Feline Lindhout

VAP–SCRN1 interaction regulates dynamic endoplasmic reticulum remodeling and presynaptic function 115 4 SCRN1 in SV cycling by conducting the Syt uptake assay in neurons depleted from SCRN1. SCRN1 knockdown also showed a marked decrease (~40%) in relative Syt internalization compared to control cells, thereby phenocopying the effect of VAP knockdown (Fig 1J,K). The presynaptic phenotype in SCRN1 knockdown neurons was rescued by expressing wild-type GFP-SCRN1 (Fig 1J, K). Together, these results illustrate that SCRN1 depletion, similarly to VAP depletion, results in impaired SV cycling. SCRN1 does not exhibit autolytic protease activity To better understand the molecular function of VAP-associated protein SCRN1, we next tested whether its conserved proteolytic domain could be involved. Like all SCRN family members, SCRN1 contains a C69 protease domain and therefore belongs to the N-terminal nucleophile (Ntn) aminohydrolases superfamily (Pei & Grishin, 2003). Proteolytic activity in this superfamily relies on autolytic cleavage of the auto-inhibitory N-terminal of the precursor protein by the mature protein (Fig S2A). This cleavage occurs right before the catalytic site of the protein, which is a cysteine residue in the SCRN family. Sequence alignment of the SCRN proteins revealed that the position of the predicted catalytic cysteine including the flanking residues is shared in SCRN2 and SCRN3, but not in SCRN1 (Fig S2B). We analyzed N-terminal SCRN cleavage by conducting Western blotting of lysates from HEK293T cells expressing wild-type GFP-SCRN1, GFP-SCRN2, or GFP-SCRN3 (Fig S2C). In lysates of GFP-SCRN2 and GFP-SCRN3 expressing cells, we identified a low molecular weight band corresponding to the predicted size of GFP fused to the N-terminal cleavage product. Conversely, this cleavage product was not observed in lysates of GFP- SCRN1 expressing cells. Moreover, mutant SCRN1, SCRN2, and SCRN3 constructs in which the predicted catalytic cysteine was replaced by a non-catalytic alanine residue also did not show a cleavage product (Fig S2C). These data suggest that SCRN1, unlike its family members SCRN2 and SCRN3, does not exhibit autolytic protease activity. SCRN1 is recruited to VAP at the ER membrane To further examine the function of VAP-associated protein SCRN1, we next assessed whether the subcellular localization of SCRN1 could be controlled by VAP. This was addressed by conducting co-expression experiments of GFP-SCRN1 and HA-VAPA or HA-VAPB in cultured neurons and COS7 cells. In COS7 cells, the ER structures are relatively less compact and easier to visualize than in neurons. GFP-SCRN1 expression alone in neurons or COS7 cells showed a diffuse cytoplasmic distribution, which only partly coincided with ER structures (Fig 2A and B). In neurons, co-expression of GFP-SCRN1 with either HA-VAPA or HA-VAPB resulted in the formation of dense VAP/SCRN1-positive clustersat neurites (Fig2C).COS7cellsco-expressingGFP-SCRN1andHA-VAPAorHA-VAPB showed marked differences in SCRN1 localization, as it induced abundant SCRN1 recruitment to VAP at the ER membrane (Fig 2D) This observation suggests that enhancing the number of VAPs at the ER membrane allows for increased SCRN1 binding, presumably because it decreases the competition with other FFAT(-