Ins in HeLa cells, we assayed whether endogenous mouse proteins coimmunoprecipitate with mouse GFP-TRPML1 in the absence of Ca2+. We immunoprecipitated GFP-TRPML1 or Derlin-1-GFP from stable RAW264.7 clones that express these fusion proteins; in these clones GFP-TRPML1 localizes predominantly to late endosomes/lysosomes and Derlin-1-GFP 
localizes predominantly to the endoplasmic reticulum with some Derlin-1 GFP localizing in endosomes [19,34]. The full-length (93 kD) GFP-TRPML1 is the main form of this protein in RAW264.7 cells, although we could also detect some higher molecular weight isoforms (Fig. 3).Proteins That Interact with TRPMLDerlin-1-GFP 
migrated at the predicted size of 55 kD by SDSPAGE. We assayed commercial antibodies that had been raised against STOML1, Rac2, and ERGIC; only the mouse anti-STOML1 antibody recognized a protein of the predicted size in RAW264.7 lysates. We therefore confined our analysis to STOML1. STOML1 co-immunoprecipitated preferentially with GFPTRPML1 (Fig. 3). In contrast, Sorting Nexin 2 (SNX2) coimmunoprecipitated preferentially with Derlin-1-GFP, consistent with our previous study showing that SNX2 associates with Derlin-1 (Fig. 3) [31]. The negative control Destrin, an actinbinding protein, did not co-immunoprecipitate with either GFPTRPML1 or Derlin-1-GFP (Fig. 3) [39]. Thus endogenous STOML1 interacts with TRPML1 (Table 1).protein is misfolded in S. cerevisiae since expression of the homologous P5KT1(BAA13031) protein is toxic to yeast. We directly tested potential interactors isolated by the Immunoprecipitation/Mass Spectrometry approach in the SUYTH screen. Through this analysis, we confirmed that TRPML1 associates with NDKA, Rac2, Cdc42, and NP9 by SU-YTH (Fig. 4; Table 1). TRPML1 also associated with Rac1 but not RhoG. Although all of these proteins were missed by the initial SU-YTH Fexinidazole site HIV-RT inhibitor 1 web screen, their confirmation in this directed approach strongly suggests that they are strong candidate TRPML1 interactors. Several other proteins did not interact with TRPML1 by the SU-YTH assay. Of note, our failure to detect a STOML1TRPML1 SU-YTH interaction is likely due to NubG-STOML1 localizing to vacuoles of yeast where there is little TRPML1-CubLexA-VP16 (Fig. 4).Testing TRPML1 Interactors by SU-YTHThe limited overlap in potential interactors using our two approaches was initially surprising, but not completely inexplicable given the potential challenges of the SU-YTH screen. Although this approach allows full-length proteins to be assayed, the disadvantage of SU-YTH is that it represents an artificial system where proteins are assayed in S. cerevisiae where they are not normally expressed and perhaps at levels and locations that are non-native. For example, based on interactions with Fur4-NubI (plasma membrane) and Ost1-NubI (endoplasmic reticulum), TRPML1-Cub-LexA-VP16 predominantly localizes to the endoplasmic reticulum (Fig. 4). Thus, proteins that associate with TRPML1 but that do not localize to the endoplasmic reticulum may not be captured in this assay (false-negatives). Finally, some exogenous proteins, such as full-length P5KT1(BAA13031), are toxic to S. cerevisiae and cannot be studied using the SU-YTH approach. Of the candidate interactors identified by SU-YTH screens, we confirmed that TRPML1 still associated with their full-length versions by SU-YTH, with the exception of P5KT1(NP_032872) (Fig. 4; Table 1). While we do not know why P5KT1(NP_032872) failed to associate with TRPML1 by SU-YTH, w.Ins in HeLa cells, we assayed whether endogenous mouse proteins coimmunoprecipitate with mouse GFP-TRPML1 in the absence of Ca2+. We immunoprecipitated GFP-TRPML1 or Derlin-1-GFP from stable RAW264.7 clones that express these fusion proteins; in these clones GFP-TRPML1 localizes predominantly to late endosomes/lysosomes and Derlin-1-GFP localizes predominantly to the endoplasmic reticulum with some Derlin-1 GFP localizing in endosomes [19,34]. The full-length (93 kD) GFP-TRPML1 is the main form of this protein in RAW264.7 cells, although we could also detect some higher molecular weight isoforms (Fig. 3).Proteins That Interact with TRPMLDerlin-1-GFP migrated at the predicted size of 55 kD by SDSPAGE. We assayed commercial antibodies that had been raised against STOML1, Rac2, and ERGIC; only the mouse anti-STOML1 antibody recognized a protein of the predicted size in RAW264.7 lysates. We therefore confined our analysis to STOML1. STOML1 co-immunoprecipitated preferentially with GFPTRPML1 (Fig. 3). In contrast, Sorting Nexin 2 (SNX2) coimmunoprecipitated preferentially with Derlin-1-GFP, consistent with our previous study showing that SNX2 associates with Derlin-1 (Fig. 3) [31]. The negative control Destrin, an actinbinding protein, did not co-immunoprecipitate with either GFPTRPML1 or Derlin-1-GFP (Fig. 3) [39]. Thus endogenous STOML1 interacts with TRPML1 (Table 1).protein is misfolded in S. cerevisiae since expression of the homologous P5KT1(BAA13031) protein is toxic to yeast. We directly tested potential interactors isolated by the Immunoprecipitation/Mass Spectrometry approach in the SUYTH screen. Through this analysis, we confirmed that TRPML1 associates with NDKA, Rac2, Cdc42, and NP9 by SU-YTH (Fig. 4; Table 1). TRPML1 also associated with Rac1 but not RhoG. Although all of these proteins were missed by the initial SU-YTH screen, their confirmation in this directed approach strongly suggests that they are strong candidate TRPML1 interactors. Several other proteins did not interact with TRPML1 by the SU-YTH assay. Of note, our failure to detect a STOML1TRPML1 SU-YTH interaction is likely due to NubG-STOML1 localizing to vacuoles of yeast where there is little TRPML1-CubLexA-VP16 (Fig. 4).Testing TRPML1 Interactors by SU-YTHThe limited overlap in potential interactors using our two approaches was initially surprising, but not completely inexplicable given the potential challenges of the SU-YTH screen. Although this approach allows full-length proteins to be assayed, the disadvantage of SU-YTH is that it represents an artificial system where proteins are assayed in S. cerevisiae where they are not normally expressed and perhaps at levels and locations that are non-native. For example, based on interactions with Fur4-NubI (plasma membrane) and Ost1-NubI (endoplasmic reticulum), TRPML1-Cub-LexA-VP16 predominantly localizes to the endoplasmic reticulum (Fig. 4). Thus, proteins that associate with TRPML1 but that do not localize to the endoplasmic reticulum may not be captured in this assay (false-negatives). Finally, some exogenous proteins, such as full-length P5KT1(BAA13031), are toxic to S. cerevisiae and cannot be studied using the SU-YTH approach. Of the candidate interactors identified by SU-YTH screens, we confirmed that TRPML1 still associated with their full-length versions by SU-YTH, with the exception of P5KT1(NP_032872) (Fig. 4; Table 1). While we do not know why P5KT1(NP_032872) failed to associate with TRPML1 by SU-YTH, w.
