ADAM5 is a unitary pseudogene at human chromosome 8p11.22, in a genomic neighbourhood that also carries the testis-expressed ADAM (a disintegrin and metalloproteinase) genes ADAM2, ADAM9, ADAM18 and the ADAM3A pseudogene. Its orthologue, historically called tMDC II, is a testis-specific type-I transmembrane protein: it is displayed on the sperm surface in macaque and identified in sperm plasma membrane preparations in pig, and the gene is conserved in mouse, rat and guinea pig. Its function has been established in mouse, where ADAM5 assembles with ADAM2 and ADAM3 into a sperm-surface complex; knockout males are severely subfertile, their spermatozoa lose ADAM2 and ADAM3, fail to transit the uterotubal junction and bind the zona pellucida poorly. Like all its orthologues, the ancestral protein is a non-catalytic member of the family, lacking the zinc-binding residues of a working metalloprotease. The human locus is transcribed - multiple alternatively spliced testis transcripts have been cloned and the RefSeq is the non-coding NR_001448 - but every characterised human transcript carries deletions and in-frame termination codons relative to the macaque orthologue, and a macaque anti-tMDC II antiserum detects no product on Western blots of human testis or sperm extracts. The 412-residue UniProt sequence is a putative conceptual translation: aligned to the 756-residue macaque protein it shows a single 246-residue internal deletion that removes the signal peptide, essentially the whole metalloprotease domain and the transmembrane anchor, and truncates the disintegrin domain. No human ADAM5 protein has been detected, and no molecular activity, subcellular location or biological process has been established for this locus. The loss is not isolated: five other testis-expressed ADAM genes - ADAM1A, ADAM1B, ADAM3A, ADAM3B and ADAM6, dispersed across chromosomes 8, 12, 14 and 16 - are also pseudogenes in humans, so the sperm ADAM adhesion complex characterised in rodents is only partly reconstructible in humans.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0003674 molecular_function | ND GO_REF:0000015 | ACCEPT | Summary: Correct. No molecular function has been established for the human ADAM5 locus, and the record positively indicates that no protein product exists to carry one. Reason: The root ND annotation records that a curator looked and found no data, which is exactly the state of this locus. HGNC types ADAM5 as a pseudogene, its RefSeq is the non-coding NR_001448, UniProt assigns PE 5 (Uncertain) and cautions that it could be a pseudogene product, and all eleven EMBL cross-references on the entry are NOT_ANNOTATED_CDS. The primary paper reports that every human tMDC II transcript is non-functional through deletions and in-frame stop codons, and that a macaque anti-tMDC II antiserum gives no signal on human testis or sperm extract - a measured protein-level negative, not merely an absence of reports. Note that the family name is not evidence of an activity here, and the structural test agrees with the curator: the 412-residue putative translation contains no HExxHxxGxxHD zinc-binding motif and in fact no HExxH substring at all, and aligned to macaque ADAM5 only 6.6% of the Peptidase M12B domain has any human counterpart. Even the intact orthologues are non-catalytic; UniProt attaches "Not expected to have protease activity" to all five reviewed ADAM5 entries. So there is no candidate molecular function to substitute for the root term, and nothing here to modify or remove. Supporting Evidence: PMID:10417343 all of which are also non-functional owing to the presence of deletions and in-frame termination codons, when compared with the macaque orthologue, a finding which is further supported by the lack of immunoreactivity on Western blots of human testis and sperm extracts probed with a macaque anti-tMDC II polyclonal antiserum PMID:41263505 While ADAM5 is a pseudogene in humans, our results provide valuable insights into the function of ADAM family proteins in mammalian reproduction. file:human/ADAM5/ADAM5-uniprot.txt CAUTION: Could be the product of a pseudogene. Not expected to have file:human/ADAM5/ADAM5-bioinformatics/RESULTS.md Human ADAM5 contains **no `HExxH` substring anywhere in its 412 residues**. |
| GO:0005575 cellular_component | ND GO_REF:0000015 | ACCEPT | Summary: Correct, and independently supported: the putative human product has lost both the signal peptide and the transmembrane anchor, so it could not reach the sperm surface where its orthologues act even if it were translated. Reason: No subcellular location has been demonstrated for a human ADAM5 product, and none can be transferred from the orthologues. The macaque and mouse proteins are type-I transmembrane sperm-surface antigens, but alignment of the human putative translation to macaque ADAM5 gives 0.000 coverage of the macaque signal peptide (1-16) and 0.000 coverage of the transmembrane helix (699-719). A product with neither secretory targeting nor a membrane anchor is not a candidate for any membrane or cell-surface term. This is the cellular-component counterpart of the molecular-function argument and it is an additional reason to keep the root ND rather than import a location by similarity. Supporting Evidence: file:human/ADAM5/ADAM5-bioinformatics/RESULTS.md the **signal peptide** and the **transmembrane helix** are lost outright, and the disintegrin domain is itself truncated because the deletion cuts through it PMID:10645274 we have localized a number of these MDC proteins (fertilin alpha, fertilin beta, tMDC I, tMDC II and tMDC III) to spermatogenic cells |
| GO:0008150 biological_process | ND GO_REF:0000015 | ACCEPT | Summary: Correct. The sperm-fertilisation role established for ADAM5 is a property of the intact orthologues; no biological process has been demonstrated for the human locus. Reason: The temptation on this gene is to import the orthologue's biology. Adam5-null mice are severely subfertile, their sperm fail to transit the uterotubal junction and bind the zona pellucida poorly, and ADAM2 and ADAM3 levels fall on the mutant sperm - but the same paper states that ADAM5 is a pseudogene in humans, and the human protein has never been detected. Transferring a fertilisation process term to this locus on orthology alone would be exactly the propagation error this review exists to catch, and GOA has correctly not done it. A separate, RNA-level hypothesis exists - that the ADAM5 pseudogene transcript acts as a competing endogenous RNA for miR-122b-5p and thereby modulates ADAM9 - but it rests on a predicted binding site plus a copy-number association, not on a demonstrated activity, and it would concern the transcript rather than a protein. It is recorded under knowledge_gaps rather than annotated. Supporting Evidence: PMID:41263505 Adam5 KO sperm failed to transit the uterotubal junction (UTJ) and displayed severe defects in zona pellucida (ZP) binding, phenotypes that resemble those observed in Adam2 and Adam3 knockout mice. PMID:41263505 While ADAM5 is a pseudogene in humans, our results provide valuable insights into the function of ADAM family proteins in mammalian reproduction. |
Loading supporting content…
Download this section (compressed HTML)Q: Should GO annotate pseudogenes at all, and if so is a set of three root ND annotations the right representation? Six testis-expressed ADAM genes are HGNC-typed pseudogenes in human - ADAM1A and ADAM1B (12q24), ADAM3A and ADAM5 (8p11.22), ADAM3B (16q12.1) and ADAM6 (14q32.33) - and ADAM5 is the only one of the six with a UniProt accession, so it alone appears in GOA. The current three ND rows are harmless and arguably correct, but they carry none of what is actually known - that the locus is transcribed and that its protein has been looked for and not found. If GO wishes to retain such entries, a machine-readable way to say "this is a pseudogene, and that is why the aspects are ND" would be more informative than ND alone.
Suggested experts: GO Consortium ontology and annotation policy, UniProt
Q: GO:0004222 metalloendopeptidase activity is annotated to ADAM5 orthologues and to human ADAM2, ADAM18 and ADAM7 by IEA:InterPro (IPR001590) and by IBA (PANTHER:PTN000224844), yet none of these proteins has any HExxH zinc-binding core in its sequence, and UniProt states of mouse Adam5 in the same entry that it is "Not expected to have protease activity". Measured across all 331 Swiss-Prot reviewed members of PANTHER PTHR11905 - the reviewed subset of a 29,886-protein family - this is not a handful of exceptions: 204 of 204 members that retain the zinc site carry GO:0004222, and so do 37 of the 40 that have the Peptidase M12B fold but no HExxH core at all (92%). Losing the catalytic site therefore has almost no effect on whether a member receives the activity term. Human ADAM5 is clean only because pseudogenisation deleted the fold the signature matches on. Since GO:0004222 is defined as catalysis in which "one or two metal ions hold the water molecule in place", should the InterPro2GO mapping from IPR001590 and the PANTHER node PTN000224844 be restricted to members that retain the zinc-binding motif?
Suggested experts: PAINT / GO Central, InterPro
Q: Should the human ADAM5 UniProt entry's DOMAIN 111..199 "Disintegrin" feature be revised? Aligned to macaque ADAM5, the human sequence has a 246-residue internal deletion fused after residue 162, so this annotated domain straddles the junction: residues 111-162 correspond to macaque 144-195, upstream of the macaque disintegrin domain, and only residues 163-199 correspond to real disintegrin sequence (macaque 442-478). The feature describes a fusion created by the deletion rather than a conserved intact domain.
Suggested experts: UniProt
Q: The 2026 mouse knockout showing ADAM5 is required for zona pellucida binding and uterotubal junction transit is not yet reflected in GO annotations on mouse Adam5, which still holds no experimental biological-process annotation for fertilisation. Should this paper be curated for the mouse gene, with the human gene explicitly excluded from the propagation on pseudogene grounds?
Suggested experts: MGI, PAINT / GO Central
Experiment: Targeted parallel-reaction-monitoring mass spectrometry on human testis lysate and ejaculated sperm, using synthetic heavy-labelled proteotypic peptides selected from the retained C-terminal half of the putative 412-residue translation - the region that aligns to macaque ADAM5 at 84% identity and is therefore genuinely encoded if anything is translated. Human ADAM2 and ADAM18 peptides serve as positive controls, since both return mapped peptides in existing repositories while ADAM5 returns none.
Hypothesis: The human ADAM5 locus produces no detectable polypeptide in testis or sperm, and the 1999 negative Western blot was not a false negative caused by the heterologous macaque antiserum failing to recognise the divergent human sequence.
Type: targeted mass spectrometry
Experiment: Test the two steps the association study leaves open. First, demonstrate occupancy directly by AGO2 CLIP or biotinylated-miRNA pulldown in an oropharyngeal cell line, asking whether miR-122b-5p is recovered on the ADAM5 transcript. Second, knock down the ADAM5 transcript alone with antisense oligonucleotides in a line with fixed ADAM3A copy number, and measure ADAM9 mRNA and protein. A luciferase reporter carrying the ADAM5-homologous segment of the ADAM9 3'-UTR, with and without the predicted seed match mutated, provides the specificity control.
Hypothesis: The ADAM5 pseudogene transcript acts as a competing endogenous RNA that sequesters miR-122b-5p and thereby raises ADAM9 expression.
Type: RNA interaction and reporter assay
Experiment: Immunoprecipitate ADAM2 from human ejaculated sperm plasma membrane preparations and identify partners by LC-MS/MS, running mouse sperm in parallel as the reference in which the ADAM2-ADAM3-ADAM5 complex is known to form. The comparison states directly how much of the rodent complex survives in humans and identifies any substitute component, which is the information needed before any rodent sperm-ADAM annotation is propagated to a human gene.
Hypothesis: In the absence of ADAM3 and ADAM5, human sperm ADAM2 either forms a reduced complex or is replaced in the sperm adhesion machinery by a different protein.
Type: interaction proteomics
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: Whether the human ADAM5 locus produces any polypeptide at all is undetermined. The only direct test is a single 1999 Western blot with a heterologous (macaque) antiserum, which was negative; no targeted assay with a human-sequence-specific reagent has been reported.
OPEN BIOLOGY WHOLLY_DARK
What is known: Firmly established: the locus is transcribed, with multiple alternatively spliced testis transcripts and two independent full-length cDNA clones; the transcripts carry deletions and in-frame termination codons relative to the macaque orthologue; HGNC, NCBI and UniProt all type it as a pseudogene or uncertain product; and no MS peptide evidence is mapped to Q6NVV9 in the EBI Proteins API, where human ADAM2, human ADAM18 and mouse Adam5 all return peptides.
Significance: A transcribed unitary pseudogene with a residual open reading frame is the classic setting for a low-abundance or truncated product that a heterologous antibody would miss. Resolving it either closes the gene out cleanly or reopens it as a real, if degenerate, gene product.
What would resolve it: Targeted parallel-reaction-monitoring mass spectrometry on human testis and ejaculated sperm using proteotypic peptides chosen from the retained C-terminal half of the putative translation, together with ribosome profiling of human testis over the ADAM5 transcripts.
Provenance (the field's own admissions):
Gap: Whether the ADAM5 pseudogene transcript has a function of its own as a regulatory RNA is unknown. The proposed competing-endogenous-RNA mechanism rests on a predicted miRNA binding site and a copy-number association, with no direct demonstration that the ADAM5 transcript binds miR-122b-5p or that it is required for the observed ADAM9 change.
OPEN BIOLOGY BP_DARK
What is known: Established: inherited copy number of the ADAM3A/ADAM5 pseudogene pair associates with oropharyngeal squamous cell carcinoma risk in a two-stage study, higher copy number accompanies higher ADAM9 expression, and the ADAM5 sequence is highly homologous to the ADAM9 3'-UTR. Not established: any measured interaction, or that ADAM5 rather than ADAM3A or another element in the amplicon carries the effect.
Significance: If real, this would be a function of the locus that survives the loss of its protein, and it would be the only reason to annotate ADAM5 at all - but at the RNA level, not as a metalloproteinase.
What would resolve it: Direct interaction evidence - miRNA pulldown or AGO-CLIP showing miR-122b-5p occupancy on the ADAM5 transcript - plus a loss-of-function test in which knocking down the ADAM5 transcript alone changes ADAM9 levels, with ADAM3A copy number held constant.
Provenance (the field's own admissions):
Gap: It is not known how the human sperm ADAM adhesion complex is constituted, given that five of its rodent components - ADAM1A, ADAM1B, ADAM3A, ADAM3B and ADAM5 - are pseudogenes in humans. Whether a reduced ADAM2-based complex forms, whether another protein substitutes, or whether humans have dispensed with the mechanism is undetermined.
OPEN BIOLOGY BP_DARK
What is known: Established in mouse: ADAM5 is part of an ADAM2-ADAM3-ADAM5 complex, its loss reduces ADAM2 and ADAM3 on sperm, and the null is severely subfertile with failed uterotubal junction transit. Established in human: the gene is a pseudogene, as are ADAM1A, ADAM1B, ADAM3A, ADAM3B and ADAM6 by HGNC locus type, while ADAM2, ADAM7 and ADAM18 remain protein-coding.
Significance: The rodent sperm-ADAM literature is routinely used to reason about human fertilisation. The extent of the human losses sets a limit on that transfer, and marks which rodent-derived annotations should not be propagated to human genes.
What would resolve it: Interaction proteomics on human ejaculated sperm to determine what, if anything, ADAM2 partners with in the absence of ADAM3 and ADAM5, compared side by side with the mouse complex.
Provenance (the field's own admissions):
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)