ALPG, long known as ALPPL2, encodes germ cell alkaline phosphatase (GCAP, the Nagao isozyme, also called placental-like alkaline phosphatase). It is the third member of the tissue-specific alkaline phosphatase cluster on chromosome 2q and is the closest paralogue of placental alkaline phosphatase, sharing about 98% sequence identity with it and differing at only seven residues. One of those residues does most of the work of telling them apart: Gly429 in GCAP, where PLAP has Glu429, makes the germ-cell enzyme strongly inhibited by L-leucine, EDTA and heat while the placental enzyme resists all three, and these classical physicochemical criteria are what historically defined the Nagao isozyme. Like the rest of the family it is a homodimeric, Zn/Zn/Mg-dependent phosphomonoesterase attached to the outer leaflet of the plasma membrane by a C-terminal glycosylphosphatidylinositol anchor, hydrolysing phosphate monoesters at alkaline pH; no substrate specific to this isozyme has been established, so its catalytic description rests on the family chemistry rather than on any dedicated pathway. Normal expression is restricted to trace amounts in testis and thymus and to germ cells. Its clinical prominence comes from ectopic re-expression: GCAP is the predominant alkaline phosphatase transcript in germ cell neoplasia in situ and in seminoma, and choriocarcinoma lines that make a placenta-like enzyme in fact derepress this gene rather than the placental one. That near-identity with PLAP is also the field's central methodological hazard, since many antibodies and activity assays cannot separate the two.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0004035 alkaline phosphatase activity | IBA GO_REF:0000033 | ACCEPT | Summary: The core molecular function, placed by PAINT at a deep node of the alkaline phosphatase family whose donors span E. coli PhoA, yeast, fly, fish and all four human isozymes, ALPG itself among them. Directly confirmed for the human protein by activity measurements on GCAP, GCAP point mutants and GCAP/PLAP chimeras expressed in COS-1 cells (PMID:1939159). Reason: Correct and at the right level of specificity. The appearance of ALPG's own accession (P10696) in the WITH/FROM list is expected rather than circular: the gene has its own experimental annotation for this term, so it is one of the descendant evidences the PAINT curator used in placing the ancestral assertion, and the IBA then adds the further claim that the activity is inherited rather than lineage-specific. Supporting Evidence: PMID:1939159 To determine the structural basis and the amino acid(s) involved in these physicochemical differences, we constructed three GCAP mutants by site-directed mutagenesis and six GCAP/PLAP chimeras and then expressed these alkaline phosphatase mutants in COS-1 cells. |
| GO:0005886 plasma membrane | IBA GO_REF:0000033 | ACCEPT | Summary: GCAP is GPI-anchored on the outer leaflet of the plasma membrane with its catalytic domain facing outward, so the plasma membrane is where it acts and is_active_in is the correct qualifier. Direct evidence exists for the human protein: the Nagao-type enzyme made by the BeWo choriocarcinoma line is released from the membrane by phosphatidylinositol-specific phospholipase C and carries a PI-glycan moiety at its C-terminus (PMID:2162249). Reason: Correct. A more precise term exists (GO:0009897, external side of plasma membrane) and is proposed below, but the IBA as written is right and its node placement -- the vertebrate GPI-anchored membrane form -- is appropriate. Supporting Evidence: PMID:2162249 Like placental AP, the tumor AP can be released from the cell membrane by a phosphaditylinositol-specific phospholipase C and has a phosphaditylinositol-glycan (PI-glycan) moiety at the COOH terminus. |
| GO:0004035 alkaline phosphatase activity | IEA GO_REF:0000003 | ACCEPT | Summary: Automatic mapping from EC 3.1.3.1, duplicating the IBA, IDA and NAS rows. Reason: The EC assignment for ALPG is experimentally grounded rather than predicted, so the mapping is sound. |
| GO:0005886 plasma membrane | IEA GO_REF:0000044 | ACCEPT | Summary: Automatic transfer from the UniProt subcellular location vocabulary (SL-0039, cell membrane). Reason: Correct, and better supported than the evidence code implies given the PI-PLC release experiment on the human enzyme. |
| GO:0016791 phosphatase activity | IEA GO_REF:0000002 | MODIFY | Summary: InterPro2GO mapping from IPR001952 and IPR018299 to the generic parent term, discarding the alkaline pH optimum that the signature itself names. Reason: The InterPro entries behind this row are the alkaline phosphatase domain and its active site, which license GO:0004035 -- a descendant of GO:0016791 that is already directly supported for this protein. Annotating only to the parent adds nothing, and for a gene with as sparse a record as ALPG the dilution matters more than it would elsewhere. Proposed replacements: alkaline phosphatase activity |
| GO:0004035 alkaline phosphatase activity | IDA PMID:1939159 Mutation of a single amino acid converts germ cell alkaline ... | ACCEPT | Summary: Direct activity measurement on GCAP and on a panel of engineered variants expressed in COS-1 cells. The study is the definitive experimental separation of GCAP from PLAP: the two differ at only seven residues, and the classical differential sensitivity to L-leucine, EDTA and heat is shown to depend critically on position 429, where GCAP has glycine and PLAP glutamate. Reason: The single most important experimental annotation for this gene. It establishes not only that ALPG has alkaline phosphatase activity but that the enzyme is biochemically distinguishable from its 98%-identical paralogue, which is what licenses treating the two as separate functional entities and what makes isozyme attribution in this family tractable at all. It is also the source of the L-leucine, EDTA and heat sensitivity that UniProt records as ALPG's activity regulation. Supporting Evidence: PMID:1939159 We report that the differential reactivity of PLAP and GCAP depends critically on a single amino acid at position 429. GCAP with Gly-429 is strongly inhibited by L-leucine, EDTA, and heat, whereas PLAP with Glu-429 is resistant. PMID:1939159 Yet, they differ by only 7 amino acids at positions 15, 67, 68, 84, 241, 254, and 429 within their respective 484 residues. |
| GO:0005576 extracellular region | TAS Reactome:R-HSA-8940388 | KEEP AS NON CORE | Summary: From the Reactome reaction in which GPLD1 cleaves GPI anchors, releasing anchored proteins from the membrane into extracellular fluid. Reason: Mechanistically sound and clinically relevant, since it is soluble placental-type alkaline phosphatase in serum that is measured as a seminoma marker. But the released pool is downstream of where the enzyme acts, and for ALPG it carries an extra caveat: serum assays for "PLAP" largely cannot distinguish the GCAP contribution from the ALPP one, so this compartment is exactly where the isozyme confusion in this family is worst. |
| GO:0005886 plasma membrane | TAS Reactome:R-HSA-8940388 | ACCEPT | Summary: Reactome's statement of the substrate side of the same reaction: ALPG is a GPI-anchored plasma membrane protein before GPLD1 acts on it. Reason: Correct and consistent with the direct PI-PLC evidence on the human enzyme. |
| GO:0004035 alkaline phosphatase activity | NAS PMID:2162249 Expression of a Nagao-type, phosphatidylinositol-glycan anch... | ACCEPT | Summary: Cited as a non-traceable author statement, but the paper contains more than an assertion. Lowe and Strauss cloned the alkaline phosphatase cDNA from the BeWo choriocarcinoma line, showed by sequence that it is the germ cell (Nagao) gene product rather than the term placental one, and demonstrated experimentally that the enzyme is released by phosphatidylinositol-specific phospholipase C with at least 95% carrying a PI-glycan anchor. Reason: The term is correct and the paper is the gene-defining study for ALPG's protein product. The NAS code understates it: this is where the identity of the tumour-associated "placenta-like" enzyme was resolved to this gene, and where its GPI anchoring was directly shown. That second result is used below as the evidence for a proposed external-side-of-membrane annotation, which is what the paper actually demonstrates. Supporting Evidence: PMID:2162249 The complementary DNA is the product of the germ cell AP (Nagao isozyme) gene and not of the term PLAP gene. PMID:2162249 These data demonstrate that BeWo AP is the product of a gene normally expressed in testis, thymus, and germ cells, but not in placenta. |
| GO:0009897 external side of plasma membrane | IDA PMID:2162249 Expression of a Nagao-type, phosphatidylinositol-glycan anch... | NEW | Summary: Proposed new annotation. GOA records that ALPG is in the plasma membrane but not which face of it, even though ecto-orientation is what makes the enzyme an ectophosphatase. Lowe and Strauss showed directly, for the human Nagao isozyme, that the enzyme is released from the cell membrane by phosphatidylinositol-specific phospholipase C and that at least 95% of it carries a PI-glycan moiety at the C-terminus. Reason: A GPI anchor places the entire mature protein on the exoplasmic leaflet, which is what allows an alkaline phosphatase to act on extracellular substrates and to be shed into body fluids by phospholipase cleavage -- the basis of its use as a serum tumour marker. GO:0009897 is a descendant of GO:0005886 and so consistent with every existing location row; it records what those rows leave unstated. The evidence here is genuinely direct on the human protein, unlike the sequence-predicted GPI signal alone, so IDA is warranted. The same annotation is proposed for the other three human isozymes, since the topology is common to the family. Supporting Evidence: PMID:2162249 Immunoprecipitation of phosphaditylinositol-specific phospholipase C-treated AP and analysis by polyacrylamide gel electrophoresis or isoelectric focusing demonstrates that at least 95% of the AP contains PI-glycan. |
| GO:0008270 zinc ion binding | ISS GO_REF:0000024 | NEW | Summary: Proposed new annotation. GOA gives ALPG no metal-binding annotation at all, even though the binuclear zinc centre is the catalytic heart of every alkaline phosphatase and UniProt annotates seven zinc-coordinating residues in ALPG (61, 111, 334, 338, 375, 376, 451) by similarity to ALPP, whose metal sites are resolved in a 1.8 A crystal structure. Reason: This is the clearest gap in ALPG's record. Its two nearest paralogues both carry zinc and magnesium binding terms, and ALPG is 98% identical to ALPP and differs from it at only seven positions, none of them a metal ligand -- so if the transfer is safe anywhere in this family it is safe here. The zinc pair polarises the substrate phosphate and activates the serine nucleophile, so this is core to the catalytic function rather than incidental metal association. ISS is the correct code: the metals were resolved in ALPP, not in ALPG. |
| GO:0000287 magnesium ion binding | ISS GO_REF:0000024 | NEW | Summary: Proposed new annotation, the counterpart of the zinc row. The third catalytic metal of the alkaline phosphatase active site is magnesium; UniProt places its coordinating residues in ALPG at 61, 174 and 330 by similarity to ALPP. Reason: The Zn/Zn/Mg triad is a single catalytic unit and annotating part of it would leave the record incomplete. As with zinc, the inference is about as secure as a sequence-based inference can be, given 98% identity to a paralogue with a crystallographically resolved active site and no differences at the ligand positions. Both ALPI and ALPP already carry this term on the same kind of evidence, so its absence for ALPG looks like an oversight rather than a considered judgement. |
| GO:0005509 calcium ion binding | ISS GO_REF:0000024 | NEW | Summary: Proposed new annotation. Beyond the catalytic metals, mammalian alkaline phosphatases carry a structural calcium site with no counterpart in the bacterial enzyme; UniProt annotates four calcium-coordinating residues in ALPG (235, 288, 289, 304) by similarity to ALPL, where the identity of the metal was established by synchrotron X-ray fluorescence. Reason: Completes the metal complement and records the mammalian-specific part of it. Worth distinguishing from the catalytic metals in interpretation: as UniProt states for ALPL, calcium binding is structural and does not influence phosphatase activity, so this is a real but non-catalytic molecular function. It also has a specific resonance for ALPG, whose defining Gly429 makes the enzyme unusually sensitive to EDTA -- a chelation phenotype that only makes sense for a metalloenzyme with several metal sites. |
| GO:0042803 protein homodimerization activity | ISS GO_REF:0000024 | NEW | Summary: Proposed new annotation. UniProt records ALPG as a homodimer by similarity to ALPP, where the crystal structure resolves an intimate interface in which a long N-terminal helix from one monomer embraces the other and a residue from one subunit reaches into the partner's active site. Reason: Ordinarily a protein-binding term would not be worth adding, but across this family the dimer is a functional rather than an incidental unit: in ALPP the interface is credited with the mammalian-specific allostery, in ALPL monomeric mutants are catalytically dead and dominant-negative, and in ALPI disease alleles act partly by impairing dimerization. Recording it for ALPG keeps the family's quaternary requirement consistent. Two of the seven residues that differ between GCAP and PLAP lie in the N-terminal region that contributes to the interface, so the dimer is also where any real functional divergence between the two would most plausibly be found. |
Loading supporting content…
Download this section (compressed HTML)Q: What does germ cell alkaline phosphatase do in germ cells? No physiological substrate, pathway or loss-of-function phenotype has been described, so the gene is functionally dark below the level of its chemistry despite having been known for decades as a tumour marker.
Q: Why do humans maintain two alkaline phosphatase genes that differ at only seven residues? If Gly429 versus Glu429 changes only inhibitor and heat sensitivity, the duplication looks regulatory rather than catalytic -- but if that residue also shifts substrate preference, ALPG and ALPP may be doing genuinely different chemistry in their respective tissues.
Q: How much of the published "placental alkaline phosphatase" literature, particularly serum marker studies and immunohistochemistry, actually measures GCAP? Given roughly 98% identity and widespread antibody cross-reactivity, resolving this retrospectively would change how much of that evidence can be attributed to either gene.
Q: What drives the derepression of ALPG in germ cell neoplasia in situ and seminoma, and in choriocarcinoma lines where the placental gene is silenced instead? The switch between two adjacent, nearly identical genes is an unusually clean system in which to study tissue-specific and oncogenic promoter control.
Experiment: Profile purified recombinant GCAP, PLAP and the reciprocal position-429 variants against a broad phosphomonoester substrate panel under matched conditions, extending the chimera set already built in PMID:1939159. A difference in substrate preference would give the first functional rationale for maintaining both genes; its absence would support a purely regulatory explanation for the duplication.
Hypothesis: GCAP and PLAP differ in substrate preference, not only in inhibitor and heat sensitivity, and Gly429 is responsible.
Type: comparative enzyme kinetics on engineered variants
Experiment: Knock out or inhibit ALPG in a human germ cell tumour line that expresses it strongly, such as a seminoma-derived line, and compare the extracellular phosphoproteome and phosphometabolome against control, using the L-leucine sensitivity of GCAP to distinguish its contribution from any co-expressed isozyme. This is the same design proposed for ALPP and is the only route to a substrate that does not depend on transferring assumptions from ALPL.
Hypothesis: ALPG has a physiological substrate at the germ cell surface that can be found by perturbing the enzyme rather than by guessing candidates.
Type: comparative phosphoproteomics with genetic perturbation
Experiment: Test the commonly used anti-PLAP antibodies and activity assays against purified recombinant ALPP and ALPG side by side, quantifying cross-reactivity, and pair the results with transcript-level discrimination in the same tumour samples. This would establish how much of the seminoma marker literature is attributable to each gene and put the isozyme assignments in both reviews on a measured rather than an assumed footing.
Hypothesis: Existing immunoassays for placental alkaline phosphatase substantially conflate ALPP and ALPG.
Type: assay specificity validation
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)