS-response, and generic
  collagen-fibril annotations are largely electronic transfers from rodent orthologs and
  should not be promoted into the core without LOXL1-specific corroboration.
• Reactome R-HSA-2002466, R-HSA-2022141, and R-HSA-2395340 explicitly describe canonical
  LOX. R-HSA-2129375 supplies general lysyl-oxidase/elastin pathway context but is not
  independent LOXL1-specific evidence.

Manual source audit

Primary records were checked against PubMed/PMC and then fetched through just fetch-pmid
so that all YAML supporting text could be validated against repository-local sources.
New decisive caches are PMID:14745449, PMID:35328709, PMID:17371835, PMID:11684696, and
PMID:18037624. No finding was imported from an LLM-generated research summary.