Extracellular domain shedding of NOTCH3 during endocytosis associated with heterogeneity between different CADASIL mutant activation mechanisms.

BACKGROUND: Mutations in NOTCH3 cause CADASIL, a dominantly inherited condition, linked to recurrent stroke and vascular dementia and associated with accumulation of the ECD of NOTCH3. The latter has a toxic effect on VSMCs. Misregulated signalling may also play a role in disease progression. ECD detachment is an obligatory step in NOTCH3 activation, but some CADASIL mutants prevent ligand-induced activation and so ligand interactions are not a common underlying requirement. Here we investigated whether basal NOTCH3 endocytosis that is associated with ligand-independent activation mechanisms can be source of ECD shedding in CADASIL mutants. METHODS: We used transient transfection of hTERT-RPE1 cells to express WT, R90C, C212Y and C455R mutant NOTCH3 constructs. Internalisation of NOTCH3 was followed using a pulse-chase endocytic uptake assay after surface NOTCH3 labelling of live cells. Immunolocalisation of NOTCH3 ECD and ICD was used to define the subcellular localisation of expressed NOTCH3 in the secretory and endocytic pathway of transfected cells, and endogenous NOTCH3 in MCF7 cells and VSMCs derived from human ES cells. To investigate NOTCH3 signalling we used a luciferase reporter assay under control of a NOTCH-responsive reporter element. RESULTS: Both WT and CADASIL NOTCH3 proteins are endocytosed before ECD shedding and then undergo dissociation and independent trafficking of the ECD and ICD in the endosome. The relative amount of ICD compared to ECD that colocalised with endosomal markers increases as NOTCH3 progresses through the endosomal trafficking pathway from early endosome to lysosome. The R90C mutant showed earlier separation of ECD compared to WT or other CADASIL mutants tested. All WT and mutant constructs activated downstream signalling when expressed in hTERT-RPE1 cells, and these basal signalling levels were not affected by the C455R mutation which removes ligand-activated signalling. R90C showed distinctly different requirements for activation being less sensitive to metalloprotease inhibition and more sensitive to inhibition of the lysosomal protein TRPML. CONCLUSIONS: Basal NOTCH3 endocytosis and signalling is a potential source of ECD shedding and accumulation in CADASIL. Different mechanisms may apply to different CADASIL mutants and understanding the variety of mechanisms by which NOTCH3 signalling and ECD shedding occur will inform new targeted approaches to treatments of small vessel disease. Tuning NOTCH3 activity through modulation of the endocytic pathway may offer better tolerated approaches than direct targeting of NOTCH3 signalling.