FAM20C is a secretory-pathway protein kinase with a strong serine preference and measured threonine activity in protein substrates. Its intracellular form is a membrane protein with a lumen-facing catalytic region, concentrated in the Golgi and able to redistribute to the endoplasmic reticulum during stress. Proteolytic processing can release an active secreted form beginning at Asp93. FAM20C phosphorylates extracellular matrix and enamel proteins, as well as secretory-pathway proteins including protein disulfide isomerase and Ero1Ξ±. These reactions contribute to biomineralization of bone and teeth and context-dependent ER proteostasis. FAM20A enhances FAM20C kinase activity. The natural 584- and 252-residue products are distinct from the processed 93β584 chain; activity and localization of the shorter natural product remain unresolved.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0000139 Golgi membrane | EXP PMID:32149426 Phosphorylation switches protein disulfide isomerase activit... | ACCEPT | Summary: FAM20C occupies the Golgi membrane before stress-dependent redistribution. Reason: PMID:32149426 Figure 3 shows WT FAM20C-mApple with GM130 in human HepG2 cells before thapsigargin treatment, followed by reversible redistribution toward the PDI-positive ER. The KDEL-tagged retention construct is a separate engineered control. Independent human HeLa extraction and fluorescence-protease-protection experiments in PMID:34349020 Figure 1 establish membrane integration and a lumen-facing C-terminal domain. This supports the Golgi membrane pool without treating every secreted or processed form as membrane-bound. |
| GO:0000139 Golgi membrane | IDA PMID:34349020 Proteolytic processing of secretory pathway kinase Fam20C by... | ACCEPT | Summary: The intracellular FAM20C precursor is an integral Golgi membrane protein. Reason: PMID:34349020 Figure 1 and Supporting Information alkaline-extraction/FPP methods distinguish membrane-associated WT FAM20C from the engineered ΞTM protein. WT remains in the membrane pellet after alkaline/high-salt extraction and becomes soluble with detergent; C-terminal mApple is protected until membrane permeabilization. GM130 colocalization and trafficking experiments support the Golgi pool. Intracellular Met4 and secreted Asp93 N termini are experimentally distinguished; the latter soluble processed pool does not invalidate precursor membrane localization. Supporting Evidence: |
| GO:0000139 Golgi membrane | IEA GO_REF:0000044 | ACCEPT | Summary: Retain the Golgi membrane location. Reason: The current SL-0134 mapping is exact. Human FAM20C membrane extraction and fluorescence-protease-protection experiments in PMID:34349020 Figure 1 establish a type-II membrane precursor with a lumen-facing catalytic region. This supports Golgi membrane directly, while the secreted Asp93 product and natural 252-residue splice product remain distinct. |
| GO:0002020 protease binding | IPI PMID:34349020 Proteolytic processing of secretory pathway kinase Fam20C by... | KEEP AS NON CORE | Summary: FAM20C associates with its processing protease S1P/MBTPS1. Reason: PMID:34349020 Figure 4 and the construct/co-immunoprecipitation methods support association with human MBTPS1/S1P, including a catalytic-mutant comparison, while cleavage-site, inhibition and knockout/rescue experiments identify S1P as the enzyme that processes FAM20C. The protease-binding definition fits this substrate interaction. Retain it as a contextual non-core binding function; it does not make FAM20C a protease or a protease inhibitor. Cellular co-association does not establish an isolated binary interface or a native stoichiometric complex. Supporting Evidence: |
| GO:0004672 protein kinase activity | IDA PMID:34349020 Proteolytic processing of secretory pathway kinase Fam20C by... | MODIFY | Summary: FAM20C catalyzes protein serine/threonine phosphorylation, with serine preferred. Reason: PMID:34349020 Figures 5 and 6 compare human recombinant 23β584/141β578 and engineered cleavage-site preparations, OPN/Ero1Ξ± substrates, and S106A/D478A controls. These establish kinase activity and serine autophosphorylation, while S1P or HRV3C performs the proteolytic step. Refine the broad protein-kinase term to the independently supported serine/threonine activity. Independent human IGFBP1 phosphorylation in PMID:26091039 (Results and Figure 5E/F) supports the additional threonine breadth; the source-specific serine experiments retain their stated scope. Proposed replacements: protein serine/threonine kinase activity |
| GO:0004674 protein serine/threonine kinase activity | IBA GO_REF:0000033 | ACCEPT | Summary: Retain inherited serine/threonine kinase activity. Reason: The displayed 584-residue human target PTN002488223 descends from the positive GO:0004674 IBD at PTN002647017 in PTHR12450 v19. Mouse FAM20C kinase assays in PMID:22900076 support inheritance; human self-WITH is valid descendant evidence. Human PMID:22582013 reports a serine preference with weaker threonine activity, and PMID:26091039 Results/Figure 5E/F provides WT-versus-D478A IGFBP1 protein-substrate evidence including Thr193. The inspected UniProt record independently represents the threonyl-protein reaction as RHEA:46608. Retain this additional substrate breadth without claiming unrestricted physiological threonine specificity. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002647017 SUPPORTS TRANSFER Actual positive IBD, target descent and displayed-sequence MSA verified. Mouse kinase and human substrate-specificity/protein-substrate experiments support inheritance. The unresolved historical rat primary trace does not negate these sources. Supporting Evidence: GO_REF:0000033 file:human/FAM20C/FAM20C-uniprot.txt |
| GO:0004674 protein serine/threonine kinase activity | IDA PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | ACCEPT | Summary: Human FAM20C preferentially phosphorylates serine while retaining measured threonine capacity. Reason: PMID:22582013 Figure 1 and the main Results describe human FLAG-FAM20C purified from HEK293T conditioned medium, peptide-array phosphorylation and stronger serine than threonine or tyrosine phosphorylation. Figure 2 supplies WT/D478A and phosphatase controls for casein phosphorylation. The 2012 supplemental array table and Figure S5 pixels were not recovered, so their individual target rows are not claimed as inspected. Independent human IGFBP1 phosphorylation in PMID:26091039 (Results and Figure 5E/F) supports the additional threonine breadth; the source-specific serine experiments retain their stated scope. |
| GO:0004674 protein serine/threonine kinase activity | IDA PMID:23754375 Crystal structure of the Golgi casein kinase. | ACCEPT | Summary: Human FAM20C has experimentally supported serine/threonine kinase activity. Reason: PMID:23754375 tests human FAM20C immunopurified from HEK293T medium on a serine-containing Ξ²-casein peptide and human active-site mutants; Supporting Information Methods and Figure S5A/B provide human Mn-ATP/Mg-ATP kinetics. The crystallographic protein is C. elegans ceFam20, not human FAM20C. Independent human IGFBP1 phosphorylation in PMID:26091039 (Results and Figure 5E/F) supports the additional threonine breadth; the source-specific serine experiments retain their stated scope. |
| GO:0004674 protein serine/threonine kinase activity | IDA PMID:25789606 A secretory kinase complex regulates extracellular protein p... | ACCEPT | Summary: FAM20C phosphorylates protein substrates and is stimulated by FAM20A. Reason: PMID:25789606 Figures 1β4 directly compare WT and D478A human FAM20C on recombinant enamel proteins and map serine-dependent phosphorylation. Purified human FAM20C 93β584 and FAM20A 63β529 establish stimulation of the kinase by the pseudokinase; FAM20A is the activator, not the substrate-defined function of FAM20C. Independent human IGFBP1 phosphorylation in PMID:26091039 (Results and Figure 5E/F) supports the additional threonine breadth; the source-specific serine experiments retain their stated scope. |
| GO:0004674 protein serine/threonine kinase activity | IDA PMID:26091039 A Single Kinase Generates the Majority of the Secreted Phosp... | ACCEPT | Summary: Human FAM20C directly phosphorylates serine and threonine in protein substrates. Reason: PMID:26091039 Results, under broader substrate specificity, explicitly names IGFBP1 Ser45, Thr193 and Ser199 after treatment with WT but not D478A FAM20C. Figure 5E supplies the controlled protein-substrate time course. The separately inspected Figure 5F and supplemental S6F concern different phosphopeptides; their exact assignments and public image/supplement links are recorded in the reference assessment and source guide. Negative corresponding short-peptide assays limit motif-only inference. Together with the curated UniProt RHEA:46608 reaction, this supports serine/threonine capacity with a strong serine preference, not universal physiological threonine phosphorylation. |
| GO:0004674 protein serine/threonine kinase activity | IDA PMID:29858230 Secretory kinase Fam20C tunes endoplasmic reticulum redox st... | ACCEPT | Summary: FAM20C directly phosphorylates Ero1Ξ± on serine within its broader serine/threonine activity. Reason: PMID:29858230 Figures 2β4 and protein-expression/kinase-assay methods use human FAM20C 141β578 purified from insect cells, WT/D478A comparisons, and Ero1Ξ± Ser145 mutants. These identify the kinase as the performer of phosphorylation; Ero1Ξ± performs its own oxidoreductase chemistry. The Ero1Ξ±-KDEL substrate is not evidence that WT FAM20C itself carries a retention signal. Independent human IGFBP1 phosphorylation in PMID:26091039 (Results and Figure 5E/F) supports the additional threonine breadth; the source-specific serine experiments retain their stated scope. |
| GO:0004674 protein serine/threonine kinase activity | IEA GO_REF:0000120 | ACCEPT | Summary: Retain the combined serine/threonine kinase inference. Reason: EC 2.7.11.1 maps to GO:0004674, and mouse Q5MJS3/ENSMUSP00000026972 has experimental kinase evidence in PMID:22900076. Human PMID:22582013 reports weaker Thr than Ser activity; PMID:26091039 Results and Figure 5E/F directly support IGFBP1 phosphorylation including Thr193. UniProt represents the threonyl-protein reaction as RHEA:46608. Retain the measured breadth beyond serine-only chemistry while keeping physiological specificity unresolved. Supporting Evidence: GO_REF:0000120 file:human/FAM20C/FAM20C-uniprot.txt |
| GO:0004674 protein serine/threonine kinase activity | TAS Reactome:R-HSA-8952289 | ACCEPT | Summary: Retain the kinase role in the phosphorylation reaction. Reason: R-HSA-8952289 has catalyst activity 8952283, whose physical entity 8952321 contains native human Q8IXL6 entity 8952276. FAM20C is the modeled enzyme, not merely a substrate. PMID:26091039 Results and Figure 5E/F compare WT and D478A kinase and support IGFBP1 phosphorylation including Thr193. This additional breadth, the PMID:22582013 serine preference and curated RHEA:46608 reaction support Ser/Thr capacity. The model-specific processed ER-lumen location is assessed separately. Supporting Evidence: Reactome:R-HSA-8952289 file:human/FAM20C/FAM20C-uniprot.txt |
| GO:0005509 calcium ion binding | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Retain calcium binding as a bounded ortholog-supported capacity. Reason: The source mouse Q5MJS3 has a calcium-binding IDA to PMID:17369251. Its original Methods p15359 and Results/Figure 8 p15363 describe full-coding-region mouse DMP4/FAM20C expressed in bacteria and a positive calcium overlay with DMP1/calmodulin and BSA controls. This is positive binding evidence, not evidence that calcium activates the kinase: PMID:22900076 found no significant calcium effect in its kinase assay. Retain the inherited capacity as non-core because a native human calcium occupancy or necessary physiological role was not established by these experiments. The current GO:0005509 definition requires calcium binding, not calcium-dependent catalysis. Absence of a resolved human calcium site or a calcium effect on kinase activity does not contradict the positive donor binding assay; it limits physiological interpretation rather than establishing over-annotation. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:Q5MJS3 SUPPORTS TRANSFER Experimental mouse calcium overlay recovered; target transfer is plausible, but its physiological human role and product specificity remain unresolved. |
| GO:0005515 protein binding | IPI PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | KEEP AS NON CORE | Summary: APOL1 is a supported FAM20C substrate-associated partner. Reason: The exact APOL1 peptide entry in PMID:22582013 Table S1 was not accessible. Independently, PMID:26091039 Figure 4C tests tagged APOL1 with WT versus D478A FAM20C and Ξ»-phosphatase, supporting target-dependent phosphorylation in cells. This corroborates the imported O14791 partner without pretending that the 2012 supplemental target row or purified APOL1 binary binding was read. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. |
| GO:0005515 protein binding | IPI PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | UNDECIDED | Summary: The exact antithrombin/SERPINC1 interaction remains unverified. Reason: The imported partner is P01008/SERPINC1. The inspected UniProt interaction block also records this pair with NbExp=2 and IntAct EBI-7147442/EBI-1039832, so curator support is explicit. However, the exact PMID:22582013 supplemental target assay remains inaccessible and no independently inspected SERPINC1 experiment resolves its scope. Other retained substrate partners have separately read target-specific experiments, including the later APOL1/CHGB/FGA tests; their actions were not based on UniProt pair-list presence alone. Retain UNDECIDED without inferring nonbinding, curator error or antithrombin inhibitor activity for FAM20C. |
| GO:0005515 protein binding | IPI PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | KEEP AS NON CORE | Summary: FAM20C engages bovine Ξ²-casein as a phosphorylation substrate. Reason: PMID:22582013 Figure 2B directly compares phosphorylation of recombinant Ξ²-casein by WT and D478A human FAM20C; the imported P02666 is bovine Ξ²-casein. The human cellular Ξ±s1-casein experiments are distinct assays and do not change the identity of this partner. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | KEEP AS NON CORE | Summary: Fibrinogen Ξ±-chain is supported as a FAM20C substrate-associated partner. Reason: The exact PMID:22582013 Table S1 entry was not recovered. Independent PMID:26091039 Figure 4A shows WT-versus-D478A phosphorylation of a purchased fibrinogen preparation, with the Ξ±-chain band identified in the figure. The supplemental Methods name the supplier but do not supply a sequence-verified human Ξ±-chain construct. Preserve the imported P02671 identity and the preparation/chain limits rather than claiming a sequenced isolated human-chain binding assay. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. |
| GO:0005515 protein binding | IPI PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | KEEP AS NON CORE | Summary: Chromogranin B/SCG1 is a supported FAM20C substrate-associated partner. Reason: The exact 2012 supplemental peptide entry was not recovered. PMID:26091039 Figure 4C/E independently tests SCG1/CHGB phosphorylation using WT/D478A FAM20C, phosphatase treatment and FAM20C-deficient cells. P05060 denotes chromogranin B, also called secretogranin I; this is not secretogranin II. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. |
| GO:0005515 protein binding | IPI PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | KEEP AS NON CORE | Summary: Osteopontin is directly phosphorylated by FAM20C. Reason: PMID:22582013 Figure 3 compares WT and D478A FAM20C on recombinant OPN and tests tagged OPN phosphorylation in cells. These are direct catalytic-substrate and cellular phosphorylation assays, not measurements of stable binding affinity. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | KEEP AS NON CORE | Summary: The OPN interaction is supported with a processed-product identity limit. Reason: PMID:22582013 Figure 3 supplies recombinant OPN and cellular phosphorylation evidence. The imported partner includes P10451-PRO_0000020321, but the historical assayed preparation was not independently matched residue-for-residue to that current processed-product identifier. Preserve this source identity and retain the supported OPN substrate interaction without asserting equivalence of all OPN cleavage products. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:25789606 A secretory kinase complex regulates extracellular protein p... | KEEP AS NON CORE | Summary: OPN phosphorylation supports the interaction without resolving the exact current processed product. Reason: PMID:25789606 Figure 1 and its supplements compare OPN phosphorylation in WT and FAM20C-deficient mouse ameloblast-like cells and human U2OS cells. The imported P10451-PRO_0000020321 remains literal; the historical construct was not sequence-matched to the current processed-product boundary. The phenotype supports substrate engagement rather than a new stable-complex mechanism. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:25789606 A secretory kinase complex regulates extracellular protein p... | KEEP AS NON CORE | Summary: Amelotin is a direct phosphorylation substrate of FAM20C. Reason: PMID:25789606 Figure 1 tests recombinant human AMTN with WT versus D478A FAM20C. Human cDNA provenance and bacterial substrate expression are described in the protein-expression Methods; the insect-cell FAM20C kinase preparation is distinct from the substrate host. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:25789606 A secretory kinase complex regulates extracellular protein p... | KEEP AS NON CORE | Summary: FAM20C associates with the activating pseudokinase FAM20A. Reason: PMID:25789606 Figures 3 and 5 show cellular co-IP, co-elution and purified-protein activation, with FAM20B as a negative comparison. Human FAM20A 63β529 and FAM20C 93β584 preparations support an activating association; analytical ultracentrifugation supports a 2:2 assembly under those conditions. This does not establish universal native stoichiometry or make FAM20C the kinase activator. Current kinase-binding/regulator and heterodimerization terms were considered, but applying regulator activity would reverse the measured role and a dimer-only claim would overstate the assembly. Retain the supported generic binding as non-core under the standing campaign exception. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:25789606 A secretory kinase complex regulates extracellular protein p... | KEEP AS NON CORE | Summary: Ameloblastin is engaged as a FAM20C phosphorylation substrate. Reason: PMID:25789606 Figure 1 and supplements use human AMBN sequences, including a recombinant SUMO-AMBN 27β48 fragment with Ser43 mutagenesis, and distinguish mouse ameloblast-like-cell from human U2OS phosphorylation contexts. This is substrate-specific catalytic evidence, not a whole native AMBN complex structure. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:25789606 A secretory kinase complex regulates extracellular protein p... | KEEP AS NON CORE | Summary: Enamelin is a direct FAM20C phosphorylation substrate. Reason: PMID:25789606 Figure 1 tests human ENAM 173β277, with S191A/S216L single and double mutants, against WT/D478A FAM20C. Loss of labeling with the double substrate mutant and kinase-dead control supports specific substrate engagement; it does not demonstrate a stable full-length enamelin complex. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:26091039 A Single Kinase Generates the Majority of the Secreted Phosp... | KEEP AS NON CORE | Summary: Ξ²-casein is a verified model substrate for FAM20C. Reason: PMID:26091039 uses Ξ²-casein/Ξ²28β40 kinase assays and WT-versus-D478A controls; independent PMID:22582013 Figure 2 directly tests recombinant bovine Ξ²-casein, matching imported P02666. Model-substrate phosphorylation is retained at that scope rather than recast as an unrelated signaling role. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. |
| GO:0005515 protein binding | IPI PMID:26091039 A Single Kinase Generates the Majority of the Secreted Phosp... | KEEP AS NON CORE | Summary: FAM20C phosphorylates the fibrinogen Ξ±-chain in the tested preparation. Reason: PMID:26091039 Figure 4A directly compares WT/D478A kinase on purchased fibrinogen and labels the Ξ±/Ξ²/Ξ³ bands. The Ξ±-chain phosphorylation supports the source partner, but the supplier-only substrate description does not establish an independently sequenced human Ξ±-chain reagent or a stable binding constant. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:26091039 A Single Kinase Generates the Majority of the Secreted Phosp... | KEEP AS NON CORE | Summary: PRP1/PRH1 is a FAM20C substrate with context-dependent recognition. Reason: PMID:26091039 Figure 5HβK tests a PRP1-derived peptide and tagged protein, with Ser8/Ser22 mutations and WT/D478A FAM20C. These experiments support the P02810 partner and reveal recognition beyond a simple acidic consensus; they do not establish a distinct PRP1 receptor or adaptor activity. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:26091039 A Single Kinase Generates the Majority of the Secreted Phosp... | KEEP AS NON CORE | Summary: IGFBP1 is directly phosphorylated by FAM20C. Reason: PMID:26091039 Figure 5E/F compares purified WT and D478A FAM20C with recombinant insect-cell IGFBP1 and directly demonstrates target phosphorylation. Negative short-peptide assays emphasize protein context. IGF-binding or transport activity belongs to IGFBP1 and is not transferred to FAM20C. Retain the supported transient enzymeβsubstrate interaction as non-core under the standing generic-binding exception; these assays do not establish stable binary affinity or an adaptor function. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:26091039 A Single Kinase Generates the Majority of the Secreted Phosp... | KEEP AS NON CORE | Summary: OPN is a supported FAM20C substrate-associated partner. Reason: PMID:26091039 Figure 1 and supplemental biochemical procedures use recombinant OPN and cellular phosphorylation/kinase controls; independent PMID:22582013 Figure 3 supplies WT/D478A substrate assays. Substrate recognition is supported without asserting stable binary affinity or a new extracellular scaffold function. Retain this supported interaction as non-core under the standing campaign exception for generic binding. Target-directed phosphorylation supports transient enzymeβsubstrate engagement, not a stable binary complex or an additional adaptor function; the binding row is not converted laterally to kinase activity. |
| GO:0005515 protein binding | IPI PMID:34349020 Proteolytic processing of secretory pathway kinase Fam20C by... | KEEP AS NON CORE | Summary: FAM20C associates with SEC24A during secretory cargo transport. Reason: PMID:34349020 Figure 2 and Supporting Information describe tagged SEC24A/SEC23A co-immunoprecipitation and RUSH/Sar1 trafficking experiments. FAM20C is the transported cargo; the association does not make it a COPII coat component or a cargo adaptor. A current specificity search did not justify a more informative performer-level MF. Retain supported generic co-association as non-core under the standing campaign exception, without a purified binary interface or native stoichiometry claim. Supporting Evidence: |
| GO:0005515 protein binding | IPI PMID:34349020 Proteolytic processing of secretory pathway kinase Fam20C by... | KEEP AS NON CORE | Summary: FAM20C associates with SEC23A in the COPII transport context. Reason: PMID:34349020 Figure 2 and Supporting Information use tagged SEC23A and SEC24A with WT/engineered FAM20C and trafficking controls. These support cargo-associated co-IP; they do not show FAM20C performing coat assembly or vesicle formation. Retain this supported generic interaction as non-core under the standing campaign exception; no specific adaptor or coat MF is warranted. Supporting Evidence: |
| GO:0005576 extracellular region | EXP PMID:26091039 A Single Kinase Generates the Majority of the Secreted Phosp... | ACCEPT | Summary: FAM20C has a secreted extracellular protein pool. Reason: PMID:26091039 Figure 1 and Supplemental Experimental Procedures recover tagged human FAM20C from conditioned medium and identify an Asp93 N terminus. Secretion controls and intracellular retention constructs distinguish pools, while independent PMID:34349020 clarifies precursor membrane topology and S1P processing. This is extracellular localization, not a claim that every phosphorylation event occurs after secretion. |
| GO:0005576 extracellular region | IBA GO_REF:0000033 | ACCEPT | Summary: Retain the inherited extracellular pool. Reason: The positive extracellular-region IBD at PTN000278680 is on the actual human target path. Human FAM20C is legitimate descendant evidence in its own WITH set; the other donor count is not a measure of inheritance quality. Mouse secretion in PMID:22900076 and human conditioned-medium recovery with an Asp93 mature N terminus in PMIDs:26091039/34349020 independently support extracellular FAM20C. This does not locate every catalytic event outside the cell or extend the result to the 252-residue product. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN000278680 SUPPORTS TRANSFER Actual positive extracellular IBD and target descent verified. Human and mouse secretion evidence supports the target; no claim that every extant donor original was reread. |
| GO:0005576 extracellular region | IDA PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | ACCEPT | Summary: A substantial tagged FAM20C pool is secreted from human cells. Reason: PMID:22582013 Figure 1B tests FLAG-FAM20C in U2OS cell extracts and conditioned medium; Figure 1D uses secreted HEK293T-derived enzyme. Figure 1C Golgi imaging uses HeLa cells and is a separate experiment. The extracellular pool is supported without transferring the paperβs original signal-peptide-only model to the later-established membrane precursor. Supporting Evidence: |
| GO:0005576 extracellular region | IDA PMID:34349020 Proteolytic processing of secretory pathway kinase Fam20C by... | ACCEPT | Summary: S1P processing releases a soluble extracellular FAM20C pool. Reason: PMID:34349020 Figures 1, 4 and 5 distinguish intracellular Met4-starting membrane-associated protein from secreted Asp93-starting protein, with S1P perturbation and processing-site controls. This directly supports the extracellular pool while keeping S1P-mediated proteolysis separate from FAM20C-catalyzed phosphorylation. Supporting Evidence: |
| GO:0005576 extracellular region | IEA GO_REF:0000120 | ACCEPT | Summary: Retain extracellular localization supported by both source components. Reason: The official Secreted/SL-0243 mapping is GO:0005576. The mouse source has experimentally supported secretion, and independent human studies recover FAM20C from conditioned medium and identify the processed Asp93 form. These source components agree on an extracellular pool without conflating secretion with the intracellular membrane precursor or inferring identical behavior of both natural products. |
| GO:0005737 cytoplasm | IEA GO_REF:0000107 | MODIFY | Summary: Refine the broad cytoplasmic location to the demonstrated Golgi membrane pool. Reason: The mouse source is real: PMID:21549343 reports Fam20c immunofluorescence in ameloblast and odontoblast cytoplasm as well as extracellular dental matrix. Cytoplasm includes organelles and is not equivalent to soluble cytosol. The human membrane/topology experiments in PMID:34349020 provide the more informative Golgi-membrane location for the conserved intracellular kinase. This cross-source refinement does not claim that the mouse whole-cell staining resolved a membrane or that all non-Golgi signal is absent. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB:Q5MJS3 SUPPORTS TRANSFER Positive mouse cytoplasmic immunofluorescence is retained as coarse localization. Independent human topology supports a more informative Golgi-membrane term, not a donor or orthology failure. Proposed replacements: Golgi membrane |
| GO:0005783 endoplasmic reticulum | EXP PMID:32149426 Phosphorylation switches protein disulfide isomerase activit... | MODIFY | Summary: WT FAM20C can redistribute to the ER membrane during stress. Reason: PMID:32149426 Figure 3 shows reversible WT FAM20C-mApple redistribution from the GM130-positive Golgi toward the PDI-positive ER in human HepG2 cells after thapsigargin. Engineered KDEL retention and condensates are separate conditions. Independent human precursor extraction/FPP assays in PMID:34349020 establish membrane integration and a lumen-facing kinase domain, supporting ER membrane as a more informative location for this intracellular pool. This does not establish a freely soluble ER-lumen 93β584 species or natural 252-product localization. Proposed replacements: endoplasmic reticulum membrane |
| GO:0005783 endoplasmic reticulum | IEA GO_REF:0000044 | MODIFY | Summary: Specify the membrane-associated ER kinase pool. Reason: SL-0095 correctly maps to endoplasmic reticulum. PMID:32149426 Figure 3 shows reversible ER redistribution of WT Fam20C-mApple during thapsigargin stress in human HepG2 cells, separately from engineered KDEL retention. Combined with the human type-II membrane topology established in PMID:34349020, this supports ER membrane for the intracellular precursor. The refinement is a cross-study synthesis, not an assertion that Figure 3 alone resolved topology or identified a natural splice product. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB-SubCell:SL-0095 SUPPORTS TRANSFER The broader ER mapping is valid. Target-specific stress localization and independently measured membrane topology justify finer location; no propagation failure is asserted. Proposed replacements: endoplasmic reticulum membrane |
| GO:0005788 endoplasmic reticulum lumen | TAS Reactome:R-HSA-8952289 | UNDECIDED | Summary: Keep the processed ER-lumen model unresolved. Reason: Reactome explicitly models native Q8IXL6 residues 93β584 as entity8952276 in the ER lumen, within the substrate-bound kinase complex of R-HSA-8952289. That is positive curated location evidence. However, PMID:34349020 distinguishes an intracellular membrane-bound precursor from a secreted Asp93 mature form, while PMID:32149426 establishes WT stress-dependent ER residence without identifying a free soluble 93β584 ER pool. The large lumen-facing domain and real ER phosphorylation activity do not by themselves establish that specific processed molecular entity. Retain UNDECIDED pending reconciliation of the modeled processing state and compartment; this is not a claim that FAM20C lacks ER activity. |
| GO:0005794 Golgi apparatus | IDA GO_REF:0000052 | MODIFY | Summary: Refine human Golgi imaging with independent membrane topology. Reason: HPA019823 ICC/IF is supported and reports Golgi staining in A-431, RPE1 and U-2OS cells; the actual saved images and metadata were inspected. The 89-residue antigen maps to displayed FAM20C186β274 and is absent from the natural 252-residue product. HPA imaging supports the Golgi compartment, while PMID:34349020 supplies the independent membrane extraction/topology evidence for Golgi membrane. This refinement does not claim membrane resolution from HPA alone or product-specific localization of all isoforms. Proposed replacements: Golgi membrane |
| GO:0005794 Golgi apparatus | IDA PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | MODIFY | Summary: The intracellular Golgi pool is more specifically membrane-associated. Reason: PMID:22582013 Figure 1C demonstrates GM130 colocalization of FLAG-FAM20C in human HeLa cells but does not itself resolve membrane topology. Independent PMID:34349020 Figure 1 supplies extraction and protease-protection evidence for an integral membrane precursor. Refine to Golgi membrane using this corroboration; retain the separate soluble processed extracellular pool rather than implying exclusive membrane residence. Proposed replacements: Golgi membrane |
| GO:0005794 Golgi apparatus | IEA GO_REF:0000107 | MODIFY | Summary: Specify the Golgi membrane rather than the whole organelle. Reason: The mouse Q5MJS3 source has an actual Golgi IDA to PMID:22900076: tagged murine Fam20c and endogenous osteoblast Fam20c overlap Golgi markers. The inherited organelle location is sound. Human FAM20C membrane extraction and fluorescence-protease-protection in PMID:34349020 justify Golgi membrane as a more informative target term without retroactively assigning those topology measurements to the mouse experiment. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB:Q5MJS3 SUPPORTS TRANSFER Mouse Golgi localization directly supports transfer. Independent human type-II membrane evidence sharpens the target location; no misplaced donor or target loss is inferred. Proposed replacements: Golgi membrane |
| GO:0006468 protein phosphorylation | IDA PMID:23754375 Crystal structure of the Golgi casein kinase. | MODIFY | Summary: FAM20C performs serine and threonine phosphorylation of protein substrates. Reason: PMID:23754375 human kinase/mutant assays and Supporting Information Figure S5 establish FAM20C as the phosphate-transfer performer rather than a consumed substrate. Refine broad protein phosphorylation to peptidyl-serine phosphorylation and peptidyl-threonine phosphorylation. Current typed is_a/part_of closures place both under GO:0006468 and neither under the other. Serine is the dominant experimentally supported specificity; the independent human IGFBP1 main Figure 5F result supplies the threonine facet, without claiming that every source separately tested threonine or that weak peptide-library tyrosine activity establishes a central tyrosine-phosphorylation process. Proposed replacements: peptidyl-serine phosphorylation peptidyl-threonine phosphorylation |
| GO:0006468 protein phosphorylation | IDA PMID:25789606 A secretory kinase complex regulates extracellular protein p... | MODIFY | Summary: FAM20C performs serine and threonine phosphorylation of protein substrates. Reason: PMID:25789606 enamel-protein, site-mutant and pseudokinase-activation experiments establish FAM20C as the phosphate-transfer performer rather than a consumed substrate. Refine broad protein phosphorylation to peptidyl-serine phosphorylation and peptidyl-threonine phosphorylation. Current typed is_a/part_of closures place both under GO:0006468 and neither under the other. Serine is the dominant experimentally supported specificity; the independent human IGFBP1 main Figure 5F result supplies the threonine facet, without claiming that every source separately tested threonine or that weak peptide-library tyrosine activity establishes a central tyrosine-phosphorylation process. Proposed replacements: peptidyl-serine phosphorylation peptidyl-threonine phosphorylation |
| GO:0006468 protein phosphorylation | IDA PMID:26091039 A Single Kinase Generates the Majority of the Secreted Phosp... | MODIFY | Summary: FAM20C performs serine and threonine phosphorylation of protein substrates. Reason: PMID:26091039 Figures 4 and 5 protein-substrate phosphorylation experiments establish FAM20C as the phosphate-transfer performer rather than a consumed substrate. Refine broad protein phosphorylation to peptidyl-serine phosphorylation and peptidyl-threonine phosphorylation. Current typed is_a/part_of closures place both under GO:0006468 and neither under the other. Serine is the dominant experimentally supported specificity; the independent human IGFBP1 main Figure 5F result supplies the threonine facet, without claiming that every source separately tested threonine or that weak peptide-library tyrosine activity establishes a central tyrosine-phosphorylation process. Proposed replacements: peptidyl-serine phosphorylation peptidyl-threonine phosphorylation Supporting Evidence: |
| GO:0030145 manganese ion binding | IDA PMID:23754375 Crystal structure of the Golgi casein kinase. | ACCEPT | Summary: Manganese supports the FAM20C catalytic metal site. Reason: PMID:23754375 Supporting Information Methods and Figure S5A/B directly measure human FAM20C kinase kinetics with Mn-ATP and Mg-ATP; the actual Mn/ADP crystal and metal coordinates are from C. elegans ceFam20 (Figure 2 and Figure S6), with conserved-site human mutational corroboration. This supports manganese binding at the catalytic cofactor site without claiming a human manganese-bound crystal, exclusive Mn use or measured cellular occupancy. The Figure S5D captionβs Mn wording differs from its Mg-labeled panel; the human A/B panels are unambiguous. Supporting Evidence: |
| GO:0031214 biomineral tissue development | IMP PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | ACCEPT | Summary: FAM20C contributes directly to biomineral matrix regulation by phosphorylation. Reason: PMID:22582013 Figure 3 shows direct phosphorylation of SIBLING matrix proteins by FAM20C and tests Raine-syndrome mutants for kinase activity/secretion. The performed contribution is matrix-protein phosphorylation, not merely disease association or being required for mineralization. These assays support the retained developmental role while not equating every clinical variant with one biochemical defect or establishing natural 252-product function. Supporting Evidence: |
| GO:0031214 biomineral tissue development | IMP PMID:25789606 A secretory kinase complex regulates extracellular protein p... | MODIFY | Summary: Specify the enamel-matrix contribution demonstrated by this source. Reason: PMID:25789606 Figure 1 and Methods demonstrate phosphorylation of human ENAM, AMTN and AMELX preparations by active FAM20C, with D478A and ENAM serine-site controls. Mouse ameloblast-like-cell knockout experiments separately support AMBN phosphorylation. Figures 4 and 5 investigate FAM20A stimulation and assembly; the performed FAM20C step remains phosphate transfer to enamel matrix proteins. Refine this source-specific broad biomineral-tissue annotation to enamel mineralization, a current is_a descendant of GO:0031214. The independently supported bone/dentin matrix role from PMID:22582013 remains broad in its own row and the core. Proposed replacements: enamel mineralization Supporting Evidence: |
| GO:0043687 post-translational protein modification | TAS Reactome:R-HSA-597592 | MODIFY | Summary: Specify the serine and threonine phosphorylation steps performed by FAM20C. Reason: The parent R-HSA-597592 reaches R-HSA-8957275 and R-HSA-8952289, where native human Q8IXL6 belongs to the kinase catalyst. PMID:26091039 Results and Figure 5E/F verify WT-dependent protein-substrate phosphorylation including Thr193, with inactive-mutant controls. Refine the parent process to peptidyl-serine and peptidyl-threonine phosphorylation, the directly performed chemistry. Current typed closures place both below protein phosphorylation and neither below the other; no typed ancestry from GO:0043687 is asserted. Proposed replacements: peptidyl-serine phosphorylation peptidyl-threonine phosphorylation |
| GO:0070062 extracellular exosome | HDA PMID:19056867 Large-scale proteomics and phosphoproteomics of urinary exos... | KEEP AS NON CORE | Summary: FAM20C was detected in a human urinary exosome-enriched fraction. Reason: PMID:19056867 Methods describe pooled urine from eight healthy adults, differential centrifugation, a 200,000 g pellet with DTT treatment and proteomic identification. The actual author-hosted target table lists FAM20C/NP_064608 with one unique peptide and two spectra. This positive target identification supports the imported fraction-localization observation. It does not resolve whether FAM20C is inside or attached to vesicles, establish a natural-product-specific peptide, or prove a functional role in exosome biogenesis. Possible soluble-protein carryover is not demonstrated contamination and is not used to remove the annotation. Supporting Evidence: |
| GO:0070062 extracellular exosome | HDA PMID:19199708 Proteomic analysis of human parotid gland exosomes by multid... | KEEP AS NON CORE | Summary: FAM20C was detected in a human parotid exosome-enriched preparation. Reason: PMID:19199708 Methods use stimulated ductal parotid saliva from one healthy adult, differential centrifugation and a 200,000 g exosome-rich pellet. Actual Supplemental Table 2 lists FAM20C/IPI00470607.3, 584 residues, five sequences and seven spectra; Supplemental Table 1 contains the corresponding peptide records. Identification used IPI 3.24, DTASelect grouping, a minimum of two peptides and an author-estimated protein FDR below 5%. This positive fraction finding is retained as non-core without asserting target-specific immunogold localization, vesicle topology, exosome activity or equivalence to the natural 252-residue product. Supporting Evidence: |
| GO:0070166 enamel mineralization | IBA GO_REF:0000033 | ACCEPT | Summary: Retain conserved enamel mineralization. Reason: The positive GO:0070166 IBD at PTN000278721 lies on the exact human FAM20C path. Its mouse MGI:2136853 donor has an IMP annotation to PMID:22732358; the inspected original describes a Fam20c-disrupted mouse with severe enamel hypomineralization. This genetic evidence is complemented by direct phosphorylation of human enamel proteins, including ENAM, by human FAM20C in PMID:25789606 Figure 1 and the kinase Methods. FAM20C performs the protein-phosphorylation step rather than being merely a mineral substrate. The mouse phenotype is not a test of the human 252-residue product. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN000278721 SUPPORTS TRANSFER Positive enamel-mineralization IBD, target path and experimental mouse donor traced. Actual mouse phenotype plus human enamel-protein phosphorylation support conserved participation; no donor-count criterion is used. |
| GO:0106310 protein serine kinase activity | EXP PMID:22582013 Secreted kinase phosphorylates extracellular proteins that r... | ACCEPT | Summary: FAM20C directly catalyzes protein serine phosphorylation. Reason: PMID:22582013 Figures 1β3 compare purified human WT/D478A kinase, serine-containing peptides and recombinant protein substrates; Ξ»-phosphatase and cellular depletion provide additional controls. The inaccessible 2012 supplemental target table is not claimed as read. Retain this specific chemistry alongside the measured serine/threonine breadth; current GO:0106310 is not an is_a descendant of GO:0004674, so the terms are not treated as interchangeable. Supporting Evidence: |
| GO:0106310 protein serine kinase activity | EXP PMID:23754375 Crystal structure of the Golgi casein kinase. | ACCEPT | Summary: FAM20C directly catalyzes protein serine phosphorylation. Reason: PMID:23754375 human Ξ²28β40 assays, active-site mutants and Supporting Information Figure S5 directly support serine phosphorylation; the crystal is from the worm ortholog and is not presented as a human structure. Retain this specific chemistry alongside the measured serine/threonine breadth; current GO:0106310 is not an is_a descendant of GO:0004674, so the terms are not treated as interchangeable. Supporting Evidence: |
| GO:0106310 protein serine kinase activity | EXP PMID:25789606 A secretory kinase complex regulates extracellular protein p... | ACCEPT | Summary: FAM20C directly catalyzes protein serine phosphorylation. Reason: PMID:25789606 Figure 1 and supplements directly map human enamel-protein serine phosphorylation with substrate-site and D478A controls; purified FAM20A activates FAM20C rather than supplying its catalytic reaction. Retain this specific chemistry alongside the measured serine/threonine breadth; current GO:0106310 is not an is_a descendant of GO:0004674, so the terms are not treated as interchangeable. Supporting Evidence: |
| GO:0106310 protein serine kinase activity | EXP PMID:29858230 Secretory kinase Fam20C tunes endoplasmic reticulum redox st... | ACCEPT | Summary: FAM20C directly catalyzes protein serine phosphorylation. Reason: PMID:29858230 Figures 2β4 use purified human FAM20C 141β578, WT/D478A and Ero1Ξ± S145A/S145E controls to establish serine phosphorylation. Ero1Ξ± performs the downstream oxidoreductase chemistry. Retain this specific chemistry alongside the measured serine/threonine breadth; current GO:0106310 is not an is_a descendant of GO:0004674, so the terms are not treated as interchangeable. Supporting Evidence: |
| GO:0106310 protein serine kinase activity | EXP PMID:32149426 Phosphorylation switches protein disulfide isomerase activit... | ACCEPT | Summary: FAM20C directly catalyzes protein serine phosphorylation. Reason: PMID:32149426 Figures 1, 2 and 4 use WT/D478A kinase, PDI Ser357 identification/mutants, purified-protein reactions and human knockout/rescue. FAM20C performs phosphorylation; the substrate PDI supplies its own chaperone/redox activities. Retain this specific chemistry alongside the measured serine/threonine breadth; current GO:0106310 is not an is_a descendant of GO:0004674, so the terms are not treated as interchangeable. Supporting Evidence: |
| GO:0106310 protein serine kinase activity | EXP PMID:34349020 Proteolytic processing of secretory pathway kinase Fam20C by... | ACCEPT | Summary: FAM20C directly catalyzes protein serine phosphorylation. Reason: PMID:34349020 Figures 5 and 6 compare human recombinant kinase preparations, OPN/Ero1Ξ± substrates and S106A/D478A controls. Ser106 autophosphorylation is distinct from S1P- or HRV3C-mediated proteolytic processing. Retain this specific chemistry alongside the measured serine/threonine breadth; current GO:0106310 is not an is_a descendant of GO:0004674, so the terms are not treated as interchangeable. Supporting Evidence: |
| GO:0106310 protein serine kinase activity | IEA GO_REF:0000116 | ACCEPT | Summary: Retain the explicit protein-serine kinase reaction. Reason: RHEA:17989 specifies ATP-dependent phosphorylation of L-seryl residues in a protein and maps exactly to GO:0106310. Human WT-versus-D478A protein-substrate assays establish serine phosphorylation. Retain this specific reaction alongside the broader Ser/Thr rows: human substrate experiments in PMID:26091039 and the curated RHEA:46608 reaction support additional threonine capacity. No claim is made that GO:0106310 is currently a typed descendant of GO:0004674. Supporting Evidence: GO_REF:0000116 file:human/FAM20C/FAM20C-uniprot.txt |
Loading supporting contentβ¦
Download this section (compressed HTML)Q: Is the natural 252-residue FAM20C product detectably expressed as a stable protein, and how do its localization and kinase activity compare with the 584-residue precursor and processed 93β584 form?
Q: Which FAM20C processing state performs phosphorylation in the ER during stress, and is a soluble mature 93β584 ER pool present independently of the membrane precursor and secreted mature protein?
Q: How much does threonine phosphorylation contribute to endogenous FAM20C function, and which protein-context features permit it despite the strong serine preference and negative corresponding short-peptide assays?
Experiment: Combine isoform-resolved RNA analysis with targeted proteomics for discriminating peptides and matched expression of sequence-verified 584- and 252-residue products. Measure protein stability, membrane association, secretion and phosphorylation of a controlled protein substrate, normalizing to recovered protein and including catalytic-dead controls. Keep the processed 93β584 preparation as a separate comparator.
Hypothesis: If the natural shorter transcript produces a stable protein, its missing N-terminal region may change trafficking or catalysis relative to the longer precursor; this must be tested independently of engineered deletion constructs.
Experiment: In human cells, combine a validated endogenous C-terminal tag, pulse-chase analysis, membrane extraction, protease protection and N-terminal mass spectrometry before, during and after ER stress. Resolve membrane, ER-soluble and extracellular fractions and measure PDI phosphorylation in parallel. Use compartment and fraction-integrity controls; establish that the tag preserves processing and avoid relying on KDEL retention as evidence for a native pool.
Hypothesis: Stress-associated ER phosphorylation may be performed by a membrane-associated precursor, a soluble processed species, or both under different conditions.
Experiment: Quantify IGFBP1 Thr193 and Ser194 independently with validated phosphopeptide standards in matched human cells lacking FAM20C and rescued with WT or D478A protein. Control substrate abundance and secretion, distinguish singly and multiply phosphorylated peptides, and compare full protein with the corresponding short-peptide substrates. Expand to additional threonine sites only after target-specific localization is secure.
Hypothesis: If protein-context-dependent threonine phosphorylation contributes physiologically, endogenous site occupancy should respond to loss of FAM20C and rescue by active kinase.
Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)