AIGR Gene Hypothesis Deep Research — A1BG (P04217)

Focus: Membrane-Receptor and Growth-Hormone Capacities (function_assignment)

Gene: A1BG (Alpha-1B-glycoprotein), Homo sapiens (NCBITaxon:9606)
UniProt: P04217 · PANTHER family: PTHR11738 · InterPro: IPR016332 / IPR050412

Summary

The seed hypothesis proposes that human A1BG (P04217) may retain (i) plasma-membrane localization
(GO:0005886), (ii) transmembrane signaling receptor activity (GO:0004888), or (iii) participation in
growth-hormone receptor signaling (GO:0060396), and asks that each be assessed independently against the
actual GO definitions. Evaluated against structure, provenance, and primary literature, all three
capacities are refuted or unsupported, and each has a specific, identifiable reason for appearing in the
annotation set that is not "A1BG performs this function." The overall verdict is REFUTED /
OVER-ANNOTATED.

Three converging lines of evidence drive this judgment. First, structure closes the receptor and
membrane-pool routes: mature A1BG is a single-chain secreted plasma glycoprotein of five Ig-like V-type
domains with a cleaved signal peptide and no transmembrane helix, GPI anchor, or lipidation in any
isoform. A Kyte-Doolittle hydropathy scan of the mature chain (residues ≥22) peaks at 0.98, far below the
~1.6 transmembrane threshold; only the cleaved signal peptide (residues 1–19, 1.69) exceeds it. Second,
the growth-hormone link is a direction-of-causality error: the one experimental source (mouse A1bg,
PMID 16723264) shows that A1bg is a GH-induced liver transcript whose expression requires an intact
GH–GHR signaling complex — a readout of GH signaling acting on the gene, not the protein transducing GH
signal. Third, the provenance is IBA-only paralog over-annotation: all three questioned terms are
GO_Central IBA (ECO:0000318, GO_REF:0000033) with zero experimental MF/BP support, inherited from the LILR
and KIR transmembrane immune receptors that share A1BG's Ig-domain fold in PANTHER PTHR11738.

The gene product's genuine, experimentally supported biology is extracellular high-affinity protein
binding: it sequesters CRISP-3 with nanomolar affinity in a 1:1 complex (surface plasmon resonance) and
directly interacts with and stabilizes NAMPT. Every experimental cellular-component annotation places A1BG
in the extracellular/secreted compartment. The recommended curation lead is to remove or
do-not-propagate GO:0004888, GO:0005886, and GO:0060396, retain the extracellular CC terms, and consider
ligand-specific binding MF annotations. The important epistemic caveats from the seed — that a soluble
location does not exclude a membrane pool, and that absence of a human assay is not loss — are respected:
the refutation rests on positive structural evidence and a corrected reading of the GH paper, not on
missing assays.

Key Findings

F001 — A1BG is a secreted plasma protein with no transmembrane helix, refuting transmembrane signaling receptor activity (GO:0004888)

Human A1BG is 495 amino acids with a cleaved signal peptide (residues 1–21), a mature chain spanning
residues 22–495, and five Ig-like V-type domains. UniProt P04217 lists the subcellular location as
Secreted, annotates zero transmembrane features, and includes the keywords Secreted and Signal
but not Membrane or Receptor. To directly test whether any internal segment could act as a hidden
membrane anchor, a Kyte-Doolittle hydropathy scan (window = 19) was run across the full sequence. Only one
window exceeded the transmembrane threshold — residues 1–19, score 1.69 — corresponding exactly to the
cleaved signal peptide. Across the entire mature chain (residues ≥22), the maximum hydropathy score was
0.98, well below the ~1.6 threshold characteristic of a genuine membrane-spanning helix or
signal-anchor.

The original amino-acid sequence determination (PMID: 3458201)
established A1BG as "a single polypeptide chain N-linked to four glucosamine oligosaccharides," homologous
to the immunoglobulin supergene family. GO:0004888 (transmembrane signaling receptor activity) is
annotated on P04217 by IBA (Inferred from Biological Ancestor; GO_Central/PANTHER) only, with no
experimental (IDA/EXP) support. Because the GO definition of a transmembrane receptor requires the protein
to span the membrane and relay a signal across it, a protein lacking any membrane-integral segment cannot
satisfy the term. This is a definitive structural incompatibility, not merely a missing assay.

{{figure:a1bg_hydropathy.png|caption=Kyte-Doolittle hydropathy scan (window = 19) of the full A1BG sequence. The only region exceeding the ~1.6 transmembrane threshold is residues 1–19 (score 1.69), corresponding to the cleaved N-terminal signal peptide. The entire mature chain (residues ≥22) peaks at 0.98, ruling out an internal transmembrane helix or signal-anchor and closing the structural route to intrinsic transmembrane receptor activity.}}

F002 — The growth-hormone receptor signaling annotation (GO:0060396) derives from A1bg being a GH-induced transcriptional target, not a signaling effector

The sole experimental basis for any GH connection is Tiong et al., 2006 (Growth Horm IGF Res;
PMID: 16723264). Reading the full paper rather than only the
abstract confirms that the mouse A1bg ortholog (referred to as "mRNA #5" / "P5") is a GH-induced liver
transcript. Its induction is abolished in GH-deficient dwarf mice, in mice treated with a GH antagonist,
and in GHR/GHBP-knockout mice. The paper's own conclusion is that "induction of mRNA #5 in the liver
requires a continuous pattern of GH secretion and an intact GH-GH receptor-signaling complex."

Critically, this describes A1bg expression as a downstream output of GH signaling — the gene is
transcriptionally switched on when the GH→GHR→JAK/STAT axis is intact. It does not describe the A1BG
protein participating in, transducing, or being required for GH signal transduction. "Gene regulated by
GH" is not equivalent to "gene product participates in GH receptor signaling." GO:0060396 on human P04217
is IBA-only, with no experimental support. The 2025 female-specific cardiomyopathy knockout phenotype
([PMID: 40270023]) is a downstream loss-of-function physiological phenotype — dilated cardiomyopathy with
altered glucose-6-phosphate/acetyl-CoA metabolism and intercalated-disc disruption — and likewise does not
demonstrate GH signal transduction by the A1BG protein. If any BP term were desired, the most defensible
would be response to growth hormone (GO:0060416) in mouse, framed as a transcriptional response, not
signal transduction.

F003 — A1BG's experimentally supported location is extracellular/secreted and its supported molecular function is high-affinity protein binding (CRISP-3, NAMPT), not plasma membrane

Every experimental cellular-component annotation on P04217 places A1BG in the extracellular/secreted
compartment: extracellular region (IDA), extracellular exosome (HDA), blood microparticle (HDA), and
platelet alpha/secretory granule and ficolin-1-rich granule lumen (TAS). By contrast, GO:0005886 (plasma
membrane) is IBA-only with no experimental support.

On the molecular-function side, the experimentally supported activity is specific, high-affinity protein
binding. CRISP-3 is a specific, high-affinity ligand of A1BG with a dissociation constant in the nanomolar
range, forming a 1:1 noncovalent complex demonstrated by surface plasmon resonance
(PMID: 15461460); this binding is conserved across mammalian
species — A1BG binds CRISPs in cow, horse, and rabbit sera
(PMID: 20116414). More recently, secreted A1BG was shown to
directly interact with and stabilize NAMPT — leading to NAD⁺ production and PARP1-dependent DNA repair —
with recombinant A1BG active in the assay ([PMID: 40560034]). Both documented functions are extracellular
protein-protein interactions. No experimental assay places A1BG at the plasma membrane or demonstrates
receptor signaling. These are separate, well-supported functions; per the seed's own logic they neither
refute nor rescue the questioned terms, and they define the informative MF that should replace bare
"protein binding."

F004 — The receptor/plasma-membrane IBA terms are paralog over-annotation from LILR/KIR membrane receptors in PANTHER PTHR11738

EBI QuickGO confirms that all three questioned terms on P04217 are GO_Central IBA (ECO:0000318,
GO_REF:0000033), that they are in fact the only MF/BP annotations A1BG carries, and that every
experimental term (IDA/HDA/TAS) is an extracellular/secreted CC — none is an MF or BP. The source of the
receptor/membrane inference is A1BG's family placement. InterPro classifies P04217 in family IPR016332
("Alpha-1B-glycoprotein/leukocyte immunoglobulin-like receptor") and in the homologous family IPR050412
("Immunoglobulin-like Receptors in Immune Regulation"). PANTHER family PTHR11738 is likewise
"Immunoglobulin-like Receptors in Immune Regulation," and the constituent CDD Ig domains (cd05751,
cd16843) are annotated as "found in Leukocyte Ig-like receptors (LILRs), Natural killer inhibitory
receptors (KIRs)." PANTHER assigns A1BG to the "immunoglobulin receptor superfamily" protein class, and
the PANTHER GO-slim CC for A1BG lists "plasma membrane."

In other words, A1BG shares an Ig-domain scaffold with a large set of bona fide transmembrane immune
receptors that carry cytoplasmic ITIM/ITAM tails and genuine plasma-membrane/receptor annotations. The
ancestral-inference machinery legitimately assigns membrane/receptor terms to the family ancestor
(PTN002621170) but then propagates them onto the A1BG leaf (PTN002482657), which is the divergent,
secreted, transmembrane-less family member. The A1BG-specific GH IBD (PTN000200788) is most plausibly
rooted in the GH-target paper — a direction-of-causality misinterpretation lifted to family level. This is
the textbook signature of paralog over-annotation.

F005 — No A1BG isoform carries a transmembrane, GPI, or lipid membrane anchor, closing the "membrane pool" structural route

The seed hypothesis correctly notes that a soluble primary location does not exclude a membrane pool. To
test whether any isoform could provide a membrane-anchoring route, the isoform records were examined.
UniProt P04217 lists two isoforms: P04217-1 (displayed) and P04217-2, which differs only at N-terminal
residues 1–122 — i.e., in the region overlapping the signal peptide, not at the C-terminus where a
membrane anchor would typically reside. No feature of type Transmembrane, Intramembrane, Lipidation, or
GPI/propeptide anchor is annotated on any isoform. Combined with the F001 hydropathy result (mature-chain
maximum Kyte-Doolittle = 0.98), there is no annotated structural route by which A1BG could become
membrane-integral. The residual, unfalsifiable possibility — transient peripheral, non-covalent
association with a membrane surface — would still not satisfy GO:0004888 (which requires transmembrane
spanning) and would require direct surface/biochemical assays to establish even as a CC.

Mechanistic Model / Interpretation

The three questioned capacities can be laid out against what A1BG actually is and does:

                 A1BG (P04217) — what the evidence shows
   ┌───────────────────────────────────────────────────────────────────────┐
   │  Signal peptide (1–21, cleaved)  ──►  SECRETED into plasma             │
   │  Mature chain (22–495): 5 Ig-like V-type domains                       │
   │  NO transmembrane helix | NO GPI anchor | NO lipidation (any isoform)  │
   │  Hydropathy(mature) max = 0.98  «  1.6 TM threshold                    │
   └───────────────────────────────────────────────────────────────────────┘
                           │
     ┌─────────────────────┼─────────────────────┐
     ▼                     ▼                     ▼
   EXPERIMENTAL CC          EXPERIMENTAL MF        QUESTIONED (IBA-only)
   extracellular region     CRISP-3 binding        GO:0004888 TM receptor  ✗
   extracellular exosome     (Kd nM, SPR, 1:1)     GO:0005886 plasma memb  ✗
   blood microparticle      NAMPT stabilization    GO:0060396 GH signaling ✗
   granule lumen            (direct interaction)

Why each questioned term fails:

The unifying interpretation is that A1BG is an Ig-domain scaffold that evolved into a soluble plasma
carrier/binding protein, retaining the fold of its LILR/KIR relatives while shedding their membrane
anchor and receptor signaling role. Its genuine, experimentally supported biology is extracellular
protein binding: sequestering CRISP-3 with nanomolar affinity and stabilizing NAMPT via direct
interaction. The receptor/membrane GO terms are a fossil of shared ancestry, propagated by ancestral
inference; the GH term is a misinterpretation of a gene that is regulated by GH rather than signaling
GH. Sex/organism specificity of both the GH-target behavior and the dilated-cardiomyopathy phenotype
(mouse, female-biased) further echoes GH's sexually dimorphic secretion pattern, reinforcing that A1bg
sits downstream of GH.

Evidence Matrix

GO Curation Implications (leads — require curator verification)

The evidence supports removing or not propagating all three questioned terms because each is IBA-only,
has zero experimental MF/BP support, and has an identifiable non-A1BG source. The experimentally grounded
CC terms (extracellular region/space/exosome/blood microparticle) should be retained, and the MF should
be upgraded from bare "protein binding" to ligand-specific binding supported by direct assay.

Mechanistic Scope

The immediate molecular function under test is whether the A1BG gene product directly (a) spans and
signals across a membrane, (b) resides in the plasma membrane, or (c) transduces GH signaling. The
analysis strictly separates direct gene-product activity from downstream consequences:

• Direct activity (supported): extracellular high-affinity binding of CRISP-3 (nM Kd, 1:1 SPR complex)
  and direct interaction with/stabilization of NAMPT.
• Direct activity (refuted): transmembrane receptor signaling — no membrane-integral segment exists to
  perform it.
• Downstream/regulatory (not gene-product function): GH-dependent transcriptional induction of A1bg is
  an upstream regulatory input onto the gene, not an activity of the protein; the female-specific KO
  cardiomyopathy (PMID 40270023) is a pleiotropic physiological outcome of A1BG loss.

The distinction is decisive: GO:0060396 and GO:0004888 both require the protein to do signaling work,
whereas the evidence only shows the gene being regulated by signaling and the protein binding
partners in the extracellular space.

Conflicts and Alternatives

1. Paralog over-annotation (primary alternative, and the supported one). A1BG shares its Ig-domain
   fold with LILR and KIR transmembrane immune receptors (InterPro IPR016332/IPR050412; PANTHER
   PTHR11738; CDD cd05751/cd16843). Ancestral inference legitimately assigns receptor/membrane terms to
   the family ancestor but incorrectly propagates them to A1BG, which lacks the membrane anchor. Most
   parsimonious account of why the terms appear.

2. Direction-of-causality (GH). A reading that treats GH-dependence as A1BG "participation" in GH
   signaling inverts the biology: GH-dependence of a transcript's abundance is regulation of the gene,
   the opposite of the protein participating in signal transduction.

3. Organism/sex specificity. GH-target behavior and the DCM phenotype are mouse and female-biased,
   echoing GH's sexually dimorphic secretion — again consistent with A1bg being downstream of GH.

4. Isoform-specific escape route (excluded). UniProt's two isoforms differ only at the N-terminus
   (signal-peptide region), not at a C-terminal anchor, and none carries TM/GPI/lipidation features.

5. Alternative MF. CRISP/venom-toxin sequestration (opossum A1BG-like proteins neutralize venom)
   points to a protease/toxin-inhibitor-like binding role, not receptor signaling. Consistent with the
   seed's note, this does not by itself refute another function; the refutation rests on independent
   structural and provenance evidence.

Limitations and Knowledge Gaps

• No direct human GH-signaling assay exists either way. We cannot experimentally exclude an unassayed
  human role, but the burden of an IBA-only term is not met and no mechanism is proposed. Resolve with: a
  human loss-of-function experiment testing GH→GHR→STAT5 output ± A1BG.

• The "membrane pool" cannot be excluded absolutely by structure alone. Peripheral/electrostatic
  association of a secreted protein with a membrane surface is conceivable and would not require a TM
  helix. Checked: isoform features, hydropathy, keyword/feature annotations — all negative for membrane
  integration. Matters because: even a peripheral association would not justify GO:0004888 but could
  touch on a "cell surface" CC. Resolve with: surface biotinylation / cell-surface proteomics or
  immuno-EM.

• NAMPT-stabilization localization (PMID 40560034) not fully dissected. The interaction is reported as
  extracellular/secreted, consistent with F003, but the precise subcellular geography of the A1BG–NAMPT
  complex merits primary-source confirmation. Resolve with: compartment-resolved interaction assays.

• PANTHER tree topology (PTN nodes) inferred from family membership, not traced node-by-node. The
  provenance conclusion is robust at the family level, but a curator may wish to inspect the exact PAINT
  annotation tree for PTN002482657 / PTN002621170 / PTN000200788.

• Methodological caveats. Hydropathy is heuristic (though unambiguous here); PMID 40560034/40270023
  read at abstract/figure-legend level; PANTHER node IDs taken from the seed context. Public-resource
  access (UniProt, QuickGO, InterPro, NCBI) succeeded; no fabricated results.

Discriminating Tests

The following would most efficiently distinguish the seed hypothesis from the paralog-over-annotation
interpretation:

1. Cell-surface proteomics / surface biotinylation on A1BG-expressing human cells — tests for any
   plasma-membrane pool (GO:0005886).
2. Reporter/phospho-signaling assay: does recombinant A1BG activate any receptor or JAK–STAT axis? —
   tests the TM-receptor MF (GO:0004888).
3. Separate "GH induces A1bg transcription" (RNA-seq ± GH) from "A1BG transduces GH signal" (STAT5
   activation in A1BG-null vs WT) — resolves the GH BP (GO:0060396).
4. Isoform TM/GPI prediction (Phobius / DeepTMHMM) across all P04217 isoforms — confirms the fully
   soluble architecture.
5. PAINT tree audit of PTHR11738 — documents node-by-node that the receptor/membrane/GH terms enter
   via ancestral LILR/KIR nodes.

Curation Leads (require curator verification)

Candidate action changes:

• Remove or mark NOT / do-not-propagate for GO:0004888 (MF) — IBA-only, refuted by absence of any TM
  segment.
• Remove or mark NOT / do-not-propagate for GO:0005886 (CC) — IBA-only, contradicted by experimental
  extracellular CC annotations.
• Remove or mark NOT / do-not-propagate for GO:0060396 (BP) — IBA-only, based on a
  transcriptional-target misinterpretation of PMID 16723264.
• Retain extracellular region / extracellular space / extracellular exosome / blood microparticle
  (CC; IDA/HDA/TAS) and consider ligand-specific binding MF (CRISP-3, NAMPT) over bare "protein
  binding."

Candidate references with exact snippets to verify:

• PMID: 3458201 — "consists of a single polypeptide chain
  N-linked to four glucosamine oligosaccharides" (secreted single-chain Ig-domain structure; no membrane
  region).
• PMID: 16723264 — "induction of mRNA #5 in the liver
  requires a continuous pattern of GH secretion and an intact GH-GH receptor-signaling complex" (A1bg is a
  GH-induced transcript, not a signal transducer).
• PMID: 15461460 — "CRISP-3 is a specific and high-affinity
  ligand of A1BG with a dissociation constant in the nanomolar range as evidenced by surface plasmon
  resonance" (true MF).
• [PMID: 40560034] — "revealed a direct interaction between A1BG and NAMPT, leading to the stabilization
  of NAMPT" (secreted protein-protein interaction, not membrane receptor / GH signaling).

Suggested curator questions:

• Is there ANY experimental (IDA/EXP) support for A1BG at the plasma membrane or as a signal transducer?
  (None found here.)
• Does any P04217 isoform encode a TM or GPI anchor? (None annotated.)
• Does the PAINT tree attach the receptor/membrane/GH terms at an ancestral LILR/KIR node rather than the
  A1BG leaf?

Suggested experiments: the discriminating tests above.

Evidence Base (literature summary)

Conclusion

The seed hypothesis's three candidate capacities for human A1BG — plasma-membrane localization,
transmembrane signaling receptor activity, and growth-hormone receptor signaling participation — are each
refuted or unsupported when assessed independently against the GO definitions and primary literature.
A1BG is a secreted, five-Ig-domain plasma glycoprotein with no membrane anchor in any isoform (closing the
receptor and membrane-pool routes), and the GH connection is a transcriptional-target relationship (mouse
A1bg is induced by GH signaling) rather than signal transduction by the A1BG protein. All three
questioned terms are IBA-only paralog over-annotations inherited from the LILR/KIR transmembrane immune
receptors that share A1BG's Ig-domain fold in PANTHER PTHR11738. The gene product's experimentally
supported biology is extracellular high-affinity protein binding (CRISP-3, NAMPT). Recommended curation
lead: remove or do-not-propagate GO:0004888, GO:0005886, and GO:0060396; retain the extracellular CC terms
and consider ligand-specific binding MF annotations.

Artifacts

• OpenScientist final report
• OpenScientist final report
• OpenScientist a1bg hydropathy