CDH23 (Q9H251) — Catenin Binding, Neuronal Projection, and Calcium Process Scope

Focus type: function_assignment
Hypothesis slug: catenin-neuronal-and-calcium-process-scope
Gene: CDH23 / Cadherin-related 23 (human, Homo sapiens, NCBITaxon:9606)
UniProt: Q9H251

Summary

This report evaluates a three-part function-assignment hypothesis for human
CDH23: that the gene product (i) binds beta-catenin or belongs to a catenin
complex, (ii) participates in neuron projection development, or (iii) contributes
to calcium transport/regulation. The seed asked that each claim be assessed
independently using the phylogenetic placement of nodes PTN008601603 and
PTN004649584 plus primary full texts, and that colocalization be distinguished
from binding and mechanical ion participation from ion-pump activity. The
investigation combined UniProt/QuickGO annotation provenance, PANTHER/Pfam domain
architecture, and targeted full-text mining of the primary literature.

Overall verdict: weakly supported / over-annotated. The catenin claim is
effectively refuted as a direct function: beta-catenin binding (GO:0008013) and
catenin complex (GO:0016342) are both IBA (phylogenetic) inferences on
Q9H251, propagated through ancestral node PTN008601603, whose experimental
seeds are exclusively classical catenin-binding cadherins (CDH1/E-cadherin,
CDH2/N-cadherin, CDH5/VE-cadherin, CDH24, CDH26) — never CDH23. CDH23 is an
atypical cadherin lacking the classical cytoplasmic catenin-binding domain
(Pfam PF01049); its tail instead anchors to the actin cytoskeleton through
harmonin PDZ interactions, described in primary work as "a novel anchorage
mode of the cadherins to the actin cytoskeleton" (PMID:12485990). The seed's
proposed rescue via beta-catenin colocalization fails: PMID:30747484 contains
zero mentions of catenin, and in PMID:22413011 beta-catenin is only an
adherens-junction marker, not a demonstrated CDH23 partner.

The neuronal claim is weakly supported at best (IBA-only via node PTN004649584;
neural expression present but no loss-of-function projection phenotype in
zebrafish, PMID:22977299). The calcium claim is partially supported but
mis-scoped: extracellular EC-domain calcium ion binding (GO:0005509) is
defensible and mechanistically central, but calcium ion transport
(GO:0006816) and regulation of cytosolic calcium (GO:0051480) over-interpret a
digenic PMCA2-pump modifier study (PMID:17234811) in which CDH23 appears only
as a genetic modifier. Importantly, none of this challenges CDH23's genuine core
functions — calcium-dependent homophilic adhesion and tip-link
mechanotransduction for hearing and vision — which have independent positive
primary support and are not the subject of this hypothesis.

Key Findings

F001 — CDH23 anchors to F-actin via harmonin PDZ, not via catenins (argues against beta-catenin binding)

UniProt Q9H251 annotates beta-catenin binding (GO:0008013) and catenin complex
(GO:0016342), but both are IBA:GO_Central phylogenetic inferences with no
experimental (IDA/IPI) support on CDH23 itself. The primary literature
documents a fundamentally different, catenin-independent cytoplasmic anchoring
mechanism. The CDH23 cytoplasmic domain binds harmonin through PDZ- and
N-domain interactions (PMID:12407180: "Two PDZ domains in harmonin interact with
two complementary binding surfaces in the CDH23 cytoplasmic domain";
PMID:19297620), and harmonin-b anchors CDH23 to the stereocilia actin core in a
manner explicitly framed as "a novel anchorage mode of the cadherins to the
actin cytoskeleton" (PMID:12485990). A later structural study confirms the
interaction "involves the harmonin Nter-PDZ1 supramodule, but not the C-terminal
PDZ-binding motif of cadherin-23" (PMID:20639393). CDH23 is classified in PANTHER
PTHR24026 "FAT atypical cadherin-related", not the classical cadherin clade
that carries the ancestral catenin-binding domain. Together these show CDH23 uses
a harmonin-based, non-catenin actin linkage — undercutting any claim of direct
beta-catenin binding or catenin-complex membership.

F002 — CDH23 "calcium ion transport" derives from a digenic PMCA2 modifier study, not from pump/transporter activity

On Q9H251, calcium ion transport (GO:0006816) and regulation of cytosolic
calcium ion concentration (GO:0051480) are annotated IMP:DFLAT; calcium ion
binding (GO:0005509) is IEA:InterPro. The experimental GOA reference for the
calcium processes, PMID:17234811 (Ficarella 2007), is fundamentally a study
of the PMCA2 (ATP2B2) plasma-membrane calcium pump. CDH23 appears in that
work only as a genetic modifier — a CDH23 T1999S substitution that
"accentuated hearing loss in a previously described human family with a PMCA2
mutation." The paper itself identifies the true Ca²⁺ machinery: "Ca²⁺ enters the
stereocilia of hair cells through mechanoelectrical transduction channels opened
by the deflection of the hair bundle and is exported back to endolymph by an
unusual splicing isoform (w/a) of plasma-membrane calcium-pump isoform 2
(PMCA2)." Neither the entry (MET channel) nor the export (PMCA2) route is CDH23.
CDH23 forms the upper tip-link that mechanically gates the MET channel. Thus
a transport or regulation-of-cytosolic-calcium annotation over-reads CDH23's
mechanical, adhesive role. Extracellular calcium ion binding by the EC
repeats, by contrast, is a genuine and central property of the protein — the seed
correctly notes that "a tip-link mechanical contribution can participate in ion
transport without being an ion pump," but that participation is captured by
mechanotransduction and calcium binding, not by a transport annotation.

F003 — "Neuron projection development" is an IBA inference; neural expression exists but loss-of-function shows no projection defect

Neuron projection development (GO:0031175) on Q9H251 is IBA:GO_Central only,
with no experimental support on CDH23. Critically, CDH23's characterized
"projections" are stereocilia — actin-based apical microvillar protrusions of
epithelial hair cells — not neuron projections; the hair-bundle role is
already captured by more accurate terms (sensory perception of sound GO:0007605
IMP:MGI; stereocilium GO:0032420 ISS). In zebrafish, cdh23 is expressed in CNS
nuclei and "a small subset of GABAergic amacrine cells," yet cdh23-mutant
larvae show "no detectable morphological retinal defects or degeneration" and a
normal optokinetic response (PMID:22977299). This is a textbook case of
expression without a demonstrated loss-of-function phenotype: neural
expression alone does not justify a neuron-projection-development function.

F004 — Catenin IBA is seeded only by classical cadherins at node PTN008601603 — paralog over-propagation

QuickGO annotation records for Q9H251 show GO:0008013 (beta-catenin binding) and
GO:0016342 (catenin complex) are both IBA (GO_REF:0000033, GO_Central) whose
with/from experimental seeds are exclusively classical type-I/II cadherins:
CDH1/E-cadherin (P12830, MGI:88354), CDH2/N-cadherin (P19022), CDH5/VE-cadherin
(P33151), CDH24 (Q86UP0), CDH26 (Q8IXH8), plus zebrafish cdh genes —
propagated through ancestral node PTN008601603, exactly the node named in the
seed hypothesis. None of the seed genes is CDH23, and no CDH23-specific
experimental catenin annotation exists. Classical cadherins bind beta-catenin
through a conserved cytoplasmic catenin-binding domain (Pfam PF01049) that is
absent from CDH23 (Q9H251 lists only PF00028 cadherin repeats plus the
CDH23-specific domain; its tail binds harmonin PDZ instead). The neuron
projection development IBA (GO:0031175) propagates from a separate node
PTN004649584, seeded by fly/mouse/zebrafish cadherins (Drosophila fat
FBgn0262018, MGI:1890219, ZDB-GENE-040513-7), again none of them CDH23. This is
the mechanistic signature of paralog over-annotation: an ancestral function
correctly inferred for the classical-cadherin subfamily is carried over onto an
atypical paralog that has diverged away from the seeding function. This finding
directly addresses the seed's request to assess PTN008601603 and PTN004649584
placement.

F005 — The seed's "beta-catenin colocalization" evidence does not demonstrate binding

Full-text mining (Europe PMC, open access) directly tests the seed's proposed
supporting evidence. PMID:30747484 (Sannigrahi 2019, PMC6487693) contains
zero occurrences of "catenin" in abstract or full text — the seed's
attribution of beta-catenin colocalization to this paper is unsupported.
PMID:22413011 (Apostolopoulou & Ligon 2012, PMC3296689) mentions β-catenin
12 times, but exclusively as an adherens-junction marker: the authors "used
an antibody to β-catenin to identify sites of possible cell-cell adhesion,"
scored contacts for β-catenin recruitment (found at 82.1% of heterotypic
contacts), and then scored those β-catenin⁺ sites for cadherin-23 labeling.
This establishes that CDH23 spatially colocalizes with β-catenin-marked
adhesion sites; it does not provide a co-IP, pulldown, or two-hybrid
demonstration of a CDH23–β-catenin molecular interaction. The paper's actually
demonstrated CDH23 function is heterotypic cell-cell adhesion between breast
cancer epithelial cells and fibroblasts — a genuine adhesion role, not a catenin
binding role. This resolves the seed's request to "compare PMID:12485990 with
beta-catenin colocalization in PMID:30747484 and PMID:22413011": one paper does
not mention catenin, and the other uses it only as a marker.

Mechanistic Model / Interpretation

The three sub-claims collapse onto a single coherent picture: CDH23 is an
atypical, tip-link cadherin whose cytoplasmic and calcium biology diverges from
the classical cadherin–catenin paradigm that seeds its questionable
annotations.

   CLASSICAL CADHERINS (E-/N-/VE-cadherin)          CDH23 (atypical, tip-link)
   ─────────────────────────────────────           ──────────────────────────
   EC repeats  (Ca2+-dependent adhesion)            27 EC repeats (Ca2+ binding,
│                                                homophilic + heterophilic
│  transmembrane                                 handshake w/ PCDH15)
▼                                                     │  transmembrane
   Cytoplasmic tail WITH catenin-binding                     ▼
   domain (Pfam PF01049)                             Cytoplasmic tail, NO PF01049
│                                                     │
▼                                                     ▼
   β-catenin ── α-catenin ── ACTIN                    HARMONIN (PDZ + N-domain)
   (catenin complex, GO:0016342)                             │
                                             ▼
                                       ACTIN core of stereocilium
                                    "novel anchorage mode" (12485990)

   IBA node PTN008601603  ──(β-catenin binding, catenin complex)──►  carried onto CDH23
seeds: CDH1, CDH2, CDH5, CDH24, CDH26  (NO CDH23)             = PARALOG OVER-PROP.

Catenin axis. The catenin-binding/complex annotations are pure phylogenetic
carry-over. The seeding node contains only classical cadherins that possess the
PF01049 catenin-binding domain; CDH23 lacks that domain and uses a
harmonin-based, structurally characterized alternative. The seed hypothesis
asked to "distinguish absence of classical cytoplasmic binding motifs from
negative binding assays." The honest answer is that both the motif is absent
and the positive cytoplasmic partner is a different protein (harmonin) — so
there is no residual positive evidence for beta-catenin binding to rescue.

Calcium axis. CDH23 is genuinely a calcium-binding protein: its EC
repeats require Ca²⁺ for rigidity and adhesive function, and Ca²⁺ occupancy at
inter-repeat linkers governs tip-link mechanics. That is a molecular-function
(binding) and mechanotransduction role — not ion transport. The "calcium ion
transport / regulation of cytosolic calcium" terms conflate the mechanical
gating of a downstream channel (and a digenic interaction with a real pump,
PMCA2) with CDH23 itself being a transporter. The correct causal chain is:

sound → hair-bundle deflection → CDH23–PCDH15 tip link tension →
MET channel opens → Ca2+ (and K+) influx → PMCA2 exports Ca2+
                   ▲                      ▲
             (not CDH23)             (not CDH23)

CDH23's node in this chain is the mechanical linker/gate, which is why
"sensory perception of sound" (GO:0007605) is the well-supported BP term.

Neuronal axis. CDH23's "projections" are epithelial stereocilia. Neural
expression is real but phenotypically silent in the available loss-of-function
model. The neuron-projection-development term is a second, independent instance
of IBA carry-over (from node PTN004649584) and should be treated as
non-core/unsupported for the human gene.

Evidence Base / Evidence Matrix

GO Curation Implications

These are leads requiring curator verification, organized by the three
sub-claims.

The recommended replacement for the catenin/neuronal/transport annotations is
not "protein binding" (uninformative). The informative, supported alternatives
already exist: calcium ion binding (MF), stereocilium / tip-link (CC),
sensory perception of sound (BP), and calcium-dependent cell-cell
adhesion (BP). The cytoplasmic interaction that a curator could positively
annotate — if seeking a real partner — is harmonin/USH1C binding (supported
by PMID:12407180, PMID:19297620, PMID:20639393), not beta-catenin.

Mechanistic Scope

The immediate molecular functions being tested are: (a) a protein–protein
binding activity toward beta-catenin, (b) a developmental process (neuron
projection development), and (c) an ion-transport/regulation activity.

• Direct gene-product activity that is supported: extracellular
  calcium-dependent homophilic and heterophilic adhesion (EC-repeat mediated),
  calcium ion binding, and mechanical tip-link formation with PCDH15 that gates
  the MET channel (PMID:23467356). Cytoplasmic tail binding to harmonin/USH1C and
  CAMSAP3 (PMID:12407180, PMID:19297620, PMID:20639393, PMID:27349180).
• Downstream phenotypes / consequences (not direct MF of CDH23): Ca²⁺ influx
  and cytosolic Ca²⁺ changes are consequences of channel gating, not CDH23
  transport. Hearing/vision loss is a disease manifestation of adhesion/tip-link
  failure. The digenic hearing-loss modification with PMCA2 (PMID:17234811) is a
  genetic-interaction effect, not evidence of CDH23 pump activity.
• Effects inferred only from loss of function or expression: the neuronal
  claim rests on expression plus phylogeny; the single available LoF model shows
  no neuronal defect (PMID:22977299).

Conflicts and Alternatives

1. Paralog confusion (primary alternative explanation). The catenin and
   neuronal terms are best explained not as CDH23 biology but as IBA
   over-propagation from classical cadherins through nodes PTN008601603 and
   PTN004649584. This is corroborated by domain architecture: CDH23 lacks
   PF01049. This is the dominant, well-supported alternative to the seed
   hypothesis.

2. Colocalization ≠ interaction. The seed treats spatial colocalization with
   β-catenin-marked junctions (PMID:22413011) as evidence of catenin membership.
   The full text shows β-catenin is only a marker; no binding assay exists, and
   the other cited paper (PMID:30747484) does not mention catenin at all.

3. Digenic modifier ≠ transporter. The calcium-transport annotation stems
   from a study whose subject is the PMCA2 pump; CDH23's contribution is genetic
   modification of a pump phenotype (PMID:17234811). An alternative, defensible
   framing is EC-repeat calcium binding, which the transport term wrongly
   promotes to transport activity.

4. Isoform/tissue caveat (partially addresses the seed's request). The seed
   asks whether alternative isoforms/tissues retain ancestral functions. CDH23
   has multiple isoforms (e.g., CDH23-C binds CAMSAP3/Marshalin, PMID:27349180;
   excretory isoforms modulate adhesion, PMID:30747484) and broad tissue
   expression (kidney, muscle, testis, heart). However, none of these isoform- or
   tissue-specific data document beta-catenin binding, neuron projection
   development, or calcium transport; they document more non-catenin cytoplasmic
   partnerships and adhesion biology. So the isoform argument, while legitimate to
   raise, does not rescue the seed claims with current evidence.

5. Stereocilia vs neuron projections. A genuine terminological conflict:
   CDH23's protrusions are epithelial stereocilia, not neurites. The
   neuron-projection term likely reflects this category error compounded by IBA.

Limitations and Knowledge Gaps

• No exhaustive interaction screen consulted. We relied on curated UniProt/
  QuickGO annotations, targeted full-text mining, and the provided literature. A
  systematic published negative binding assay (e.g., CDH23 tail vs β-catenin
  co-IP) is not in the record; the argument rests on absence of the motif +
  presence of an alternative partner, which is strong but not equivalent to a
  published negative assay. Resolution: a direct co-IP / ITC of the CDH23
  cytoplasmic domain against β-catenin.

• Isoform coverage is incomplete. CDH23 has many splice isoforms; we
  documented harmonin- and CAMSAP3-binding cytoplasmic behaviors but did not
  survey every isoform's tail for a cryptic catenin-binding capacity.
  Resolution: isoform-resolved domain analysis and pulldowns.

• Neuronal function relies on a single zebrafish LoF model. PMID:22977299
  reports no retinal phenotype, but a mammalian CNS-specific CDH23 knockout with
  detailed neurite/synapse morphometry has not been examined here. Absence of a
  reported phenotype is not proof of no function. Resolution: conditional
  neuronal Cdh23 knockout with axon/dendrite morphometry.

• Calcium binding stoichiometry not quantified here. We accept GO:0005509 on
  InterPro/structural grounds but did not compute EC-linker Ca²⁺ occupancy from
  structure. Resolution: structural/biophysical Ca²⁺-binding measurement (not
  needed to reject transport, which is already refuted mechanistically).

• QuickGO records were read at a point in time. GO_Central annotations can be
  updated; a curator should re-verify current IBA seeds for GO:0008013,
  GO:0016342, and GO:0031175 on Q9H251.

Discriminating Tests

1. Direct CDH23 tail vs β-catenin binding assay (co-IP, GST-pulldown, or ITC
   using the CDH23 cytoplasmic domain). A clean negative would convert
   "no-motif + alternative-partner" into a published negative binding result and
   fully close the catenin question.

2. Domain-swap / motif search across CDH23 isoforms for any PF01049-like
   catenin-binding module. Positive would reopen the isoform-rescue argument;
   negative confirms domain loss.

3. Conditional neuronal Cdh23 knockout (mouse) with quantitative
   neurite/axon-guidance morphometry and synapse counts, to test neuron
   projection development directly rather than by expression inference.

4. Ca²⁺-flux electrophysiology in hair cells separating CDH23's mechanical
   gating role from any putative transport: CDH23 mutation should alter MET-channel
   gating (tension) without conferring intrinsic ion-transport activity.

5. Re-audit of GO_Central IBA seeds for the three terms on Q9H251, confirming
   nodes PTN008601603 / PTN004649584 and the absence of a CDH23 experimental seed.

Proposed Follow-up Experiments / Actions (Curation Leads)

Candidate action changes (leads requiring curator verification):

• Do not accept beta-catenin binding (GO:0008013) and catenin complex
  (GO:0016342) for CDH23. Provenance = IBA over-propagation from classical
  cadherins (node PTN008601603); domain PF01049 is absent; the real cytoplasmic
  partner is harmonin.
• Do not accept / demote neuron projection development (GO:0031175) to
  non-core. Provenance = IBA (node PTN004649584); LoF model shows no neuronal
  phenotype (PMID:22977299).
• Remove or generalize calcium ion transport (GO:0006816) and regulation of
  cytosolic calcium (GO:0051480). These over-read a digenic PMCA2-pump modifier
  study (PMID:17234811).
• Retain calcium ion binding (GO:0005509), stereocilium (GO:0032420),
  sensory perception of sound (GO:0007605), and calcium-dependent cell-cell
  adhesion as the informative, supported terms.
• Consider adding a positively supported cytoplasmic MF: harmonin/USH1C
  binding (PMID:12407180, PMID:19297620, PMID:20639393) in place of the
  unsupported catenin term.

Candidate references with exact snippets to verify:

• PMID:12485990 — "thereby identifying a novel anchorage mode of the cadherins to
  the actin cytoskeleton."
• PMID:12407180 — "Two PDZ domains in harmonin interact with two complementary
  binding surfaces in the CDH23 cytoplasmic domain."
• PMID:17234811 — "which accentuated hearing loss in a previously described human
  family with a PMCA2 mutation"; "Ca²⁺ enters the stereocilia ... through
  mechanoelectrical transduction channels ... and is exported back to endolymph
  by an unusual splicing isoform (w/a) of ... PMCA2."
• PMID:22977299 — "revealed no detectable morphological retinal defects or
  degeneration"; "cdh23 is expressed by a small subset of GABAergic amacrine
  cells."
• PMID:22413011 — "cadherin-23 mediates heterotypic cell-cell adhesion between
  breast cancer epithelial cells and fibroblasts" (β-catenin used as junction
  marker only).

Suggested curator questions:

• Is there any CDH23-specific experimental (IDA/IPI) annotation for GO:0008013 or
  GO:0016342? (Expected: no.)
• Do current GO_Central seeds at PTN008601603 include CDH23? (Expected: no.)
• Should the review explicitly flag stereocilia vs neuron projections to prevent
  recurring category error?

Bottom Line

For human CDH23 (Q9H251), the seed hypothesis's three claims are weakly
supported to refuted as direct gene-product functions. The catenin binding /
complex terms are paralog IBA carry-over from classical cadherins and are
contradicted by CDH23's domain architecture (no PF01049) and its harmonin-based
actin anchorage; the neuron-projection term is IBA-only with a null LoF
phenotype; and the calcium-transport/regulation terms over-interpret a digenic
PMCA2 modifier study. Only calcium ion binding survives as a defensible
molecular-function annotation, alongside CDH23's genuine, independently supported
core roles in calcium-dependent homophilic adhesion and tip-link
mechanotransduction for hearing and vision.

Artifacts

• OpenScientist final report
• OpenScientist final report