DLL3 is a single-pass type I membrane glycoprotein of the Delta/Serrate/LAG-2 (DSL) family, with an extracellular DSL domain and six EGF-like repeats. Unlike the canonical ligands DLL1 and DLL4, DLL3 does not activate Notch on neighbouring cells. Endogenous DLL3 is retained in intracellular membranes, chiefly the Golgi apparatus and the late endocytic compartment, where it binds Notch1 and routes newly synthesised receptor to lysosomal degradation before it reaches the cell surface, acting as a cis-inhibitor of Notch signalling. This activity operates in the presomitic mesoderm, and biallelic loss-of-function mutations in humans cause autosomal recessive spondylocostal dysostosis, a vertebral and rib segmentation disorder. In high-grade pulmonary neuroendocrine tumours DLL3 is strongly overexpressed and aberrantly trafficked to the plasma membrane, which is the basis for DLL3-directed antibody-drug conjugate and T-cell-engager therapeutics.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0001501 skeletal system development | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Orthology transfer from mouse Dll3 captures the axial skeletal phenotype, which is the developmental consequence of DLL3's cis-inhibitory activity in the presomitic mesoderm. Reason: Mouse Dll3 (pudgy) and human DLL3 loss both disrupt vertebral and rib segmentation, so the transferred term is correct. It is a downstream developmental outcome rather than the molecular core function, hence non-core. Supporting Evidence: PMID:10742114 we have cloned and sequenced human DLL3 to evaluate it as a candidate gene for SD and identified mutations in three autosomal recessive SD families |
| GO:0001501 skeletal system development | IMP PMID:10742114 Mutations in the human delta homologue, DLL3, cause axial sk... | KEEP AS NON CORE | Summary: Human biallelic DLL3 mutations cause spondylocostal dysostosis, with vertebral and rib malsegmentation. Reason: Direct human genetic evidence supports a requirement in axial skeletal development. The phenotype is a developmental consequence of defective somite segmentation, so the term is kept as a real but non-core role. Supporting Evidence: PMID:10742114 Spondylocostal dysostosis (SD, MIM 277300) is a group of vertebral malsegmentation syndromes with reduced stature resulting from axial skeletal defects. PMID:10742114 identified mutations in three autosomal recessive SD families |
| GO:0005112 Notch binding | IBA GO_REF:0000033 | ACCEPT | Summary: Notch binding through the DSL domain is retained in DLL3 and is its core molecular activity, even though the binding is in cis and inhibitory rather than trans-activating. Reason: The inherited DSL-family assertion holds for DLL3 specifically. Chapman et al. show direct interaction of Dll3 with Notch1 in the late endocytic compartment, and the UniProt record states that the DSL domain mediates receptor binding. Notch binding describes the interaction without asserting activation, so the divergent signalling output of DLL3 is not an argument against this term. Supporting Evidence: PMID:21147753 Dll3 is not presented on the surface of presomitic mesoderm (PSM) cells in vivo, but instead interacts with Notch1 in the late endocytic compartment |
| GO:0005112 Notch binding | NAS PMID:10742114 Mutations in the human delta homologue, DLL3, cause axial sk... | ACCEPT | Summary: Notch binding is the molecular activity of the DSL domain of DLL3. Reason: The author statement in the spondylocostal dysostosis cloning paper is corroborated by direct interaction data for Dll3 and Notch1, so the claim stands on independent evidence. Supporting Evidence: PMID:21147753 instead interacts with Notch1 in the late endocytic compartment |
| GO:0005509 calcium ion binding | IEA GO_REF:0000002 | KEEP AS NON CORE | Summary: The EGF-like calcium-binding repeat signature supports calcium binding by the ectodomain. Reason: DLL3 carries tandem EGF-like repeats of the calcium-binding class, matched by both InterPro and SMART signatures, so the domain-based inference is legitimate. Calcium binding is a structural property of these repeats rather than the signalling function of the protein, so it is retained as non-core. No DLL3-specific calcium measurement was found. Supporting Evidence: file:human/DLL3/DLL3-uniprot.txt InterPro; IPR001881; EGF-like_Ca-bd_dom. file:human/DLL3/DLL3-uniprot.txt SMART; SM00179; EGF_CA; 5. |
| GO:0005886 plasma membrane | IBA GO_REF:0000033 | MODIFY | Summary: DLL3 is not active at the plasma membrane. Endogenous protein is retained in the Golgi and the late endocytic compartment, where it engages Notch1. Reason: The ancestral DSL-ligand node is correctly placed for Delta and Serrate ligands that signal in trans from the cell surface, but there is target-specific evidence that DLL3 diverged in trafficking. Geffers et al. localise endogenous DLL3 predominantly to the Golgi while DLL1 is on the surface of the same cells, and Chapman et al. show Dll3 is not presented at the presomitic mesoderm cell surface. Replace with Golgi membrane and late endosome, the compartments in which the protein resides and in which the Notch1 interaction occurs. Surface DLL3 is seen in neuroendocrine tumours, which is a pathological overexpression state rather than the site of normal activity. Propagation Review Root cause: TERM SCOPING PROBLEM Failure modes: COMPARTMENT OR COMPLEX MISMATCH FUNCTIONAL DIVERGENCE Sources checked: PANTHER:PTN001170801 SUPPORTS SOURCE BUT NOT TARGET The ancestral DSL-ligand node asserts surface activity, which holds for DLL1, DLL4 and the invertebrate ligands but not for the divergent DLL3 branch. Proposed replacements: Golgi membrane late endosome Supporting Evidence: PMID:17664336 In contrast to endogenous DLL1 located on the surface of presomitic mesoderm cells, we find endogenous DLL3 predominantly in the Golgi apparatus. PMID:21147753 Dll3 is not presented on the surface of presomitic mesoderm (PSM) cells in vivo |
| GO:0005886 plasma membrane | IEA GO_REF:0000107 | MODIFY | Summary: The mouse-ortholog transfer assigns a plasma-membrane location that the direct mouse data contradict. Reason: The donor is mouse Dll3, and the mouse experiments that address its localization place endogenous DLL3 in the Golgi, not at the cell surface. Replace with Golgi membrane and late endosome, consistent with the compartments in which DLL3 is detected and in which it binds Notch1. Propagation Review Root cause: SOURCE BAD Failure modes: COMPARTMENT OR COMPLEX MISMATCH Sources checked: UniProtKB:O88516 · mouse Dll3 SOURCE BAD The donor carries a plasma-membrane assignment, but the mouse localization experiments place endogenous DLL3 in the Golgi rather than at the surface. Proposed replacements: Golgi membrane late endosome Supporting Evidence: PMID:17664336 we find endogenous DLL3 predominantly in the Golgi apparatus PMID:21147753 interacts with Notch1 in the late endocytic compartment |
| GO:0007219 Notch signaling pathway | IBA GO_REF:0000033 | MODIFY | Summary: DLL3 acts on Notch signalling, but as a cis-inhibitor rather than as a pathway ligand that transmits the signal. Reason: The inherited assertion places DLL3 in the pathway as a canonical activating ligand. Both studies that tested DLL3 directly show it does not activate Notch in trans, in mice or in Drosophila wing discs, and instead suppresses receptor output from within the same cell. This is a sign correction, so the replacement is negative regulation of Notch signaling pathway. Propagation Review Root cause: TERM SCOPING PROBLEM Failure modes: REGULATORY SIGN INVERSION FUNCTIONAL DIVERGENCE Sources checked: PANTHER:PTN001170801 SUPPORTS SOURCE BUT NOT TARGET Pathway membership as an activating ligand is right for the ancestral node and for DLL1 and DLL4, while DLL3 inherited the binding module but inverted the signalling output. Proposed replacements: negative regulation of Notch signaling pathway Supporting Evidence: PMID:17664336 Dll3 does not compensate for Dll1; DLL1 activates Notch in Drosophila wing discs, but DLL3 does not. PMID:21147753 DLL3 is a divergent member of the DSL family of Notch ligands that does not activate signalling in adjacent cells, but instead inhibits signalling when expressed in the same cell as the Notch receptor. |
| GO:0007219 Notch signaling pathway | IEA GO_REF:0000002 | MODIFY | Summary: The DSL-domain InterPro mapping assigns pathway membership generically, whereas for DLL3 the directional term is the accurate one. Reason: As for the phylogenetic row, the DSL domain is present and does bind Notch, but the effect of DLL3 on the pathway is inhibitory and acts in cis. Replace with negative regulation of Notch signaling pathway. Proposed replacements: negative regulation of Notch signaling pathway Supporting Evidence: PMID:21147753 instead inhibits signalling when expressed in the same cell as the Notch receptor |
| GO:0016020 membrane | IEA GO_REF:0000120 | MODIFY | Summary: DLL3 is an integral membrane protein, but the specific membrane is known. Reason: The UniProt subcellular-location mapping is correct but less precise than the evidence allows. Endogenous DLL3 is a Golgi membrane protein, so the specific descendant term should be used. Proposed replacements: Golgi membrane Supporting Evidence: PMID:17664336 we find endogenous DLL3 predominantly in the Golgi apparatus |
| GO:0016020 membrane | NAS PMID:10742114 Mutations in the human delta homologue, DLL3, cause axial sk... | MODIFY | Summary: Author statement of membrane localization, refinable to the Golgi membrane. Reason: The transmembrane topology is established by the UniProt feature table, and the compartment was subsequently determined to be the Golgi for endogenous protein. Use the specific descendant term. Proposed replacements: Golgi membrane Supporting Evidence: file:human/DLL3/DLL3-uniprot.txt Single-pass type I membrane protein PMID:17664336 we find endogenous DLL3 predominantly in the Golgi apparatus |
| GO:0045746 negative regulation of Notch signaling pathway | IBA GO_REF:0000033 | ACCEPT | Summary: Cis-inhibition of Notch is the core biological role of DLL3. Reason: The PAINT node placement is supported by direct DLL3 evidence. The protein interacts with Notch1 intracellularly and promotes lysosomal degradation of newly synthesised receptor before surface presentation, lowering Notch output in the expressing cell. Supporting Evidence: PMID:21147753 with Dll3 targeting newly synthesized Notch1 for lysosomal degradation prior to post-translational processing and cell surface presentation of the receptor |
| GO:0160108 animal gross anatomical part developmental process | IEA GO_REF:0000117 | TOO GENERAL | Summary: A high-level developmental grouping term that says nothing specific about DLL3. Reason: Skeletal system development, retained in this review, is a descendant of this term and already entails it. The claim is true but uninformative. Supporting Evidence: PMID:10742114 identified mutations in three autosomal recessive SD families |
| GO:0009986 cell surface | IDA PMID:26311731 A DLL3-targeted antibody-drug conjugate eradicates high-grad... | NEW | Summary: In high-grade pulmonary neuroendocrine tumours, overexpressed DLL3 reaches the cell surface, where it is accessible to antibodies on non-permeabilised cells. Reason: Flow cytometry of non-permeabilised cells with DLL3-specific antibodies establishes a surface pool of DLL3, and the same study detects surface DLL3 on SCLC and LCNEC tumour cells. This is a real location of the protein, distinct from the Golgi and late-endosome pool of normal cells, and it is the location exploited by DLL3-directed therapeutics. Cell surface rather than plasma membrane is proposed, because the evidence is antibody accessibility on intact cells. Supporting Evidence: PMID:26311731 establishing that DLL3 protein can localize to the cell surface and is not necessarily confined to the Golgi PMID:26311731 DLL3 protein was detectable at the surface of SCLC and LCNEC tumor cells |
Loading supporting content…
Download this section (compressed HTML)Q: What retains DLL3 in the Golgi and late endosomes, and which sorting determinants in the transmembrane and cytoplasmic regions are responsible?
Q: Does the surface pool of DLL3 seen in neuroendocrine tumours have any signalling activity, in cis or in trans, or is it functionally inert cargo that happens to be accessible?
Q: Are the six EGF-like repeats of DLL3 genuinely calcium-loaded, and does calcium occupancy affect the Notch1 interaction?
Experiment: Pulse-chase metabolic labelling of Notch1 in cells with and without DLL3 expression, combined with surface biotinylation and lysosomal inhibitors, to measure the fraction of nascent Notch1 that reaches the plasma membrane versus the lysosome.
Hypothesis: DLL3 cis-inhibits Notch by diverting newly synthesised Notch1 from the secretory pathway to lysosomes.
Experiment: Swap the DLL3 transmembrane and cytoplasmic regions onto DLL1 and vice versa, then quantify steady-state surface versus Golgi pools by surface biotinylation and immunofluorescence in presomitic-mesoderm-like cells.
Hypothesis: Golgi retention of DLL3 is encoded in its transmembrane and juxtamembrane sequences.
Experiment: Titrate DLL3 expression in a non-neuroendocrine cell line and in SCLC lines and measure the surface-to-total ratio by flow cytometry on non-permeabilised cells, to test whether the surface fraction appears only above a threshold of total expression.
Hypothesis: Surface DLL3 in SCLC arises from saturation of a retention mechanism rather than from a tumour-specific trafficking pathway.
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)