model: gomodel:67e5e74400000148 title: Arabidopsis MRG domain proteins bridge two histone modifications taxon: NCBITaxon:3702 status: COMPLETE summary: >- A four-activity chromatin model built almost entirely on Xu et al. 2014 (PMID:25183522), describing how Arabidopsis MRG proteins couple H3K36 trimethylation to histone H4 acetylation at flowering-time genes. The chain is: ASHH3/SDG7 (AT2G44150, histone H3K36 dimethyltransferase, has input histone H3.3 HTR4) "directly provides input for" ASHH2/SDG8 (AT1G77300, histone H3K36 trimethyltransferase, has input HTR4), which "directly provides input for" the MRG2 (AT1G02740) histone H3K36me3 reader activity (has input HTR4 and AT2G45160), which "directly positively regulates" a histone H4K5 acetyltransferase activity (has input histone H4, AT2G28740). Every activity occurs in chromatin and is part_of transcription elongation-coupled chromatin remodeling (GO:0140673). The core story (SDG8-deposited H3K36me3 is read by the MRG2 chromodomain, and MRG1/2 recruit the MYST acetyltransferases HAM1/HAM2 to raise H4K5 acetylation at FT and FLC) is well supported by the cited paper. Two problems dominate. (1) Wrong gene product: the H4K5 acetyltransferase is enabled by AT2G45160, which is SCARECROW-LIKE 27 (UniProtKB:Q7XJM8), a GRAS-family transcription regulator whose alternative symbol is also HAM1 (HAIRY MERISTEM 1). The MYST acetyltransferase HAM1 studied in all three cited papers is AT5G64610 (UniProtKB:Q9FLF7; HAM2 is AT5G09740). The same wrong locus is reused as a has input of the MRG2 reader. This looks like a symbol collision when the entity was picked. (2) Process: GO:0140673 is defined as restoring chromatin after RNA polymerase II passes during elongation, to suppress cryptic intragenic initiation. That is the role of animal and yeast MRG15/Eaf3. The cited paper argues that the Arabidopsis MRGs diverged from it and instead raise acetylation at the promoter and 5' region to elevate transcription. A positive regulation of transcription / flowering-gene expression process would fit the paper's claim better. No gene product in this model has a gene review in this repository (genes/ARATH has none of MRG1, MRG2, SDG8, SDG7, HAM1, HAM2 or SCL27). The downstream targets FT, FLC and SOC1 are reviewed, but they are not represented in the model. notes: >- This is the only production GO-CAM with NCBITaxon:3702 in the GO production index (gocam-models.json, 2036 models, checked 2026-10-14). It is also the only model with TAIR/AGI gene products in gocam-gps.json (49,416 model entries), and the only model returned by the GO API taxon endpoint (api.geneontology.org/api/taxon/NCBITaxon:3702/models). Molecular-function associations carry no evidence of their own. The ECO:0000314 evidence sits on enabled_by, occurs_in, part_of and the edges. Some of it supports genetic (mutant) observations, for which ECO:0000315 would be more accurate. activity_reviews: - activity_id: gomodel:67e5e74400000148/67e5e74400000163 gene_product: AGI_LocusCode:AT2G44150 molecular_function: GO:0140954 biological_process: GO:0140673 cellular_component: GO:0000785 verdict: UNCERTAIN consistency: NO_GENE_REVIEW qc_flags: - DIRECT_VS_INDIRECT_CAUSAL supporting_text: >- The double mutant with another close homologue of SDG8, sdg7 sdg8 fully abolishes both the di- and tri-methylation marks at H3K36 (Figure 4A). evidence: - source_id: PMID:25183522 statement: >- SDG7 (ASHH3) contributes to the residual H3K36me2 left in sdg8. The evidence is genetic: immunostaining of mutant nuclei. supporting_text: >- mutation of the Arabidopsis H3K36 methyltransferase SDG8 leads to a near complete abolishment of H3K36me3 and a slight reduction in H3K36me2 - source_id: PMID:25183522 statement: >- Model-wide process issue: in Arabidopsis H3K36me3 peaks at the 5' end of the coding region rather than following elongating Pol II, so elongation-coupled chromatin remodeling (GO:0140673) is a poor fit. supporting_text: >- the H3K36me3 level in Arabidopsis peaks at the 5′-end of the coding region notes: >- ASHH3/SDG7 (UniProtKB:Q945S8) is a SET-domain H3K36 methyltransferase homologue of SDG8. The cited paper supports a contribution to H3K36me2 in vivo only through the sdg7 sdg8 double mutant. It does not include an enzyme assay, so ECO:0000314 overstates the evidence type (ECO:0000315 / IMP would be accurate). The MF H3K36 dimethyltransferase is plausible but rests on genetic inference here. The edge "directly provides input for" ASHH2 encodes a strict sequential model (SDG7 makes me2, SDG8 converts it to me3). The paper does not test that order. The sdg8 single mutant already lowers H3K36me2, so SDG8 likely also makes some me2 itself, and the direct-input edge over-states what the genetics show. Has input HTR4 (histone H3.3, AT4G40030) is a reasonable stand-in for histone H3 as substrate. The verdict rests on the evidence type and the over-strict input edge; the model-wide process issue (GO:0140673) is recorded in the summary and does not drive this activity's verdict. - activity_id: gomodel:67e5e74400000148/67e5e74400000172 gene_product: AGI_LocusCode:AT2G45160 molecular_function: GO:0043995 biological_process: GO:0140673 cellular_component: GO:0000785 verdict: WRONG consistency: NO_GENE_REVIEW supporting_text: >- Specific HATs acetylate histone H4K5 (HAM1, HAM2), H4K12 (HAG2), and H3K14 (HAG1) evidence: - source_id: PMID:17877703 statement: The MYST-family acetyltransferases HAM1 and HAM2 acetylate H4K5 in vitro. supporting_text: >- Specific HATs acetylate histone H4K5 (HAM1, HAM2), H4K12 (HAG2), and H3K14 (HAG1) - source_id: PMID:23273925 statement: >- HAM1/HAM2 are MYST histone acetyltransferases that act on H4K5, with in vivo support. supporting_text: >- two MYST histone acetyltransferases HAM1 and HAM2 work redundantly to acetylate histone H4 lysine 5 (H4K5ace) in vitro. - source_id: PMID:25183522 statement: MRG1/2 recruit the H4-specific acetyltransferases HAM1/HAM2. supporting_text: >- interact with the histone H4-specific acetyltransferases (HAM1 and HAM2) - source_id: PMID:25183522 statement: >- Model-wide process issue: in Arabidopsis H3K36me3 peaks at the 5' end of the coding region rather than following elongating Pol II, so elongation-coupled chromatin remodeling (GO:0140673) is a poor fit. supporting_text: >- the H3K36me3 level in Arabidopsis peaks at the 5′-end of the coding region notes: >- The molecular function (histone H4K5 acetyltransferase, GO:0043995), its input (histone H4, AT2G28740) and chromatin location are correct for the MYST acetyltransferase HAM1. The enabler is the wrong gene, though. AT2G45160 is SCARECROW-LIKE 27 (UniProtKB:Q7XJM8, a GRAS-family transcription regulator, also called HAM1/HAIRY MERISTEM 1 and LOM1). It has no acetyltransferase domain and none of the three cited papers studies it. The MYST HAM1 of PMID:17877703, PMID:23273925 and PMID:25183522 is AT5G64610 (UniProtKB:Q9FLF7, "Histone acetyltransferase of the MYST family 1"; UniProt cites PMID:17877703 and PMID:23273925 for its H4K5 activity). HAM2 (AT5G09740) acts redundantly and could be added as a second enabler or a separate activity. The enabled_by should be changed to AT5G64610. As cached, this annoton asserts an acetyltransferase activity for a GRAS protein and would propagate a false MF to SCL27. It is also the terminal node, with no causal output to transcription of FT/FLC, so the model's output on flowering-gene expression is not represented (the paper's readout). The verdict rests on the wrong enabler; the model-wide process issue (GO:0140673) is recorded in the summary and does not drive it. source_defects: - defect_type: WRONG_GENE_PRODUCT resource: GO_CAM record: gomodel:67e5e74400000148/67e5e74400000172 slot: enabled_by requested_change: REPOINT asserted_value: AGI_LocusCode:AT2G45160 corrected_value: AGI_LocusCode:AT5G64610 corrected_label: HAM1, MYST histone acetyltransferase 1 (UniProtKB:Q9FLF7) cause: SYMBOL_COLLISION confidence: CONFIRMED explanation: >- The histone H4K5 acetyltransferase activity (GO:0043995) is enabled_by AT2G45160, which is SCARECROW-LIKE 27 (UniProtKB:Q7XJM8). SCL27 is a GRAS transcription regulator that also carries the alias HAM1 (HAIRY MERISTEM 1). The acetyltransferase in all three cited papers (PMID:17877703, PMID:23273925, PMID:25183522) is the MYST protein HAM1, AT5G64610 (UniProtKB:Q9FLF7; UniProt gene names HAM1 HAG4 At5g64610). The enabler should be AT5G64610. HAM2 (AT5G09740) is redundant and could be added. supported_by: - reference_id: PMID:23273925 supporting_text: >- two MYST histone acetyltransferases HAM1 and HAM2 work redundantly to acetylate histone H4 lysine 5 (H4K5ace) in vitro. - activity_id: gomodel:67e5e74400000148/67e5e74400000149 gene_product: AGI_LocusCode:AT1G02740 molecular_function: GO:0140003 biological_process: GO:0140673 cellular_component: GO:0000785 verdict: UNCERTAIN consistency: NO_GENE_REVIEW qc_flags: - HAS_INPUT_MISUSE supporting_text: >- MRG1/2 binds to H3K36me3 with their chromodomain and interact with the histone H4-specific acetyltransferases (HAM1 and HAM2) evidence: - source_id: PMID:25183522 statement: MRG2 reads H3K36me3 in vivo; binding depends on SDG8-made H3K36me3. supporting_text: >- These data indicate that H3K36me3 is essential for MRG protein binding in vivo. - source_id: PMID:25183522 statement: MRG2 physically engages HAM1/2 (PLA, BiFC, Co-IP). supporting_text: >- The interaction between MRG1/2 and MYST family proteins HAM1/2, characterised by an in situ PLA assay, BiFC assay and Co-IP analysis - source_id: PMID:25183522 statement: >- Model-wide process issue: in Arabidopsis H3K36me3 peaks at the 5' end of the coding region rather than following elongating Pol II, so elongation-coupled chromatin remodeling (GO:0140673) is a poor fit. supporting_text: >- the H3K36me3 level in Arabidopsis peaks at the 5′-end of the coding region notes: >- MRG2 (UniProtKB:Q4V3E2) as a histone H3K36me3 reader (GO:0140003) is well supported: there is chromodomain peptide binding in vitro, and the PLA signal is lost in sdg8. Has input HTR4 (histone H3.3) is appropriate. The second has input, AT2G45160 (SCL27), is wrong twice over. It is the wrong locus (the partner is MYST HAM1, AT5G64610), and a recruited effector enzyme is not the input of a reader activity: the "directly positively regulates" edge to the acetyltransferase already captures the recruitment. The MRG2 -> HAM edge (RO:0002629) is defensible: the paper shows direct physical interaction and loss of H4K5ac at FT in mrg1 mrg2. MRG1 (AT4G37280), which acts redundantly with MRG2, is not modelled. Process GO:0140673 (elongation-coupled suppression of cryptic transcription) contradicts the paper's conclusion that plant MRGs diverged from that role and act at promoter/5' regions to raise FT and FLC expression. That process issue is model-wide and recorded in the summary; this activity's UNCERTAIN verdict rests on the has-input misuse. source_defects: - defect_type: WRONG_GENE_PRODUCT resource: GO_CAM record: gomodel:67e5e74400000148/67e5e74400000149 slot: has_input requested_change: DELETE_VALUE asserted_value: AGI_LocusCode:AT2G45160 cause: SYMBOL_COLLISION confidence: CONFIRMED explanation: >- The MRG2 H3K36me3 reader activity has AT2G45160 (SCL27, a GRAS protein with the alias HAM1) as a has_input. The partner the paper reports is the MYST acetyltransferase HAM1, AT5G64610 (UniProtKB:Q9FLF7). Separately, a recruited effector enzyme is not the input of a reader activity. The MRG2-to-HAM1 relationship is already captured by the causal edge to the acetyltransferase activity, so the has_input should be removed rather than re-pointed. supported_by: - reference_id: PMID:25183522 supporting_text: >- interact with the histone H4-specific acetyltransferases (HAM1 and HAM2) - activity_id: gomodel:67e5e74400000148/67e5e74400000156 gene_product: AGI_LocusCode:AT1G77300 molecular_function: GO:0140955 biological_process: GO:0140673 cellular_component: GO:0000785 verdict: OK consistency: NO_GENE_REVIEW supporting_text: >- mutation of the Arabidopsis H3K36 methyltransferase SDG8 leads to a near complete abolishment of H3K36me3 and a slight reduction in H3K36me2 evidence: - source_id: PMID:25183522 statement: SDG8 (ASHH2) is required for H3K36me3 and for MRG2 chromatin engagement. supporting_text: >- MRG2 binds to H3K36me3 specifically in vivo because the binding of MRG2 is completely abolished in the sdg8 mutant - source_id: PMID:25183522 statement: >- Model-wide process issue: in Arabidopsis H3K36me3 peaks at the 5' end of the coding region rather than following elongating Pol II, so elongation-coupled chromatin remodeling (GO:0140673) is a poor fit. supporting_text: >- the H3K36me3 level in Arabidopsis peaks at the 5′-end of the coding region notes: >- ASHH2/SDG8/EFS (UniProtKB:Q2LAE1) is the main Arabidopsis H3K36 trimethyltransferase, so the MF is correct. In this model it rests on mutant immunostaining (a genetic observation tagged ECO:0000314). The "directly provides input for" edge to the MRG2 reader is appropriate: the H3K36me3 product of SDG8 is what the MRG2 chromodomain binds, and MRG2 chromatin binding is lost in sdg8. Has input HTR4 (histone H3.3) is acceptable. The model-wide process issue (GO:0140673) is recorded in the summary and does not drive this activity's verdict, so the MF and edge, both well supported, leave it OK. references: - id: PMID:25183522 title: Arabidopsis MRG domain proteins bridge two histone modifications to elevate expression of flowering genes. - id: PMID:17877703 title: 'In vitro specificities of Arabidopsis co-activator histone acetyltransferases: implications for histone hyperacetylation in gene activation.' - id: PMID:23273925 title: Requirement of histone acetyltransferases HAM1 and HAM2 for epigenetic modification of FLC in regulating flowering in Arabidopsis.