Worked example: LDLRAP1

Species: human

Worked example: LDLRAP1 / ARH (Q5SW96)

Back to the project page.

Unlike the GRB2 example, which re-read an existing
review, LDLRAP1 was unreviewed when this project picked it — so it is a real
review driven by the project, not a narration of one. It is a clean
clathrin-cargo adaptor (a different child from GRB2's signaling adaptor),
a Mendelian disease gene, and compact enough (90 GOA annotations) to review in
full. See the full review of LDLRAP1 (linked from the gene symbol) and its notes
at genes/human/LDLRAP1/LDLRAP1-notes.md.

The gene in one line

LDLRAP1 (ARH) is the cargo-selective CLASP (clathrin-associated sorting
protein) that couples the LDL receptor to the clathrin endocytic machinery: its
PID/PTB domain reads the FxNPxY internalization motif in the LDLR tail, while
separate C-terminal motifs bind clathrin and the β2-adaptin subunit of
AP-2, and it binds PtdIns(4,5)P2. Loss of function causes autosomal recessive
hypercholesterolemia [PMID:12221107].

Starting state (GOA, 90 rows)

What the decision rules produced

Action n Representative calls
ACCEPT 48 the adaptor MFs and endocytic-contact bindings; endocytosis / LDL-clearance / cholesterol-homeostasis processes; plasma-membrane, coated-vesicle and endosome locations
REMOVE 17 bare protein binding rows from interactome screens (protein-binding policy)
KEEP_AS_NON_CORE 10 AP-1 binding, broad cholesterol-transport/metabolism, APP-related and regulatory-framing rows
MODIFY 9 see below
MARK_AS_OVER_ANNOTATED 6 neurofilament/axon localisation; vague "regulation of protein binding"; cytokine-response and VSMC-proliferation rows that make LDLRAP1 the target of regulation, not a participant

The nine MODIFY calls are the instructive part, and they exercise two
different senses of the action:

  1. Partner → specific binding (refinement of protein binding). Six
    GO:0005515 rows whose interactor is AP2B1 / β2-adaptin (P63010) were
    recast to GO:0035612 AP-2 adaptor complex binding — the same
    "IntAct rows are a partner list" move as in GRB2, here consolidating six
    screen hits onto the informative term.
  2. Lateral correction of a mis-typed adaptor. GO:0035591 signaling adaptor
    activity and GO:0030159 signaling receptor complex adaptor activity were
    changed to GO:0035615 clathrin-cargo adaptor activity. LDLR endocytosis is
    not a signaling event; the same experiments (PMID:12221107,
    PMID:15166224) show LDLRAP1 coupling LDLR to the endocytic machinery, so the
    signaling-branch terms are the wrong child and the clathrin-cargo child is
    right. This is MODIFY as a lateral/branch correction, not a generalise/
    specialise refinement.
  3. One location refinement: GO:0005737 cytoplasm → GO:0005829 cytosol (the
    gene's own IDA term).

Core functions settled on two: GO:0035615 clathrin-cargo adaptor activity
(the integrative function — clathrin + AP-2 + PIP2 contacts) and GO:0050750
LDL particle receptor binding (the PTB domain reading the LDLR NPxY motif for
cargo selection).

Term-precision note. An earlier draft used GO:0001784 phosphotyrosine
residue binding for the cargo-recognition core function. That is wrong for
LDLRAP1: its ARH/Dab2-class PID/PTB domain binds the unphosphorylated FxNPxY
motif (PMID:12451172, "PTB domains bind to the consensus sequence FXNPXY") —
"PTB" is a fold name inherited from the phospho-dependent Shc/IRS class, not a
claim of phosphotyrosine binding here. core_functions ids are author-supplied
and hard-validated, so the core function now uses GO:0050750 (already present
as an IPI row); the GOA GO:0001784 IDA row is kept non-core by deference to the
curator. Contrast GRB2, where GO:0001784 is genuinely phospho-dependent (its
SH2 domain) — the same term is right there and wrong here.

What this example adds beyond GRB2