Membrane proteins. We expressed a TRPML1-Cub-LexA-VP16 fusion protein in yeast and monitored its interaction with Fur4 (plasma membrane localized) Ost1 (endoplasmic reticulum localized). When these test proteins were fused to NubG, which reduces its affinity for Cub, no interaction was detected, as was the case for the unfused NubG control (see below). In contrast, TRPML1-Cub-LexA-VP16 interacted with both Fur4-NubI and Ost1-NubI fusions, as expected. Intriguingly, the TRPML1-Fur4 interaction was stronger than the TRPML1Ost1 interaction, suggesting that more TRPML1-Cub-LexAVP16 protein remains in the endoplasmic reticulum than is secreted to reach the plasma membrane [30].. We then transformed NubG-fused mouse cDNA libraries into yeast expressing TRPML1-Cub-LexA-VP16 and assayed for growth on selective media. We identified several potential TRPML1 interactors, which included Lysosomal-Associated Protein Transmembrane 4B that was previously identified as a TRPML1 interactor (Table S3) [29]. However, there were only a few candidate proteins that were identified using both the Immunoprecipitation/Mass Spectrometry and the SU-YTH approaches (highlighted in blue, red, and green in MedChemExpress BMS5 Tables S2 and S3). These included the same glyceraldehyde 3-phosphate dehydrogenase, homologous though not identical cadherins that are encoded by different genes, and homologous though not the same sodium channel alpha subunit encoded by two different genes. SU-YTH may fail to detect bona fide interactors that are unable to associate with TRPML1 in yeast or may also yield falsepositives that only associate in the context of this assay. We therefore carried out additional assays to probe the effectiveness of Immunoprecipitation/Mass Spectrometry and SU-YTH for the purpose of identifying TRPML1 interacting proteins.Figure 1. Cloning Strategy for Analyzing Candidate Interactors. Shown is a schematic of the GTWY cloning strategy for constructing the epitope-fused candidate proteins in the proper expression Indolactam V web vectors. doi:10.1371/journal.pone.0056780.gto candidate proteins. Cells were fixed for 1 hour in 1 formaldehyde/1XPBS, washed three times with 1XPBS, and mounted in Slowfade mounting medium (Invitrogen) on slides for viewing. Confocal images 15857111 were taken with a Nikon PCM 2000, using HeNe 543 excitation for the red dye and argon 488 for the green dye.ImmunofluorescenceRAW264.7 macrophages stably expressing GFP-TRPML1 were transfected with plasmids expressing V5 fused to candidate proteins. Immunofluorescence was carried out as previously described [19]. Primary antibodies used were Rabbit anti-GFP (Abcam) and Mouse anti-V5 (Abcam).Determining Co-Localization with GFP-TRPMLThe percent co-localization is defined as the number of GFPTRPML1-stained structures that co-localized with TagRFP(S158T)-fused or V5-fused structures divided by the total number of GFP-TRPML1-stained structures in a section and multiplied by 100. The graphs show the average from sections of at least eight different cells, with at least twenty GFP-TRPML1stained structures per cell.Results Identification of TRPML1 Interactors by Immunoprecipitation and Mass SpectrometryOur first approach for identifying TRPML1 interactors was immunoprecipitation combined with Mass Spectrometry. We immunoprecipitated GFP-TRPML1 or Derlin-1-GFP (an integral membrane protein found in the endoplasmic reticulum and endosomes) from stably expressing RAW264.7 clones in the absence of Ca2+ and then used Mass Spectrometry to i.Membrane proteins. We expressed a TRPML1-Cub-LexA-VP16 fusion protein in yeast and monitored its interaction with Fur4 (plasma membrane localized) Ost1 (endoplasmic reticulum localized). When these test proteins were fused to NubG, which reduces its affinity for Cub, no interaction was detected, as was the case for the unfused NubG control (see below). In contrast, TRPML1-Cub-LexA-VP16 interacted with both Fur4-NubI and Ost1-NubI fusions, as expected. Intriguingly, the TRPML1-Fur4 interaction was stronger than the TRPML1Ost1 interaction, suggesting that more TRPML1-Cub-LexAVP16 protein remains in the endoplasmic reticulum than is secreted to reach the plasma membrane [30].. We then transformed NubG-fused mouse cDNA libraries into yeast expressing TRPML1-Cub-LexA-VP16 and assayed for growth on selective media. We identified several potential TRPML1 interactors, which included Lysosomal-Associated Protein Transmembrane 4B that was previously identified as a TRPML1 interactor (Table S3) [29]. However, there were only a few candidate proteins that were identified using both the Immunoprecipitation/Mass Spectrometry and the SU-YTH approaches (highlighted in blue, red, and green in Tables S2 and S3). These included the same glyceraldehyde 3-phosphate dehydrogenase, homologous though not identical cadherins that are encoded by different genes, and homologous though not the same sodium channel alpha subunit encoded by two different genes. SU-YTH may fail to detect bona fide interactors that are unable to associate with TRPML1 in yeast or may also yield falsepositives that only associate in the context of this assay. We therefore carried out additional assays to probe the effectiveness of Immunoprecipitation/Mass Spectrometry and SU-YTH for the purpose of identifying TRPML1 interacting proteins.Figure 1. Cloning Strategy for Analyzing Candidate Interactors. Shown is a schematic of the GTWY cloning strategy for constructing the epitope-fused candidate proteins in the proper expression vectors. doi:10.1371/journal.pone.0056780.gto candidate proteins. Cells were fixed for 1 hour in 1 formaldehyde/1XPBS, washed three times with 1XPBS, and mounted in Slowfade mounting medium (Invitrogen) on slides for viewing. Confocal images 15857111 were taken with a Nikon PCM 2000, using HeNe 543 excitation for the red dye and argon 488 for the green dye.ImmunofluorescenceRAW264.7 macrophages stably expressing GFP-TRPML1 were transfected with plasmids expressing V5 fused to candidate proteins. Immunofluorescence was carried out as previously described [19]. Primary antibodies used were Rabbit anti-GFP (Abcam) and Mouse anti-V5 (Abcam).Determining Co-Localization with GFP-TRPMLThe percent co-localization is defined as the number of GFPTRPML1-stained structures that co-localized with TagRFP(S158T)-fused or V5-fused structures divided by the total number of GFP-TRPML1-stained structures in a section and multiplied by 100. The graphs show the average from sections of at least eight different cells, with at least twenty GFP-TRPML1stained structures per cell.Results Identification of TRPML1 Interactors by Immunoprecipitation and Mass SpectrometryOur first approach for identifying TRPML1 interactors was immunoprecipitation combined with Mass Spectrometry. We immunoprecipitated GFP-TRPML1 or Derlin-1-GFP (an integral membrane protein found in the endoplasmic reticulum and endosomes) from stably expressing RAW264.7 clones in the absence of Ca2+ and then used Mass Spectrometry to i.