Portal de Periódicos da CAPES

Rapid Mapping of Interactions between Human SNX-BAR Proteins Measured In Vitro by AlphaScreen and Single-molecule Spectroscopy

2014; Elsevier BV; Volume: 13; Issue: 9 Linguagem: Inglês

10.1074/mcp.m113.037275

1535-9484

Emma Sierecki, Loes M. Stevers, Nichole Giles, Mark Polinkovsky, Mehdi Moustaqil, Sergey Mureev, Wayne A. Johnston, Mareike Dahmer‐Heath, Dubravka Škalamera, Thomas J. Gonda, Brian Gabrielli, Brett M. Collins, Kirill Alexandrov, Yann Gambin,

Biotin and Related Studies

Protein dimerization and oligomerization is commonly used by nature to increase the structural and functional complexity of proteins. Regulated protein assembly is essential to transfer information in signaling, transcriptional, and membrane trafficking events. Here we show that a combination of cell-free protein expression, a proximity based interaction assay (AlphaScreen), and single-molecule fluorescence allow rapid mapping of homo- and hetero-oligomerization of proteins. We have applied this approach to the family of BAR domain-containing sorting nexin (SNX-BAR) proteins, which are essential regulators of membrane trafficking and remodeling in all eukaryotes. Dimerization of BAR domains is essential for creating a concave structure capable of sensing and inducing membrane curvature. We have systematically mapped 144 pairwise interactions between the human SNX-BAR proteins and generated an interaction matrix of preferred dimerization partners for each family member. We find that while nine SNX-BAR proteins are able to form homo-dimers, several including the retromer-associated SNX1, SNX2, and SNX5 require heteromeric interactions for dimerization. SNX2, SNX4, SNX6, and SNX8 show a promiscuous ability to bind other SNX-BAR proteins and we also observe a novel interaction with the SNX3 protein which lacks the BAR domain structure. Protein dimerization and oligomerization is commonly used by nature to increase the structural and functional complexity of proteins. Regulated protein assembly is essential to transfer information in signaling, transcriptional, and membrane trafficking events. Here we show that a combination of cell-free protein expression, a proximity based interaction assay (AlphaScreen), and single-molecule fluorescence allow rapid mapping of homo- and hetero-oligomerization of proteins. We have applied this approach to the family of BAR domain-containing sorting nexin (SNX-BAR) proteins, which are essential regulators of membrane trafficking and remodeling in all eukaryotes. Dimerization of BAR domains is essential for creating a concave structure capable of sensing and inducing membrane curvature. We have systematically mapped 144 pairwise interactions between the human SNX-BAR proteins and generated an interaction matrix of preferred dimerization partners for each family member. We find that while nine SNX-BAR proteins are able to form homo-dimers, several including the retromer-associated SNX1, SNX2, and SNX5 require heteromeric interactions for dimerization. SNX2, SNX4, SNX6, and SNX8 show a promiscuous ability to bind other SNX-BAR proteins and we also observe a novel interaction with the SNX3 protein which lacks the BAR domain structure. The last decade has witnessed an unprecedented increase in the number of identified human protein–protein interactions (PPIs) 1The abbreviations used are: PPI, protein–protein interaction; AlphaScreen, Amplified Luminescent Proximity Homogeneous Assay Screen; BAR, Bin/Amphiphysin/Rvs; C, coincidence; cps, count per second; G, intensity of GFP fluorescence; GFP, green fluorescent protein; GW, Gateway technology; IP (co-), immunoprecipitation (co-); LTE, Leishmania tarentolae extract; PX, Phox; SH3, SRC homology 3; SNX, sorting nexin; R, intensity of Cherry fluorescence; RT, room temperature. 1The abbreviations used are: PPI, protein–protein interaction; AlphaScreen, Amplified Luminescent Proximity Homogeneous Assay Screen; BAR, Bin/Amphiphysin/Rvs; C, coincidence; cps, count per second; G, intensity of GFP fluorescence; GFP, green fluorescent protein; GW, Gateway technology; IP (co-), immunoprecipitation (co-); LTE, Leishmania tarentolae extract; PX, Phox; SH3, SRC homology 3; SNX, sorting nexin; R, intensity of Cherry fluorescence; RT, room temperature., and these commonly form sophisticated interaction networks that mediate adaptive signaling responses to environmental stimuli. Validation of PPIs in general remains challenging, particularly when examining interactions within a family of proteins. Self-interaction is even more difficult to detect but can play an important physiological role. In human cells homo- or hetero-dimerization is a common feature of proteins regulating cell signaling, including tyrosine kinase receptors, G-protein coupled receptors, chemokines, cytokines, and transcription factors (1.Marianayagam N.J. Sunde M. Matthews J.M. The power of two: protein dimerization in biology.Trends Biochem. Sci. 2004; 29: 618-625Abstract Full Text Full Text PDF PubMed Scopus (424) Google Scholar). Dimerization also has the potential to increase the efficiency and specificity of enzymatic reactions, with more than 70% of enzymes able to self-associate (1.Marianayagam N.J. Sunde M. Matthews J.M. The power of two: protein dimerization in biology.Trends Biochem. Sci. 2004; 29: 618-625Abstract Full Text Full Text PDF PubMed Scopus (424) Google Scholar). Dimerization and higher-order oligomerization is also an important and widespread feature among proteins central to membrane trafficking. In this study we focus on a family of proteins, the sorting nexins (SNXs), which regulate vesicular and tubulovesicular transport in the endocytic system (2.Seet L.F. Hong W. The Phox (PX) domain proteins and membrane traffic.Biochim. Biophys. Acta. 2006; 1761: 878-896Crossref PubMed Scopus (163) Google Scholar, 3.Teasdale R.D. Collins B.M. Insights into the PX (phox-homology) domain and SNX (sorting nexin) protein families: structures, functions, and roles in disease.Biochem. J. 2012; 441: 39-59Crossref PubMed Scopus (207) Google Scholar). In particular, we examine the propensity for homo- and hetero-association among SNX proteins containing the Bin/Amphiphysin/Rvs (BAR) domain, a family of molecules that utilize dimerization to sense and drive membrane curvature. In vivo studies have shown that the SNX-BAR proteins are localized on tubular and vesicular membrane structures throughout the endocytic network (4.Cullen P.J. Phosphoinositides and the regulation of tubular-based endosomal sorting.Biochem. Soc. Trans. 2011; 39: 839-850Crossref PubMed Scopus (10) Google Scholar, 5.van Weering J.R. Verkade P. Cullen P.J. SNX-BAR proteins in phosphoinositide-mediated, tubular-based endosomal sorting.Semin. Cell Dev. Biol. 2010; 21: 371-380Crossref PubMed Scopus (127) Google Scholar). They have been shown to be involved in a growing array of endosomal sorting events (6.Wang J.T. Kerr M.C. Karunaratne S. Jeanes A. Yap A.S. Teasdale R.D. The SNX-PX-BAR family in macropinocytosis: the regulation of macropinosome formation by SNX-PX-BAR proteins.PLoS One. 2010; 5: e13763Crossref PubMed Scopus (46) Google Scholar, 7.Wassmer T. Attar N. Harterink M. van Weering J.R. Traer C.J. Oakley J. Goud B. Stephens D.J. Verkade P. Korswagen H.C. Cullen P.J. The retromer coat complex coordinates endosomal sorting and dynein-mediated transport, with carrier recognition by the trans-Golgi network.Dev. Cell. 2009; 17: 110-122Abstract Full Text Full Text PDF PubMed Scopus (219) Google Scholar, 8.van Weering J.R. Verkade P. Cullen P.J. SNX-BAR-mediated endosome tubulation is co-ordinated with endosome maturation.Traffic. 2012; 13: 94-107Crossref PubMed Scopus (113) Google Scholar, 9.Traer C.J. Rutherford A.C. Palmer K.J. Wassmer T. Oakley J. Attar N. Carlton J.G. Kremerskothen J. Stephens D.J. Cullen P.J. SNX4 coordinates endosomal sorting of TfnR with dynein-mediated transport into the endocytic recycling compartment.Nat. Cell Biol. 2007; 9: 1370-1380Crossref PubMed Scopus (210) Google Scholar) and both clathrin-dependent and independent endocytosis (10.McGough I.J. Cullen P.J. Clathrin is not required for SNX-BAR-retromer-mediated carrier formation.J. Cell Sci. 2013; 126: 45-52Crossref PubMed Scopus (20) Google Scholar). The 12 members of the SNX-BAR subfamily contain a phox-homology (PX) domain required for membrane association and a C-terminal BAR domain, composed of three α-helices that can dimerize to form a rigid banana-shaped structure (Fig. 1A). The concave surface of the dimeric BAR domains contains basic residues that mediate association with the phospholipid bilayer through electrostatic interactions. These proteins are then able to sense local bending of the membrane, and even drive membrane deformation by forming higher order helical arrays that stabilize high curvature membrane tubules and vesicles (11.Masuda M. Mochizuki N. Structural characteristics of BAR domain superfamily to sculpt the membrane.Semin. Cell Dev. Biol. 2010; 21: 391-398Crossref PubMed Scopus (68) Google Scholar, 12.Qualmann B. Koch D. Kessels M.M. Let's go bananas: revisiting the endocytic BAR code.EMBO J. 2011; 30: 3501-3515Crossref PubMed Scopus (187) Google Scholar, 13.Rao Y. Haucke V. Membrane shaping by the Bin/amphiphysin/Rvs (BAR) domain protein superfamily.Cell. Mol. Life Sci. 2011; 68: 3983-3993Crossref PubMed Scopus (83) Google Scholar). The exact structure of polymerized SNX-BAR proteins has not been elucidated. However, it is known that BAR domain-driven dimerization of the proteins is required. Both homo-dimerization and hetero-dimerization have been observed. For example, SNX9, SNX18, and SNX33 have been shown to form homo-dimers (14.Childress C. Lin Q. Yang W. Dimerization is required for SH3PX1 tyrosine phosphorylation in response to epidermal growth factor signaling and interaction with ACK2.Biochem. J. 2006; 394: 693-698Crossref PubMed Scopus (17) Google Scholar, 15.Dislich B. Than M.E. Lichtenthaler S.F. Specific amino acids in the BAR domain allow homodimerization and prevent heterodimerization of sorting nexin 33.Biochem. J. 2011; 433: 75-83Crossref PubMed Scopus (16) Google Scholar, 16.Zhang J. Zhang X. Guo Y. Xu L. Pei D. Sorting nexin 33 induces mammalian cell micronucleated phenotype and actin polymerization by interacting with Wiskott-Aldrich syndrome protein.J. Biol. Chem. 2009; 284: 21659-21669Abstract Full Text Full Text PDF PubMed Scopus (25) Google Scholar, 17.Wang Q. Kaan H.Y. Hooda R.N. Goh S.L. Sondermann H. Structure and plasticity of endophilin and sorting nexin 9.Structure. 2008; 16: 1574-1587Abstract Full Text Full Text PDF PubMed Scopus (62) Google Scholar). On the other hand, SNX2, SNX5, and SNX6, which assemble with the retromer trafficking complex, have been reported to form a series of restricted hetero-dimers that coat common endosomal membrane tubules (18.Haft C.R. de la Luz Sierra M. Barr V.A. Haft D.H. Taylor S.I. Identification of a family of sorting nexin molecules and characterization of their association with receptors.Mol. Cell. Biol. 1998; 18: 7278-7287Crossref PubMed Scopus (213) Google Scholar, 19.van Weering J.R. Sessions R.B. Traer C.J. Kloer D.P. Bhatia V.K. Stamou D. Carlsson S.R. Hurley J.H. Cullen P.J. Molecular basis for SNX-BAR-mediated assembly of distinct endosomal sorting tubules.EMBO J. 2012; 31: 4466-4480Crossref PubMed Scopus (130) Google Scholar). Although still poorly understood, one potential advantage of hetero-dimerization of the SNX-BAR proteins is that formation of different combinations may allow for fine-tuning of membrane trafficking processes either in a spatially restricted manner within individual cells, or via tissue-specific expression patterns of the different proteins. Here we present a systematic analysis of the in vitro SNX-BAR interaction landscape, using a combination of cell-free protein expression, AlphaScreen proximity assay, co-immunoprecipitation and single molecule fluorescence techniques. All possible pairs of 11 different SNX-BAR proteins (SNX1, SNX2, SNX4, SNX5, SNX6, SNX7, SNX8, SNX9, SNX30, SNX32, and SNX33; Fig. 1B), plus the PX domain-only SNX protein SNX3, were co-expressed in a Leishmania tarentolae-based cell-free expression system and analyzed for their ability to form homo- or hetero-oligomers in vitro by AlphaScreen. The homo-dimerization propensity was further validated by co-immunoprecipitation assays and by single-molecule brightness analysis. Finally, single molecule coincidence was used to analyze the stoichiometry of proteins in the SNX assemblies, confirming the presence of monomers and dimers, and revealing unexpected hetero-assemblies. The genetically encoded tags used here are enhanced GFP (GFP), mCherry (Cherry), and c-Myc (myc). The ORFs (Open Reading Frames) encoding SNX proteins were cloned into the following Gateway cloning compatible Leishmania expression vectors*: pCellFree-N-terminal-GFP, pCellFree-N-terminal-myc-Cherry, and pCellFree-C-terminal-Cherry-myc at the ARVEC Facility, UQ Diamantina Institute. These genes were sourced from the Human ORFeome collection version 1.1 and 5.1 or the Human Orfeome collaboration OCAA collection (Open Biosystems, Huntsville, AL) (20.Škalamera D. Ranall M.V. Wilson B.M. Leo P. Purdon A.S. Hyde C. Nourbakhsh E. Grimmond S.M. Barry S.C. Gabrielli B. Gonda T.J. A high-throughput platform for lentiviral overexpression screening of the human ORFeome.PLoS One. 2011; 6: e20057Crossref PubMed Scopus (37) Google Scholar). Briefly, the SNX gene in entry clones pDONOR223 or pENTR201 vectors were exchanged with the ccdB gene in the expression plasmid by LR recombination (Invitrogen, Australia). The Leishmania tarentolae Extract (LTE) was prepared as published earlier (21.Kovtun O. Mureev S. Jung W. Kubala M.H. Johnston W. Alexandrov K. Leishmania cell-free protein expression system.Methods. 2011; 55: 58-64Crossref PubMed Scopus (67) Google Scholar). A 10 μl expression volume containing 30 nm of pCellFree-N-terminal-GFP and 60 nm of pCellFree-C-terminal-Cherry-myc was incubated for 2.5 h at 27 °C for expression. Expressed proteins were mixed with NuPAGE LDS sample buffer (Invitrogen, Australia), denatured by heating at 72 °C for 3 min and resolved on NuPage Novex 4%-12% gel (Invitrogen, Australia). The proteins were detected by scanning the gel for fluorescence using ChemiDoc MP System (Bio-Rad, Australia). The AlphaScreen Assay was performed using the AlphaScreen cMyc detection kit and Proxiplate-384 Plus plates from Perkin Elmer. Expressed protein pairs were diluted in ¼ serial dilutions in Buffer A (25 mm HEPES, 50 mm NaCl). A volume of 2 μl of diluted protein was added to each well containing 12.5 μl (0.4 μg) of Acceptor beads in Buffer B (25 mm HEPES, 50 mm NaCl, 0.001% v/v casein, and 0.001% v/v Nonidet P-40). This was followed by addition of 2 μl biotin labeled GFP-nanotrap diluted in Buffer A to a final concentration of ∼2.5 nm. The proteins and Acceptor beads were incubated for 45 min at RT before addition of 2 μl (0.4 μg) of Streptavidin coated Donor Beads. This was followed by a second incubation of 45 min at RT in the dark. The AlphaScreen signal was measured using the PE Envision Multilabel Platereader according to the manufacturer's recommended settings (excitation: 680/30 nm for 0.18 s, emission: 570/100 nm after 37 ms). A binding index is defined as the average of all the signal intensities for each protein pair. For each experiment, the signal intensity from a negative control (where GFP alone is present in the solution) is subtracted. The signals are then normalized to a reference value. In this case, we chose the value obtained for the interaction between GFP-SNX4 and SNX4-Cherry-myc, as a standard in all measurements. An average of all the normalized data for both configuration of each protein pair (GFP-protein A/protein B-Cherry OR GFP-protein B/protein A-Cherry) was calculated to provide the binding index. GFP and Cherry labeled pairs of SNX proteins were co-expressed in 170 μl of LTE for 3.5 h at 27 °C. The co-expression was performed with 6 nm of the vector coding for N-terminal GFP "bait" SNX proteins and 18 nm of the vector coding for a C-terminal Cherry-myc "prey" SNX protein. A GFP construct was used as a negative control bait. NaCl was added to the expressed protein (to a final concentration of 200 mm) and the samples were incubated with 10 μl of GFP-nanotrap coated beads (NHS-activated Sepharose coupled with MBP-GFP-Nanotrap) for 30 min at 4 °C with gentle mixing by rotation. Subsequently, the beads were washed six times with 200 μl of wash buffer (PBS with 0.1% Triton X-100 and 200 mm NaCl). The proteins were released from the beads by heating for 3 min at 72 °C in 15 μl of 2x NuPAGE LDS loading buffer and resolved as described above for gel electrophoresis. Fluorescence intensity was analyzed using ImageJ software and a leakage of 10% of the GFP fluorescence into the Cherry channel was accounted for in the quantification of the pull-down results. Single molecule spectroscopy was performed as described previously (22.Gambin Y. Schug A. Lemke E.A. Lavinder J.J. Ferreon A.C. Magliery T.J. Onuchic J.N. Deniz A.A. Direct single-molecule observation of a protein living in two opposed native structures.Proc. Natl. Acad. Sci. U.S.A. 2009; 106: 10153-10158Crossref PubMed Scopus (71) Google Scholar, 23.Gambin Y. VanDelinder V. Ferreon A.C. Lemke E.A. Groisman A. Deniz A.A. Visualizing a one-way protein encounter complex by ultrafast single-molecule mixing.Nat. Methods. 2011; 8: 239-241Crossref PubMed Scopus (110) Google Scholar). Proteins were expressed in LTE and immediately diluted ∼400 times in buffer A, to a final concentration of ∼50 pm. A volume of 20 μl of each sample was placed into a custom-made silicone 192-well plate that self-seals on top of a large 70 × 80 mm glass coverslip (ProSciTech, Australia). Plates were analyzed at room temperature on a Zeiss LSM710 microscope with a FCS Confocor3 module. For intensity measurements, N-terminal GFP-labeled SNX proteins were expressed in 10 μl of LTE using 50 nm of DNA and incubation for 3 h at 27 °C. One laser (488 nm) was focused in solution using a 40×/1.2 NA water-immersion objective. Fluorescence from GFP was filtered by a 505–540 band pass filter. The number of photons collected in 1 ms time bins (IGFP(t)) was recorded. A single-molecule event was detected when the total intensity was above a threshold of 50 photons. Single-molecule brightness analysis is the simple "counting" of photons emitted by a protein complex and comparison to the typical number of photons obtained from a GFP monomer, to calculate the number of GFP fluorophores present in the complex. This is however complicated by the fact that proteins diffuse randomly through the focal volume and that different trajectories are observed. The optimal trajectory maximizes the time spent in the focal volume and the number of photons emitted. If one considers that at any time step, the proteins can escape the detection volume, this optimal trajectory is exponentially rare and smaller bursts are more frequent. Therefore, the distribution of burst size typically follows an exponential decay. The number of events for each intensity range was counted and normalized to the total number of events to give a probability P(I). Intensity measurement graphs were obtained by measuring more than 1000 events; P(I) is plotted as a function of burst intensity (I) (photons per ms). For coincidence experiments, N-terminal GFP-labeled and N-terminal Cherry labeled SNX proteins were co-expressed using 40 nm of DNA in 10 μl of LTE for 3 h at 27 °C. Two lasers (488 nm and 561 nm) were focused in solution using a 40×/1.2 NA water-immersion objective. Fluorescence was collected and separated using a 565 nm dichroic; signal from GFP (IGFP(t)) was filtered by a 505–540 band pass filter, whereas fluorescence from Cherry (ICherry(t)) was filtered by a 580 nm long pass. The fluorescence of the two channels was recorded simultaneously and separately, adding the number of photons collected in 1 ms time bins. A single-molecule event was detected when the total intensity of the two channels was above a threshold of 50 photons. For each event, the intensities of the GFP and Cherry bursts were corrected for background and leakage (10% leakage of the GFP into the Cherry channel). The coincidence C was then measured ratiometrically as the corrected Cherry signal, divided by the total intensity of the burst (C = ICherry/[IGFP + ICherry]). In the absence of Cherry fluorescence, C is close to zero, and in the absence of GFP, C tends toward 1. Events with 0.25<C 1000 events per interaction, and fitted by Gaussian peaks for GFP-only, coincidence and Cherry-only contributions. The bound fraction was calculated as the proportion of coincidence (0.25<C<0.75) to total events. We set out to systematically analyze the pairwise interactions between 12 SNX proteins, following co-expression of all 144 possible pairs. Due to the large number of possible combinations, a cell-free approach was used to accelerate conversion of DNA to protein while controlling of the co-expression ratios. We employed the recently developed eukaryotic L. tarentolae-based cell free expression system (LTE). This system enables expression of complex human proteins in their functional and full-length form (24.Guo Z. Johnston W. Kovtun O. Mureev S. Brocker C. Ungermann C. Alexandrov K. Subunit organization of in vitro reconstituted HOPS and CORVET multisubunits membrane tethering complexes.Plos One. 2013; 8: e81534Crossref PubMed Scopus (12) Google Scholar). As L. tarentolae is a fast-growing and fermentable organism, the translation-competent lysate can be generated rapidly and inexpensively (21.Kovtun O. Mureev S. Jung W. Kubala M.H. Johnston W. Alexandrov K. Leishmania cell-free protein expression system.Methods. 2011; 55: 58-64Crossref PubMed Scopus (67) Google Scholar, 25.Mureev S. Kovtun O. Nguyen U.T. Alexandrov K. Species-independent translational leaders facilitate cell free expression.Nat. Biotechnol. 2009; 27: 747-752Crossref PubMed Scopus (108) Google Scholar, 26.Kovtun O. Mureev S. Johnston W. Alexandrov K. Towards the construction of expressed proteomes using a Leishmania tarentolae based cell-free expression system.PLoS One. 2010; 5: e14388Crossref PubMed Scopus (30) Google Scholar). The SNX proteins were expressed in vitro as GFP and Cherry-myc fusion proteins. GFP and Cherry fluorescence were used to detect expressed proteins and GFP and myc serve as affinity tags (27.Kubala M.H. Kovtun O. Alexandrov K. Collins B.M. Structural and thermodynamic analysis of the GFP:GFP-nanobody complex.Protein Sci. 2010; 19: 2389-2401Crossref PubMed Scopus (196) Google Scholar). Genetically encoded tags are relatively large and may interfere with protein folding and the ability to engage in protein-protein interactions. To address these issues, we iteratively tagged each SNX with an N-terminal GFP tag and a C-terminal Cherry-myc tag and each interaction was tested in two combinations. Cloning of the 24 different fusion proteins was greatly simplified by the development of a cell-free expression vector compatible with Gateway recombination technology* that enabled us to source the SNX-BAR genes from a human Open Reading Frame (ORF) library (supplemental Fig. S2). All fusion proteins were expressed in LTE and analyzed by SDS-PAGE, and as can be seen in the Fig. 1C, all proteins migrated corresponding to their expected molecular weights. All proteins could be expressed at similar levels, typically between 1 and 1.5 μm (supplemental Figs. S4 and S5). Note that SNX6 was a truncated clone, with an intact BAR domain. The only human SNX-BAR protein not included in these studies is SNX18, as its ORF was not found in available libraries. AlphaScreen is a sensitive bead-based proximity assay able to detect interactions with a wide range of affinities (from pm to mm) (28.Eglen R.M. Reisine T. Roby P. Rouleau N. Illy C. Bosse R. Bielefeld M. The use of AlphaScreen technology in HTS: current status.Curr. Chem. Genomics. 2008; 1: 2-10Crossref PubMed Google Scholar, 29.Taouji S. Dahan S. Bosse R. Chevet E. Current screens based on the AlphaScreen technology for deciphering cell signaling pathways.Curr. Genomics. 2009; 10: 93-101Crossref PubMed Scopus (48) Google Scholar). Fig. 2A shows a schematic of how this assay was employed to analyze the pair-wise interactions between SNX protein pairs following their co-expression in the cell-free system, without any purification or enrichment steps. AlphaScreen signals can occur even if beads are relatively far apart (distance up to 200 nm), and it is therefore technically a proximity assay as opposed to a direct binding assay. However, as we are using an orthogonal in vitro expression system, bridging by intermediate proteins from the LTE is not expected and interactions are most likely direct. The luminescence signal upon binding depends strongly on protein concentration (Fig. 2B), thus different dilutions of the proteins were used to determine the maximum response of each AlphaScreen assay. The system is therefore self-calibrating, adjusting for the differences in initial protein concentrations so the protein expression levels do not have to be tightly controlled. This feature of the binding signal also serves as an internal quality control, as nonspecific interactions do not display a strong dose dependent response. The typical AlphaScreen data presented in Fig. 2C for a pair of interacting proteins (SNX8-SNX8) and for a noninteracting pair (SNX3-SNX3) demonstrates that the method can detect specific interactions between SNX proteins with confidence. In the interaction screen, all 144 protein pairs were tested in triplicate and for each pair 5 dilutions were performed to accurately measure the maximal response of the assay. Because of the excellent scalability of the AlphaScreen assay we were able to perform it in a 384-well format. We averaged the results obtained with the N- and C-terminal tag combinations and plotted the aggregated results as a matrix of interactions (Fig. 3). An AlphaScreen binding index was calculated relative to a strong binding pair (SNX4-SNX4), which was used as internal control in every AlphaScreen plate. We chose an arbitrary threshold of 50% of the SNX4-SNX4 binding index as a cut-off for a positive interaction, and then sought independent validation of these results using other methods. Under these conditions, all proteins except for SNX1, SNX3, and SNX5 are able to form homo-dimers, and most show significant selectivity for different heteromeric binding partners. In nearly all cases, the signal intensities obtained with probes carrying tags at different termini were comparable, providing a high degree of confidence in the binding specificities we observe. Only one significant outlier was seen, and interestingly this interaction is between SNX8 and the nonBAR domain protein SNX3, where we observe an interaction using GFP-tagged SNX8 and cherry-tagged SNX3, but little association when tags are reversed (supplemental Fig. S1). We believe however, that this is a specific interaction based on further validation as discussed below. The propensity to form homo-oligomers was also assessed by a co-immunoprecipitation (co-IP) assay (Fig. 4). The bait proteins were expressed with an N-terminal GFP tag, whereas C-terminal Cherry-myc fusion proteins were used as prey. Immunoprecipitation was carried out using GFP-antibody functionalized Sepharose, and samples were resolved by SDS-PAGE and scanned for GFP and Cherry fluorescence. To compensate for variations in expression levels, we measured the intensities of Cherry fluorescence before and after pull-down and calculated the ratio of intensities as a measure of binding. Qualitative comparison of the two methods (AlphaScreen and co-IP) validated our choice of a binding index of 50 as a threshold for positive interactions (Figs. 4B and 4C). Furthermore, a higher AlphaScreen index roughly correlates with a stronger interaction as estimated by co-IP. This analysis also revealed limitations of the two methods, with co-IPs failing to reproduce the self-association of SNX2, and AlphaScreen underestimating the strength of the SNX9 self-interaction. Although the results of co-IP experiments correlate well with AlphaScreen data, both methods measure interactions of proteins coupled to solid supports that may potentially contribute to nonspecific association. To address this issue, we turned to single molecule spectroscopy of freely diffusing fluorescently labeled proteins. Single-molecule spectroscopy is typically performed after the purification of recombinant proteins and labeling with organic fluorescent dyes. We found that GFP and Cherry fluorophores can be measured at the single-molecule level, even on freely diffusing proteins (30.Gambin Y. Ariotti N. McMahon K.-A. Bastiani M. Sierecki E. Kovtun O. Polinkovsky M. Magenau A. Jung W. Okano S. Zhou Y. Leneva N. Mureev S. Johnston W. Gaus K. Hancock J.F. Collins B.M. Alexandrov K. Parton R.G. Single molecule analysis reveals self-assembly and nanoscale segregation of two distinct cavin subcomplexes on caveolae.eLife. 2013; 3: e01434PubMed Google Scholar). The fluorescence intensity of a monomeric fluorescent protein can be calibrated precisely and subsequently used to "count" the number of proteins in diffusing protein complexes. The SNX-BAR proteins were expressed with an N-terminal GFP tag in LTE, and the samples were analyzed on a confocal microscope configured for single-molecule studies. For brightness analysis, we used a 488 nm laser to excite the GFP fluorophores. Time traces of GFP emission were recorded in 1 ms time "bins" to improve signal/noise ratio. Proteins were diluted to picomolar concentrations where only single proteins or protein complexes were present in the confocal volume. When GFP-tagged proteins diffuse through the confocal volume, a fluorescent burst is recorded. The burst intensity indicates how many fluorophores are present in the confocal volume at the given time. In the case of dimer formation, the burst