ABHD18 is a mitochondrial alpha/beta-hydrolase-fold enzyme that deacylates cardiolipin. It hydrolyses acyl chains from tetra-acylated cardiolipin, first giving the tri-acylated intermediate monolysocardiolipin and then continuing stepwise to dilyso-cardiolipin, and it prefers substrates with shorter, saturated chains - species with more than five double bonds resist it. It is functionally homologous to yeast Cld1. That reaction is the long-sought first half of cardiolipin acyl-chain remodelling: nascent cardiolipin is deacylated by ABHD18 and then re-acylated by tafazzin with longer, unsaturated chains, and repeated cycles of hydrolysis and reacylation give mature cardiolipin the acyl composition required for inner-membrane and respiratory-chain function. Whether that pairing is best read as a remodelling cycle or as degradation opposed by tafazzin is framed differently by the two 2025 studies that characterised the enzyme. The identity of the deacylase had remained unknown until ABHD18 was found in a genetic suppressor screen. That screen also gives the gene direct clinical significance: Barth syndrome is caused by inactivating mutations in tafazzin and is characterised by accumulation of monolysocardiolipin, so inactivating the upstream deacylase relieves the block. Loss of ABHD18 rescues mitochondrial defects in Barth syndrome cells and rescues morbidity and mortality in mouse models, and a selective covalent ABHD18 inhibitor rescues tafazzin-mutant phenotypes in patient fibroblasts and fish embryos.
Definition: Catalysis of the hydrolysis of acyl chains from cardiolipin, producing monolysocardiolipin and, on further hydrolysis, dilyso-cardiolipin, with release of fatty acids. Registered in Rhea as RHEA:32935 for the first step.
Justification: GO covers the rest of cardiolipin metabolism at substrate-specific resolution - cardiolipin synthase activity (GO:0008808), cardiolipin synthase (CMP-forming) activity (GO:0043337), cardiolipin phospholipase D activity (GO:0035755), cardiolipin dehydrogenase (NAD+) activity (GO:0160241) - but has no term for the deacylation step, even though that step is half of the cycle described by GO:0035965 cardiolipin acyl-chain remodeling. The gap is historical: the responsible enzyme was unidentified until ABHD18 was described, so the most specific molecular function available is GO:0004623 phospholipase A2 activity - and even that conveys neither substrate nor product, while its generic parent GO:0004620 glycerophospholipase activity conveys less still. The reaction is registered in Rhea as RHEA:32935 and recorded in UniProt with that cross-reference, so the chemistry is unambiguous; RHEA:32935 also places the removed chain at sn-2, which is why the A2-type term is used as the interim parent - though that placement is a Rhea reaction representation behind a UniProt ECO:0000305 curator inference rather than measured regiospecificity. The activity is conserved - ABHD18 is functionally homologous to yeast Cld1 - and it is stepwise rather than single-turnover, proceeding past monolysocardiolipin to dilyso-cardiolipin, so a term naming only the first product would itself be incomplete. A term at this granularity would also let the deacylase and the reacylase (tafazzin) be distinguished by molecular function rather than only by their shared process term.
Parent term: A2-type glycerophospholipase activity
Supporting Evidence:
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005739 mitochondrion | IEA GO_REF:0000044 | ACCEPT | Summary: Electronic mapping from the UniProt "Mitochondrion" subcellular location, which is itself experimentally supported. Correct and core: cardiolipin is synthesised and remodelled in the mitochondrion, so this is where ABHD18's deacylase activity must act. The compartment could in principle be stated more precisely - cardiolipin remodelling occurs at the inner membrane - but no direct sub-mitochondrial assay exists for this protein yet. Reason: Examined as a synthesis row (ORGANELLES policy). The GO_REF:0000044 mapping comes from a UniProt location line whose evidence is ECO:0000269 to two papers (the MitoCoP proteome and the 2025 functional study), and UniProt annotates a predicted N-terminal transit peptide (residues 1-55), so the organelle claim is right. The source line names only the organelle; the inner membrane, inferred from several agreeing lines rather than assayed, is added alongside in the NEW GO:0005743 row. Supporting Evidence: file:human/ABHD18/ABHD18-uniprot.txt CC -!- SUBCELLULAR LOCATION: Mitochondrion {ECO:0000269|PubMed:34800366, |
| GO:0005739 mitochondrion | HTP PMID:34800366 Quantitative high-confidence human mitochondrial proteome an... | ACCEPT | Summary: Assignment to the high-confidence human mitochondrial proteome (MitoCoP). High-throughput, but it is one of the two ECO:0000269 sources behind UniProt's own subcellular location statement for this protein, and it agrees with the compartment the enzyme's substrate occupies. Correct. Reason: Direct (high-throughput) evidence for the organelle. A whole-organelle proteome cannot resolve the sub-compartment, so the row stays at the root; the inner membrane is added in the NEW GO:0005743 row. Supporting Evidence: file:human/ABHD18/ABHD18-uniprot.txt CC -!- SUBCELLULAR LOCATION: Mitochondrion {ECO:0000269|PubMed:34800366, |
| GO:0004623 A2-type glycerophospholipase activity | IDA PMID:40903572 Genetic suppression features ABHD18 as a Barth syndrome ther... | NEW | Summary: Proposed new annotation supplying the molecular function, which GOA lacks entirely. ABHD18 hydrolyses acyl chains from cardiolipin, shown directly in vitro; UniProt records the reaction with a Rhea cross-reference (RHEA:32935). The regiochemistry justifies the A2-type child rather than the generic parent. RHEA:32935 writes the product as 1'-[1,2-diacyl-sn-glycero-3-phospho],3'-[1-acyl-sn-glycero-3-phospho]-glycerol - the remaining chain on the deacylated glycerol is at sn-1, implying the chain removed was at sn-2, which is A2 regiochemistry. This is a caveat rather than a demonstration: UniProt tags the CATALYTIC ACTIVITY block ECO:0000305 (curator inference), so RHEA:32935 is a reaction REPRESENTATION and neither 2025 paper determines regiochemistry experimentally. GO:0004623 is used as the more informative of the two available terms on that basis; the generic GO:0004620 would also be defensible, and if regiospecificity is later measured and differs, this row should move. Neither term names the substrate. There is no cardiolipin-specific hydrolase term - GO has cardiolipin synthase (GO:0008808), cardiolipin phospholipase D (GO:0035755) and cardiolipin dehydrogenase (GO:0160241), but nothing for deacylation - so that gap is filed under proposed_new_terms rather than forced. Reason: The gene has no molecular function annotation at all despite a directly demonstrated, Rhea-registered catalytic activity. A2-type is chosen over the generic parent because RHEA:32935 places the removed chain at sn-2 - but that is the Rhea reaction representation behind a UniProt ECO:0000305 curator inference, not measured regiospecificity, so the choice is provisional. Supporting Evidence: PMID:40903572 ABHD18 converts CL into monolysocardiolipin (MLCL) in vitro, and its inactivation in cells and mice results in a shift to nCL in serum and tissues. file:human/ABHD18/ABHD18-uniprot.txt CC -!- FUNCTION: Catalyzes the hydrolysis of an acyl chain from the tetra- |
| GO:0035965 cardiolipin acyl-chain remodeling | IMP PMID:40903572 Genetic suppression features ABHD18 as a Barth syndrome ther... | NEW | Summary: Proposed new annotation. Cardiolipin acyl-chain remodelling is the cycle in which nascent cardiolipin is deacylated and then re-acylated with longer, unsaturated chains; ABHD18 performs the deacylation and tafazzin the reacylation. Tafazzin already carries this term, so annotating ABHD18 completes a process that until 2025 had only one identified enzyme. Two independent 2025 studies frame the same enzymology differently and the difference is worth recording. PMID:40903572 presents ABHD18 as the missing deacylase of the remodelling pathway. PMID:40378955 presents it as a degradative lipase functionally homologous to yeast Cld1, concluding that ABHD18 catalyses the breakdown of cardiolipin whereas tafazzin protects cardiolipin from degradation - and shows the deacylation is stepwise, proceeding past monolysocardiolipin to dilyso-cardiolipin, with species carrying more than five double bonds resistant. GO:0035965 is retained because both accounts agree ABHD18 supplies the deacylation step of the ABHD18/tafazzin pair, and because GO has no cardiolipin catabolic process term to express the degradative reading. A curator taking the PMID:40378955 framing at face value would want one; that gap is noted in suggested_questions. Reason: Places the gene in the pathway its catalytic activity belongs to, using the term its partner enzyme already carries; GOA currently records no biological process for ABHD18. Supporting Evidence: PMID:40903572 ABHD18 converts CL into monolysocardiolipin (MLCL) in vitro, and its inactivation in cells and mice results in a shift to nCL in serum and tissues. file:human/ABHD18/ABHD18-uniprot.txt Together with Tafazzin (TAZ), it participates in the mechanism of CL PMID:40378955 Knockdown of Abhd18 decreased the concentration of MLCL in murine, Taz-knockout myoblasts. |
| GO:0005743 mitochondrial inner membrane | IC PMID:40378955 ABHD18 degrades cardiolipin by stepwise hydrolysis of fatty ... | NEW | Summary: Proposed sub-compartment. ABHD18 deacylates cardiolipin, a lipid made and largely held in the mitochondrial inner membrane, and the most likely place for it is the matrix face of that membrane. The existing mitochondrion rows stop at the organelle. Reason: Several independent lines agree, but none is a direct sub-mitochondrial assay on ABHD18, hence IC. (1) Targeting - UniProt annotates an N-terminal mitochondrial transit peptide (residues 1-55) before the mature chain, and the entry has no transmembrane or intramembrane segment, so the protein is imported past the outer membrane rather than anchored in it. (2) Substrate - the enzyme hydrolyses cardiolipin, which is found predominantly in the inner membrane and is synthesised on its matrix-facing leaflet; a lipase without a membrane anchor must act at that membrane surface rather than free in the matrix. (3) Functional homologue - yeast Cld1, whose cardiolipin-deacylating role ABHD18 replicates (PMID:40378955), is an interfacial inner-membrane protein facing the matrix by protease protection and carbonate extraction (PMID:23637464). Line 3 is an analogy rather than orthology - the same paper notes that ABHD4, not ABHD18, is the mammalian ABHD closest to Cld1 by phylogeny - so it corroborates but does not carry the call. Lines 2 and 3 share one premise, that cardiolipin lives in the inner membrane, so the call rests on two independent lines (import past the outer membrane, and an inner-membrane substrate); the protease-protection experiment in suggested_experiments would settle it. Lines 1 and 2 alone point to the inner membrane rather than the matrix lumen; which leaflet is not established for ABHD18, so the generic inner-membrane term is used rather than GO:0099617 matrix side of mitochondrial inner membrane. Not added to core_functions locations, since the human activity has not been shown at that membrane. Supporting Evidence: file:human/ABHD18/ABHD18-uniprot.txt FT TRANSIT 1..55 FT /note="Mitochondrion" PMID:40378955 In eukaryotes, it is found predominantly in the inner mitochondrial membrane PMID:40378955 The present data establish Abhd18 as a functional homolog of cld1 PMID:40378955 Among the 23 mammalian ABHD proteins annotated in UniProt, ABHD4 is most closely related to Cld1 by phylogenetic analysis PMID:23637464 Cld1p was protected from the protease, unless the IM was solubilized, similar to the matrix protein Kgd1p PMID:23637464 Cld1p is embedded in the IM facing the mitochondrial matrix PMID:23637464 how does immature CL synthesized on the matrix-facing leaflet of the IM gain access to tafazzin |
Loading supporting content…
Download this section (compressed HTML)Q: Where in the mitochondrion does ABHD18 act? Cardiolipin remodelling happens at the inner membrane, and its partner tafazzin is annotated to both the inner and outer mitochondrial membranes, but ABHD18's only localisation evidence is an unqualified "mitochondrion" from a proteomic survey. Submitochondrial localisation would sharpen the cellular component annotation and constrain how the deacylase and the reacylase meet their shared substrate.
Q: Is ABHD18 the only cardiolipin deacylase, or one of several? It was found as a suppressor of tafazzin loss rather than by a systematic search, so other contributing enzymes may exist. That matters therapeutically, since redundancy would limit how completely an ABHD18 inhibitor can normalise the monolysocardiolipin-to-cardiolipin ratio in Barth syndrome.
Q: What determines the acyl-chain preference? ABHD18 prefers cardiolipin with shorter, saturated chains, which is what makes the remodelling cycle directional - nascent species are consumed and mature ones spared. The structural basis is unknown, and it is the key to designing inhibitors that spare mature cardiolipin.
Q: Does ABHD18 have substrates besides cardiolipin? It is an alpha/beta-hydrolase, a fold with broad substrate range across the family, and only cardiolipin has been tested. A selective covalent inhibitor now exists, so off-target lipid changes can be surveyed directly.
Q: Is ABHD18 better described as remodelling or as degradation, and does GO need a cardiolipin catabolic process term? PMID:40903572 presents ABHD18 as the missing deacylase of the remodelling cycle; PMID:40378955 concludes that ABHD18 catalyses the breakdown of cardiolipin whereas tafazzin protects it, and shows deacylation continues past monolysocardiolipin to dilyso-cardiolipin. GO has cardiolipin biosynthetic process and cardiolipin acyl-chain remodeling but no catabolic term, so the degradative reading currently cannot be expressed.
Experiment: Determine the submitochondrial localisation and topology of ABHD18 by protease protection on isolated mitochondria and mitoplasts combined with immuno-electron microscopy, and test for proximity to tafazzin and CRLS1 by split-reporter or proximity-labelling assays.
Hypothesis: ABHD18 acts at the mitochondrial inner membrane, in proximity to tafazzin and cardiolipin synthase.
Type: cell biology
Experiment: Measure kinetics of purified ABHD18 against a defined panel of cardiolipin species varying systematically in chain length and unsaturation, then test whether that selectivity, modelled together with tafazzin's known reacylation preferences, is sufficient to explain the observed steady-state cardiolipin composition.
Hypothesis: ABHD18's preference for short, saturated acyl chains is what makes cardiolipin remodelling directional.
Type: enzymology
Experiment: Compare the lipidomic consequences of complete ABHD18 loss with those of the covalent inhibitor at saturating concentration; residual monolysocardiolipin production in the null background would indicate additional deacylases and set an upper bound on what inhibition can achieve therapeutically.
Hypothesis: ABHD18 is the sole non-redundant cardiolipin deacylase.
Type: lipidomics and genetics
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)