id: A0A9L0RGD6 gene_symbol: SHLD2 taxon: id: NCBITaxon:9796 label: Equus caballus status: COMPLETE description: Nuclear and DNA-break localization are supported by conserved SHLD2 recruitment biology. The core OB1/OB2 end-protection module supports the NHEJ and anti-HR predictions despite a likely C-terminal gene-model artifact; actin localization and isotype switching remain unresolved. source_documents: - projects/PROTNLM_EVALUATION/mammal-benchmark/horse40-predictions.csv - projects/PROTNLM_EVALUATION/mammal-benchmark/predictions.jsonl.gz - projects/PROTNLM_EVALUATION/mammal-benchmark/paired-sequences/SHLD2.json - genes/HORSE/SHLD2/SHLD2-hypotheses/horse40-nhej-promotion/openscientist.md references: - id: PMID:29789392 title: FAM35A associates with REV7 and modulates DNA damage responses of normal and BRCA1-defective cells. full_text_unavailable: false findings: [] - id: file:HORSE/SHLD2/SHLD2-uniprot.txt title: HORSE/SHLD2/SHLD2-uniprot.txt findings: [] - id: file:HORSE/SHLD2/SHLD2-bioinformatics/RESULTS.md title: HORSE/SHLD2/SHLD2-bioinformatics/RESULTS.md findings: [] - id: file:human/SHLD2/SHLD2-hpa.xml title: human/SHLD2/SHLD2-hpa.xml findings: [] - id: PMID:29656893 title: DNA Repair Network Analysis Reveals Shieldin as a Key Regulator of NHEJ and PARP Inhibitor Sensitivity. full_text_unavailable: false findings: [] - id: PMID:30022168 title: The shieldin complex mediates 53BP1-dependent DNA repair. full_text_unavailable: false findings: [] - id: file:HORSE/SHLD2/SHLD2-hypotheses/horse40-nhej-promotion/openscientist.md title: 'OpenScientist hypothesis run: SHLD2 NHEJ promotion' publication_type: DEEP_RESEARCH findings: - statement: OpenScientist found that the horse C-terminal frameshift leaves the SHLD2 OB1/OB2 execution module intact. supporting_text: the frameshift leaves OB1 and OB2 — the domains that execute ssDNA binding and resection blockade — intact predictions: - source_method: ProtNLM2 source_version: UniProt API snapshot 2026-09-08 predicted_term: id: GO:2001034 label: positive regulation of double-strand break repair via nonhomologous end joining predicted_term_type: GO_BP review: assessment: COR confidence_score: 2 summary: Mammalian SHLD2 experiments support this repair-process role, and the OpenScientist audit resolved the selected horse sequence as a justified orthology transfer. A likely C-terminal frameshift corrupts OB3, but it leaves the conserved SHLD3/REV7 recruitment segment and the OB1/OB2 ssDNA-binding execution module intact, so the horse model retains the apparatus that restricts DNA end resection and promotes NHEJ. This is a homology-based ISS/IBA transfer, not direct horse evidence. supported_by: - reference_id: PMID:29789392 supporting_text: Expression of FAM35A isoform 1 in 293T cells (Fig 4D) restored NHEJ to normal levels (Fig 4E). - &id003 reference_id: file:HORSE/SHLD2/SHLD2-uniprot.txt supporting_text: DR InterPro; IPR029715; FAM35A. - &id001 reference_id: file:HORSE/SHLD2/SHLD2-bioinformatics/RESULTS.md supporting_text: share 70.2% identity among 792 paired residues. - &id004 reference_id: PMID:30022168 supporting_text: Importantly, both mutants localized to DSB sites (ED Fig 10cd) and interacted with the other members of the Shieldin complex (ED Fig 10e). Therefore, the SHLD2-m1 and SHLD2S mutants are defective in suppressing HR. - &id005 reference_id: file:HORSE/SHLD2/SHLD2-bioinformatics/RESULTS.md supporting_text: Beyond that point the horse C-terminus is substantially divergent and gapped. This is not a complete match to either human isoform. - reference_id: file:HORSE/SHLD2/SHLD2-hypotheses/horse40-nhej-promotion/openscientist.md supporting_text: the frameshift leaves OB1 and OB2 — the domains that execute ssDNA binding and resection blockade — intact - source_method: ProtNLM2 source_version: UniProt API snapshot 2026-09-08 predicted_term: id: GO:0015629 label: actin cytoskeleton predicted_term_type: GO_CC review: assessment: UNC confidence_score: 1 summary: HPA reports an additional actin-filament pool of human SHLD2 with supported antibody-based reliability, so this prediction is not contradicted merely by SHLD2 being a nuclear repair protein. However, the secondary localization is cell-line dependent and its targeting mechanism is uncharacterized; family membership and global sequence identity do not establish conservation of this particular pool in the selected horse isoform. The main nuclear localization has substantially stronger mechanistic support. A conserved targeting determinant or independent localization evidence is needed to resolve this exact transfer. supported_by: - &id002 reference_id: file:human/SHLD2/SHLD2-hpa.xml supporting_text: Mainly localized to the nucleoplasm. In addition localized to the actin filaments. - *id001 - source_method: ProtNLM2 source_version: UniProt API snapshot 2026-09-08 predicted_term: id: GO:0035861 label: site of double-strand break predicted_term_type: GO_CC review: assessment: COR confidence_score: 2 summary: The selected horse protein belongs to the SHLD2 family and retains its OB-fold architecture; its alignment with human SHLD2 covers 94.9% of the human sequence with 70.2% identity among paired residues. Human SHLD2 recruitment to double-strand breaks and conserved shieldin domains support this location in horse. This is a supported transfer new relative to the frozen horse GOA, not a claim that the model training set lacked mammalian SHLD2 annotations. supported_by: - &id006 reference_id: PMID:29656893 supporting_text: "Recruitment of shieldin to DSBs, via the \nATM-RNF8-RNF168-53BP1-RIF1 axis, promotes\ \ NHEJ-dependent repair of \nintrachromosomal breaks, immunoglobulin class-switch recombination\ \ (CSR), and \nfusion of unprotected telomeres." - reference_id: file:HORSE/SHLD2/SHLD2-uniprot.txt supporting_text: /note="Shieldin complex subunit 2 first OB fold" - *id001 - source_method: ProtNLM2 source_version: UniProt API snapshot 2026-09-08 predicted_term: id: GO:0005654 label: nucleoplasm predicted_term_type: GO_CC review: assessment: COR confidence_score: 2 summary: The selected horse protein belongs to the SHLD2 family and retains its OB-fold architecture; its alignment with human SHLD2 covers 94.9% of the human sequence with 70.2% identity among paired residues. The human HPA nucleoplasmic signal agrees with nuclear damage-focus studies, supporting conservation of the principal nuclear compartment. This is a supported transfer new relative to the frozen horse GOA, not a claim that the model training set lacked mammalian SHLD2 annotations. supported_by: - *id002 - reference_id: PMID:29789392 supporting_text: FAM35A re-localized in damaged cell nuclei - *id001 - source_method: ProtNLM2 source_version: UniProt API snapshot 2026-09-08 predicted_term: id: GO:2000042 label: negative regulation of double-strand break repair via homologous recombination predicted_term_type: GO_BP review: assessment: COR confidence_score: 2 summary: Mammalian SHLD2 experiments support this repair-process role, and the OpenScientist audit resolved the selected horse sequence as a justified orthology transfer. The likely OB3 frameshift should be verified against a corrected transcript, but the OB1/OB2 ssDNA-binding execution module and N-terminal recruitment segment are conserved, supporting the anti-resection mechanism that negatively regulates double-strand break repair by homologous recombination. supported_by: - reference_id: PMID:29789392 supporting_text: However, the BRCA1‐mutant cells formed damage‐dependent nuclear RAD51 foci following FAM35A depletion, suggesting that 5′ end resection was more active in the absence of FAM35A (Fig 5B). - *id003 - *id001 - *id004 - *id005 - reference_id: file:HORSE/SHLD2/SHLD2-hypotheses/horse40-nhej-promotion/openscientist.md supporting_text: Human ortholog carries this (IDA); mechanistically the more *direct* description of resection blockade - source_method: ProtNLM2 source_version: UniProt API snapshot 2026-09-08 predicted_term: id: GO:0045830 label: positive regulation of isotype switching predicted_term_type: GO_BP review: assessment: UNC confidence_score: 1 summary: Mammalian SHLD2 and shieldin experiments connect this repair pathway to immunoglobulin class-switch recombination, but GO:0045830 is a tissue-restricted downstream immunological process. The OpenScientist audit supports transfer of the conserved OB1/OB2-dependent NHEJ and anti-HR repair module to horse SHLD2, but it did not evaluate GO:0045830 or provide horse B-cell or CSR evidence. This more specific process therefore remains unresolved until the conserved repair activity is tied to the relevant horse immune context. supported_by: - *id006 - *id003 - *id001 - *id004 - *id005