Subscriber access provided by Szilard Library EMBL Article Triple SILAC to determine stimulus specific interactions in the Wnt pathway Maximiliane Hilger, and Matthias Mann J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/pr200740a • Publication Date (Web): 20 Oct 2011 Downloaded from http://pubs.acs.org on November 8, 2011

Just Accepted

“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Proteome Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties. Page 1 of 43 Journal of Proteome Research

1 2 3 4 Triple SILAC to determine stimulus specific interactions 5 6 7 8 in the Wnt pathway 9 10 11 12 13 14 15 16 Maximiliane Hilger 1,2 and Matthias Mann 1* 17 18 19 1 20 Department of Proteomics and Signal Transduction, Max-Planck Institute of Biochemistry, Am 21 22 Klopferspitz 18, D-82152 Martinsried, Germany 23 24 25 2 26 New address: Pharma Research & Early Development, Roche Diagnostics GmbH, Nonnenwald 27 28 2, D-82372 Penzberg, Germany 29 30 31 32 Keywords: interaction, interactome, Signaling pathways, SILAC, dynamic interactions, 33 34 APC, Axin-1, Wnt3a,DVL2, CTBP2 35 36 37 *Corresponding author: Matthias Mann, Department of Proteomics and Signal Transduction, 38 39 40 Max-Planck Institute of Biochemistry, Am Klopferspitz 18, D-82152 Martinsried, Germany. 41 42 E-mail: [email protected] ; Ph.: 49-89-8578-2557; Fax: +49 89 8578 2219. 43 44 45 46 Abbreviations: AP ‑MS, affinity purification coupled with mass spectrometry; BAC, bacterial 47 48 artificial ; FDR, false discovery rate; GFP, green fluorescent protein; GO, 49 50 51 Ontology; LTQ, linear trap quadrupole; QUBIC, quantitative BAC interactomics; SILAC, stable 52 53 isotope labeling by amino acids in cell culture; TAP, tandem affinity purification. 54 55 56 57 58 59 60 [1]

ACS Paragon Plus Environment Journal of Proteome Research Page 2 of 43

1 2 3 4 ABSTRACT 5 6 7 Many important regulatory functions are performed by dynamic multi-protein complexes 8 9 10 that adapt their composition and activity in response to different stimuli. Here we employ 11 12 quantitative affinity purification coupled with mass spectrometry to efficiently separate 13 14 15 background from specific interactors but add an additional quantitative dimension to 16 17 explicitly characterize stimulus dependent interactions. This is accomplished by SILAC in a 18 19 20 triple-labeling format, in which pull-downs with bait, with bait and stimulus and without bait 21 22 are quantified against each other. As baits we use full-length fused to the green 23 24 25 fluorescent protein and expressed under endogenous control. We applied this technology to 26 27 Wnt signaling, which is important in development, tissue homeostasis and cancer, and 28 29 30 investigated interactions of the key components APC, Axin-1, DVL2 and CtBP2 with 31 32 differential pathway activation. Our screens identify many known Wnt signaling complex 33 34 35 components and link novel candidates to Wnt signaling, including FAM83B and Girdin, which 36 37 we found as interactors to multiple Wnt pathway players. Girdin binds to DVL2 independent 38 39 40 of stimulation with the ligand Wnt3a but to Axin-1 and APC in a stimulus dependent manner. 41 42 . The core destruction complex itself, which regulates beta-catenin stability as the key step in 43 44 45 canonical Wnt signaling, remained essentially unchanged. 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 [2]

ACS Paragon Plus Environment Page 3 of 43 Journal of Proteome Research

1 2 3 Introduction 4 5 6 7 The study of protein-protein interactions is of pivotal importance because most biological 8 9 functions are mediated by protein complexes. In contrast to most other techniques, affinity 10 11 12 purification coupled with mass spectrometry (AP-MS) is unbiased in that it does not require 13 14 knowledge of potential interaction partners and it can be used for systems-wide analysis of 15 16 1 17 protein-protein interaction networks . AP-MS has often been performed with the goal of a high 18 19 degree of purification of protein complexes using tandem affinity purification (TAP) tagging 20 21 2 22 approaches . However, this requires large amounts of starting material because of the two 23 24 purification steps. Furthermore the stringent washing conditions involved in TAP purifications 25 26 27 often lead to loss of weakly bound protein complex members. Quantitative mass spectrometry 28 29 can overcome these limitations by distinguishing specific interactors from unspecific 30 31 3-5 32 background binders by the ratios of proteins in bait versus control pull-downs . This allows 33 34 single step low stringent purification and high confidence interaction mapping including weak 35 36 37 interactors. 38 39 40 There are many different formats for performing AP-MS in a qualitative, semi-quantitative 41 42 6-22 43 and quantitative fashion . Recently our laboratory has established an integrated quantitative 44 45 workflow for AP-MS using bacterial artificial (BACs) containing the gene of 46 47 48 interest fused to the green fluorescent protein (GFP), which leads to expression of the full- 49 50 length, GFP-tagged proteins from their endogenous promoters 23 . This system, termed QUBIC 51 52 53 for QUantitative BAC InteraCtomics, has several advantages. Most importantly the bait protein 54 55 is expressed close to endogenous levels because the entire gene encoding the bait protein, 56 57 58 59 60 [3]

ACS Paragon Plus Environment Journal of Proteome Research Page 4 of 43

1 2 3 including up- and downstream regulatory elements, is stably integrated into the genome of the 4 5 24-27 6 cell . As tagged transcripts and proteins are processed by the cell equally to the endogenous 7 8 counterpart, different splice isoforms can be expressed and proteins are post-translationally 9 10 11 modified in the correct manner. Furthermore, cell lines expressing tagged versions of very large 12 13 proteins can be created. In contrast to APs of the endogenous proteins, the QUBIC strategy 14 15 16 does not rely on the availability of highly specific and immunoprecipiating antibodies for each 17 18 19 protein of interest. 20 21 22 Most protein complexes, especially those with regulatory functions, are dynamic structures 23 24 that form or change their composition and activity in response to cellular perturbations 28 . 25 26 27 Stimulation-dependent changes in protein conformation, subcellular localization or 28 29 modification determine the interaction properties of the different complex members. AP-MS 30 31 29, 30 32 using stable isotope labeling with amino acids in cell culture (SILAC) in a double-labeling 33 34 format is frequently employed for the characterization of protein interactions. SILAC with three 35 36 31- 37 isotope states has previously mainly been used to study the time dimension of the proteome 38 39 33 but has also enabled comparison of bead proteomes 34 , differentiation of isoform specific 40 41 42 interactors 6 and the change in composition of RNA polymerase upon inhibition of 43 44 transcription 35 . 45 46 47 48 Here we wished to establish and characterize a general method for characterizing 49 50 constitutive and stimulation dependent dynamic interaction partners of regulatory protein 51 52 53 complexes. We combined the QUBIC approach with triple SILAC labeling to differentiate 54 55 background binders from specific binders and, in the same experiment, constitutive interactors 56 57 58 59 60 [4]

ACS Paragon Plus Environment Page 5 of 43 Journal of Proteome Research

1 2 3 from those that associate with a complex in a stimulus dependent manner. We applied this 4 5 6 method to the analysis of complexes in the Wnt signaling pathway and investigated differential 7 8 complex formation dependent on stimulation of cells with the Wnt ligand Wnt3a. 9 10 11 12 The canonical Wnt pathway regulates cell fate, proliferation and self-renewal of adult stem 13 14 and progenitor cells during the entire lifespan of metazoan organisms36-40 . Aberrant regulation 15 16 17 of this pathway leads to different diseases, most prominently sporadic colon cancer. The key 18 19 step in canonical Wnt signaling is the regulation of β-catenin. In the absence of Wnt ligands 20 21 22 β-catenin levels are low as a result of its continuous phosphorylation by the destruction 23 24 complex, which triggers ubiquitylation and subsequent proteasomal degradation. Core 25 26 27 components of the destruction complex are APC (Adenomatous Polyposis Coli) and Axin-1, 28 29 which both function as scaffolds, and the kinases glycogen synthase kinase-3β (GSK-3β) and 30 31 32 casein kinase I-α (CKI-α). Upon Wnt ligand binding to the receptors Frizzled and LRP5/6, the 33 34 destruction complex function is attenuated, at least in part through relocalization to the plasma 35 36 36-40 37 membrane and interactions with Dishevelled (DVL) . Levels of β-catenin then accumulate in 38 39 the cytoplasm and β-catenin translocates to the nucleus where it binds to TCF/LEF transcription 40 41 42 factors and coactivates transcription of target . 43 44 45 Because of its central importance, the Wnt pathway is intensively studied and new 46 47 48 pathway players that may be potential therapeutic targets are still found using a variety of 49 50 approaches 41-43 . Although canonical Wnt signaling has been investigated in depth the exact 51 52 53 mechanism by which the destruction complex is inhibited and β-catenin is stabilized is still not 54 55 fully understood. Our Wnt pathway interactome study identifies potential novel Wnt pathway 56 57 58 59 60 [5]

ACS Paragon Plus Environment Journal of Proteome Research Page 6 of 43

1 2 3 members and sheds light on the dynamics of the complexes involved. 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 [6]

ACS Paragon Plus Environment Page 7 of 43 Journal of Proteome Research

1 2 3 4 5 6 7 EXPERIMENTAL PROCEDURES 8 9 Cell culture. HeLa cells stably expressing BACs with human GFP-tagged APC, Axin-1, DVL2 10 11 12 or CtBP2 were grown at 37ºC and 5% CO 2 for at least five passages in SILAC media (Dulbecco's 13 14 modified eagle medium minus L-arginine and L-lysine (Invitrogen) with 10% dialyzed fetal 15 16 17 bovine serum (Invitrogen) and 100 U/mL penicillin/streptomycin (Invitrogen)) containing 18 19 49 mg/mL L-arginine (Arg0) and L-lysine (Lys0) (light), 13 C 14N -L-arginine (Arg6) (Euriso-top) and 20 6 4 21 13 15 13 15 22 4,4,5,5-D4-L-lysine (Lys4) (medium) or C6 N4-L-arginine (Arg10) and C6 N2-L-Lysine (Lys8) 23 24 (heavy) each supplemented with 400 µg/mL geneticin (Invitrogen). The untransfected control 25 26 27 HeLa Kyoto cells were only light or heavy SILAC labeled. Cells were expanded to four, 80% 28 29 confluent 15 cm dishes per affinity purification and per SILAC label (in total 12 dishes for one 30 31 32 triple SILAC experiment). The cell lines were generated by the BAC recombineering 33 34 technology 24, 25 and used as transgenic cell pools. 35 36 37 38 Wnt stimulation and cell harvest. In the ‘forward’ experiment heavy labeled transgenic cell 39 40 lines were stimulated for two hours with 200 ng/mL recombinant mouse Wnt3a (RD Systems), 41 42 43 dissolved in carrier solution (0.1% BSA in PBS) (Suppl. Figure 1). The corresponding medium 44 45 labeled transgenic cells and light labeled untransfected control cells were incubated with the 46 47 48 carrier solution for two hours. In the ‘reverse’ experiment the labels in the previous ‘heavy’ and 49 50 ‘light’ conditions were interchanged, whereas the medium labeled condition was unchanged. 51 52 53 Subsequently cells were trypsinized, pelleted, resuspended in PBS and counted. Equal cell 54 55 numbers of each SILAC condition were separately pelleted, snap frozen and stored at -80°C. 56 57 58 59 60 [7]

ACS Paragon Plus Environment Journal of Proteome Research Page 8 of 43

1 2 3 Preparation of cell extract. Cell pellets were thawed on ice and resuspended in 2 mL ice- 4 5 6 cold lysis buffer (basic buffer (150 mM NaCl, 50 mM Tris-HCl (pH 7.5), 5% glycerol), 1% IGPAL- 7 8 CA-630 (Sigma), 1 mM MgCl , 1% Benzonase (Merck) and 1x EDTA-free complete protease 9 2 10 11 inhibitors (Roche). After incubation for 40 min on a rotation wheel at 4°C lysates were 12 13 centrifuged at 4,000xg for 15 min at 4°C. Supernatants were subjected to affinity purification. 14 15 16 17 Affinity purification. Each cleared SILAC extract was incubated separately with 100 µL 18 19 μMACS mouse monoclonal anti-GFP antibody coupled magnetic microbeads (Miltenyi Biotech) 20 21 22 for 15 min. One µColumn (Miltenyi Biotech) per SILAC extract was equilibrated with 250 µL 23 24 basic buffer containing 1% IGPAL-CA-630 (Sigma) using a hand magnet (Miltenyi Biotech). After 25 26 27 incubation with the beads, lysates were applied to the columns. Subsequently columns were 28 29 rinsed four times with 1 mL basic buffer containing 0.05% IGPAL-CA-630 (Sigma). For unspecific 30 31 32 protein elution 25 µL of preheated (95°C) SDS gel loading buffer (50 mM Tris HCl (pH 8), 50 mM 33 34 DTT, 1% SDS, 0.005% bromophenol blue, 10% glycerol) were added and incubated for 5 min. 35 36 37 Eluates were collected by adding additional 30 µL preheated SDS gel loading buffer to each 38 39 column. Corresponding eluates of the triple SILAC experiment and 30 µL NuPAGE LDS sample 40 41 42 buffer (Invitrogen) were combined. 43 44 45 Protein digestion. Combined eluates were separated by 1D-SDS PAGE (4-12 % Novex mini- 46 47 48 gel) (Invitrogen) and visualized by colloidal Coomassie staining (Invitrogen). Proteins were 49 50 separated in three adjacent lanes that were subsequently cut into 8 slices. All gel slices were 51 52 44, 45 53 subjected to in-gel digestion with trypsin (Promega) . Resulting tryptic peptides were 54 55 extracted with 30 % ACN in 3 % TFA, concentrated until full evaporation of organic solvent and 56 57 58 59 60 [8]

ACS Paragon Plus Environment Page 9 of 43 Journal of Proteome Research

1 2 3 further concentrated and desalted on reversed phase C StageTips46, 47 . Shortly prior to high 4 18 5 6 resolution liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis peptides 7 8 were eluted twice from the StageTips with 20 µL buffer B (80% ACN in 0.5% acetic acid) solution 9 10 11 into a 96 sample well plate (Abgene), concentrated in a SpeedVac concentrator until removal of 12 13 the organic solvent and reconstituted with buffer A* (2% ACN in 0.1 % TFA). 14 15 16 17 LC-MS/MS analysis. Eluted peptides were analyzed by a nanoflow HPLC (Proxeon 18 19 Biosystems) coupled on-line via a nanoelectrospray ion source (Proxeon Biosystems) to a linear 20 21 22 trap quadrupole (LTQ)-Orbitrap Velos mass spectrometer (Thermo Fisher Scientific). Peptide 23 24 mixtures were loaded with an IntelliFlow of maximal 500 nl/min onto a C -reversed phase 25 18 26 27 column (15 cm long, 75 μm inner diameter, packed in-house with ReproSil-Pur C 18 -AQ 3 μm 28 29 resin (Dr. Maisch)) in buffer A (0.5% acetic acid). Peptides were eluted with a multi-segment 30 31 32 linear gradient of 5–60% buffer B (80% ACN and 0.5% acetic acid) at a constant flow rate of 33 34 250 nl/min over 107 min. Mass spectra were acquired in the positive ion mode applying a data- 35 36 37 dependent automatic switch between survey scan and tandem mass spectra (MS/MS) 38 39 acquisition. A ‘top 10’ method was applied that acquires one Orbitrap survey scan in the mass 40 41 42 range of m/z 300-1650 followed by MS/MS of the ten most intense ions in the LTQ. The target 43 44 value in the LTQ-Orbitrap was 1,000,000 for survey scan at a resolution of 60,000 at m/z 400. 45 46 47 Fragmentation in the LTQ was performed by collision-induced dissociation with a target value 48 49 of 5,000 ions. The ion selection threshold was 500 counts. Selected sequenced ions were 50 51 52 dynamically excluded for 90 seconds. 53 54 55 56 57 58 59 60 [9]

ACS Paragon Plus Environment Journal of Proteome Research Page 10 of 43

1 2 3 Data analysis. Raw mass spectrometric data were analyzed with the MaxQuant software 4 5 48, 49 6 (version 1.0.14.9) . A false discovery rate (FDR) of 0.01 for proteins and peptides and a 7 8 minimum peptide length of 6 amino acids were required. A time-dependent mass recalibration 9 10 11 algorithm was used instead of lock masses for recalibration to improve the mass accuracy of 12 13 precursor ions. MS/MS spectra were searched by Mascot (version 2.2.04, Matrix Science) 14 15 16 against the IPI human data base (version 3.62) (containing 83,947 entries) combined with 17 18 19 262 common contaminants and concatenated with the reversed versions of all sequences. For 20 21 the Mascot search trypsin allowing for cleavage N-terminal to proline was chosen as enzyme 22 23 24 specificity. Cysteine carbamidomethylation was selected as a fixed modification, while protein 25 26 N-terminal acetylation and methionine oxidation were selected as variable modifications. 27 28 29 MaxQuant was used for scoring of the peptides for identification. It also determined the SILAC 30 31 state of peptides by the mass differences between SILAC peptide pairs and this information was 32 33 34 used to perform searches with fixed Arg6 and Lys4 or Arg10 and Lys8 modifications as 35 36 appropriate. Maximally two missed cleavages and three labeled amino acids were allowed. 37 38 39 Initial mass deviation of precursor ion was up to 7 ppm, mass deviation for fragment ions was 40 41 0.5 units on the m/z scale. Protein identification required two peptides one of which had to be 42 43 44 unique to the protein group. 45 46 47 Quantification in MaxQuant was performed as described 48 . The ‘Match between runs’ 48 49 option was selected, which enabled the transfer of identifications between the MS analysis of 50 51 52 the same and the adjacent gel slices of all replicates and their quantification across the 53 54 55 replicates. The ‘Requantify’ option was enabled, which in effect integrates noise levels for 56 57 undetected SILAC partners in order to estimate a lower bound on the SILAC ratio. Data analysis 58 59 60 [10]

ACS Paragon Plus Environment Page 11 of 43 Journal of Proteome Research

1 2 3 plots were either performed in the MaxQuant environment (Perseus) or in the R environment 50 . 4 5 6 The label swap of the control cell line to light SILAC label was additionally used to filter out non- 7 8 assigned contaminants such as rare not contained in the contaminant database (Suppl. 9 10 11 Figure 4). (GO) analysis was performed with AMIGO 51 . STRING 9.0 analyses on 12 13 the detected interactomes retrieved known interactors (Suppl. Table 1). Cytoscape 52, 53 was 14 15 16 used to visualize APC and Axin-1 interactome overlap. All protein group files, containing the 17 18 19 information of all protein pull-downs, are uploaded to TRANCHE (see instructions at the end of 20 21 the manuscript) . 22 23 24 Fluorescence microscopy. HeLa cells stably expressing BACs with human GFP-tagged 25 26 27 FAM73A were grown in 35 mm glass bottom dishes (MatTek). After staining cells with 28 29 MitoTracker Red CM-H2XRos (Molecular Probes, Invitrogen) cells were imaged with a spinning- 30 31 32 disk confocal microscope (TiLL iMIC CSU22; Andor) using a back-illuminated EM charge-coupled 33 34 device camera (iXonEM 897; Andor) and a 60x 1.4 NA oil immersion objective (Olympus). 16-bit 35 36 37 images were collected using Image iQ (version 1.9; Andor) in the linear range of the camera. 38 39 They were deconvoluted with Huygens Software and cropped with ImageJ 40 41 42 (http://rsbweb.nih.gov/ij/ ). 43 44 45 Western Blot. Input material for affinity purifications (1/1000 of the protein lysate) was 46 47 48 boiled with 5 µL LDS sample buffer (Invitrogen) for 5 min at 95 °C and separated on a 4-12 % 49 50 NOVEX gradient gel using MOPS buffer (Invitrogen). Proteins were transferred to a PVDF 51 52 53 membrane in a wet chamber (Biorad) at 300 mA for 1.25 h. The membrane was blocked with 54 55 PBST + 1 % BSA for 1 h at room temperature and subsequently incubated with Anti- β-Tubulin 56 57 58 59 60 [11]

ACS Paragon Plus Environment Journal of Proteome Research Page 12 of 43

1 2 3 antibody (1:20,000) (Sigma) and Anti-β-catenin antibody (1:2,000) (BD) in PBST + 1 % BSA for 4 5 6 1 h at room temperature. The membrane was washed with PBST three times and subsequently 7 8 incubated with HRP-anti-mouse (Amersham) (1:10,000) in PBST + 1 % BSA for 1 h at room 9 10 11 temperature. The membrane was washed with PBST three times prior to detection with the ECL 12 13 Western blotting detection reagent (Amersham) according to manufacturer’s instruction. 14 15 16 Chemiluminescence screens (GE Healthcare) were used to visualize the band patterns. 17 18 19 20 21 22 23 RESULTS AND DISCUSSION 24 25 26 27 Triple SILAC enables detection of constitutive and dynamic interactions. We established 28 29 QUBIC 54 in a triple-encoding SILAC format to allow analysis of both constitutive and dynamic 30 31 32 interactions of bait proteins that belong to diverse levels of canonical Wnt signaling under 33 34 differential pathway activation. These encompassed APC and Axin-1 scaffold components of the 35 36 37 destruction complex, DVL2, a mediator of the Wnt signal from the membrane to the 38 39 destruction complex, as well as CtBP, a coregulator of Wnt target gene transcription 36-40 . We 40 41 42 chose a SILAC-based approach, because a precise quantification strategy was mandatory to 43 44 45 resolve interaction dynamics. SILAC is known to allow highly accurate and reproducible 46 47 quantification and to be superior of label-free quantification methods for complex and dynamic 48 49 50 experiments where small fold-changes in binding are expected. 51 52 53 The cells expressing GFP-tagged protein were light (L) and medium (M) SILAC labeled, while 54 55 the control cell line without BAC transgene was heavy (H) SILAC labeled (Figure 1A). The light 56 57 58 59 60 [12]

ACS Paragon Plus Environment Page 13 of 43 Journal of Proteome Research

1 2 3 labeled transgenic cells were stimulated with Wnt3a for two hours. Wnt stimulation was 4 5 6 controlled by detection of β-catenin protein levels of the input material via Western blot (Suppl. 7 8 Figure 1). To prevent light to heavy or medium to heavy exchange of specific interaction 9 10 11 partners during the immunopurification procedure 16, 55 , GFP pull-downs were performed 12 13 separately for each SILAC condition and eluates were mixed afterwards. For in-depth 14 15 16 interactome characterization we reduced the sample complexity by one-dimensional gel 17 18 19 separation into eight slices. Eluates were characterized at high sensitivity on the high-resolution 20 21 LTQ-Orbitrap Velos instrument 56 . Detected peptides can be classified according to their SILAC 22 23 24 triplet peak patterns (Figure 1B). For unspecific background binders to either beads or GFP- 25 26 antibody, this pattern shows no change between the three states (‘one to one to one’). A 27 28 29 specific interactor with the GFP-bait will have peptide ratios between the pull-down with the 30 31 non-stimulated cell population and the untransfected control cell population (M/H ratio in this 32 33 34 case) and/or between the stimulated cell population and the control (L/H ratio). A stimulus 35 36 dependent interactor has a significant ratio for the peptide intensity of the GFP-bait pull-down 37 38 39 from the stimulated cell population compared to the GFP-bait pull-down from the non- 40 41 stimulated cell population (L/M ratio). 42 43 44 Via a two-dimensional plot proteins can be grouped into constitutive interactors and 45 46 47 dynamic interactors according to the ratios of GFP-bait to control cells (here M/H) and GFP-bait 48 49 with stimulus to GFP-bait without stimulus (L/M). Figure 1C illustrates this principle in cartoon 50 51 52 form, with outliers in the positive x-direction (M/H) representing specific binding to the bait 53 54 55 protein. This dimension contains the information of a standard SILAC interaction experiment. 56 57 Plotting the ratio between stimulated and non-stimulated cell populations (L/M) on the y-axis 58 59 60 [13]

ACS Paragon Plus Environment Journal of Proteome Research Page 14 of 43

1 2 3 adds the stimulus dependent dimension. Outliers in the positive y-direction bind more strongly 4 5 6 upon Wnt stimulus, whereas outliers in the negative direction selectively dissociate from the 7 8 complex upon stimulation. The three main protein classes are therefore proteins that bind 9 10 11 specifically to the bait but not in a signal dependent manner (blue oval in Figure 1C), proteins 12 13 that increase their binding upon stimulation (green oval) and proteins that decrease their 14 15 16 binding (red oval). 17 18 19 While two of the three possible ratios of the triplets are in principle sufficient to represent 20 21 22 the constitutive and stimulus dependent aspects of interaction, in practice all three dimensions 23 24 are often needed. This is because the third ratio can be more accurately determined directly 25 26 27 rather than estimated from the other two. Furthermore, proteins that only bind specifically in 28 29 one stimulus state are not optimally represented in the graph plotting the specificity ratio for 30 31 32 the other state on the x-axis. They are only separated from background binding proteins in one 33 34 dimension (positive outlier on the y-axis). 35 36 37 38 Heat maps turned out to be a very valuable additional visualization method, combining all 39 40 the information from triple SILAC pull-downs into a single picture. This was particularly true for 41 42 43 integrating the data from reverse labeling experiments (see below). To generate these heat 44 45 maps, we placed all pull-down ratios between the triplet states on the horizontal axis and 46 47 48 performed one-dimensional hierarchical clustering of the multiple ratios of each quantified 49 50 prey protein in the vertical dimension. 51 52 53 54 We typically detected about 1,200 proteins per pull-down experiment of which about 55 56 1,100 proteins were quantified with at least two ratio counts. These large numbers reflect the 57 58 59 60 [14]

ACS Paragon Plus Environment Page 15 of 43 Journal of Proteome Research

1 2 3 single step purification procedure and our low stringency washing conditions. However, the 4 5 6 quantitative information encoded in the SILAC ratios immediately excluded the vast majority of 7 8 these proteins from consideration (typically more than 95% of all quantified proteins). 9 10 11 12 For each bait, experiments were repeated after swapping the SILAC labels between the 13 14 control and the stimulated cell populations. This introduced an additional dimension of 15 16 17 specificity and provided a minimum of two biological replicates. Because we required at least 18 19 two peptides and two ratio counts, there was a minimum of four data points for quantitation 20 21 22 per protein. ‘Forward’ and ‘reverse’ experiments together took 1.5 days of measurement time. 23 24 25 For an initial check of the results of our screen after filtering for significant and 26 27 28 reproducible SILAC ratios, we inspected the interactomes of the different baits. These range in 29 30 size from four to 28. According to STRING database analysis they contained 33 to 75% of 31 32 33 already known interactors (Suppl. Table 1), a high value considering the diversity of systems in 34 35 which Wnt signaling has been studied. APC is an interaction scaffold that is altered by Wnt3a 36 37 38 stimulation. Adenomatous polyposis coli (APC) is a large (~310kDa) tumor suppressor protein 39 40 that is mutated in most sporadic colorectal cancers in early tumorigenesis. Apart from its role in 41 42 43 Wnt signaling as a member of the β-catenin destruction complex, APC is involved in various 44 45 other cellular processes such as cell migration, cell division, transcriptional regulation and DNA 46 47 57 48 repair . Consequently it has been reported to be localized in many different compartments of 49 50 the cell, including the nucleus, mitochondria, mitotic spindle, centrosome, and 51 52 53 the plasma membrane. Determination of interactors would help to further elucidate these 54 55 diverse APC functions. Previous AP-MS of this protein has provided important information 41, 42 . 56 57 58 59 60 [15]

ACS Paragon Plus Environment Journal of Proteome Research Page 16 of 43

1 2 3 However, systematic study of its interactions has been challenging due to its large size 4 5 6 (necessitating the use of cDNA fragments instead of full-length cDNA) and to the difficulty in 7 8 distinguishing specific from non-specific proteins in the absence of quantitative methods. 9 10 11 12 We analyzed the Wnt3a stimulation dependent and independent APC interactome with the 13 14 triple SILAC strategy described above and show the results of one experiment in Figure 2A. 15 16 17 More than 1,000 quantified proteins are plotted according to their interaction specificity (x- 18 19 axis) against their stimulus specificity (y-axis). Unassigned contaminants are not metabolically 20 21 22 labeled and are therefore readily apparent by their pattern after the label-swapping 23 24 experiment and were removed (Suppl. Figure 4). A large majority of proteins cluster around the 25 26 27 origin, indicating that they bound equally well in the presence or absence of the bait and 28 29 stimulus. We extracted the fold-change distribution of the background binders (Suppl. Figure 2) 30 31 32 and determined significant outliers with box plot statistics (Suppl. Figure 3). Many proteins are 33 34 clearly separated from this background in the x-direction (specific interactors) but not in the 35 36 37 stimulus dependent dimension. These proteins cluster around the x-axis and are colored in 38 39 blue. Eight proteins are in the upper right quadrant, indicating that they bound specifically to 40 41 42 APC and that this binding was increased upon Wnt stimulation (colored in green). Conversely, 43 44 there were nine proteins whose binding to APC decreased upon Wnt stimulation (colored in 45 46 47 red, lower right quadrant). Interestingly, several proteins showed no specific binding to APC 48 49 without stimulus at all but were recruited upon Wnt stimulation. As explained above, these 50 51 52 proteins are more easily visualized when using the ratio of binding to GFP-APC with Wnt 53 54 55 stimulation against control as the x-axis (Figure 2B). In that plot APC binders that are recruited 56 57 in a stimulus dependent manner are located in the upper right quadrant (colored in green). 58 59 60 [16]

ACS Paragon Plus Environment Page 17 of 43 Journal of Proteome Research

1 2 3 In total we performed four biological replicates, two of these with swaped heavy and light 4 5 6 labels (Experimental Procedures). To statistically assess the biological reproducibility of the APC 7 8 interactions, we employed a one sample t-test and separately plotted the median of all protein 9 10 11 ratios from the four replicates in volcano plots for the cases without and with Wnt stimulation 12 13 (Figure 3A and B). This revealed 28 reproducible APC interactors (p-value < 0.1 combined with a 14 15 16 minimum ratio of 4). Inspecting the corresponding region in the left hand side of the graph 17 18 19 revealed no and one protein for the stimulated and unstimulated case, respectively (Figure 3), 20 21 consistent with a false-positive rate of a few percent. Next, we created a heat map of the 22 23 24 median ratios of all proteins (Suppl. Figure 6). The 28 reproducible APC interactors clustered 25 26 together in two subgroups (Figure 4). We also visualized the p-values determined from the t- 27 28 29 tests as a heat map (Figure 4). Together, these two heat maps conveniently combine the 30 31 information obtained from the replicate triple pull-down experiments. 32 33 34 Our APC interactome includes well-known binders such as β-catenin and the transcription 35 36 37 regulator CtBP2. It covers proteins with GO cellular component annotation of all described APC 38 39 localizations (Suppl. Table 1). For example, the novel APC interactor Cep170 localizes to the 40 41 42 centrosome 58 and another novel interactor, the kinesin family member KIF2A, localizes to 43 44 microtubules59 . The novel protein FAM73A had no known compartmental localization. 45 46 47 Microscopy of a GFP-BAC line of this protein showed co-staining with mitochondrial outer 48 49 membranes (Suppl. Figure 5). 50 51 52 53 Significant APC binders also include Axin-1, CKI-α and Wtx, the known binding partners of 54 55 APC in the cytoplasmic β-catenin destruction complex. This complex is usually thought to partly 56 57 58 59 60 [17]

ACS Paragon Plus Environment Journal of Proteome Research Page 18 of 43

1 2 3 disassemble upon Wnt stimulation, although evidence against this has also been reported 60 . 4 5 6 Interestingly, in our triple SILAC experiments, we did not detect dynamic APC interaction 7 8 changes for members of this complex upon Wnt pathway activation with Wnt3a, at least not 9 10 11 after two hours of stimulation. 12 13 14 While most specific interactors showed stimulus independent binding to APC, we also 15 16 17 identified dynamic interaction changes upon Wnt3a activation (Figure 2 and 4) such as the 18 19 enhanced binding of APC to β-catenin. Since the major mechanism of Wnt activation is 20 21 22 stabilization of β-catenin leading to its accumulation, this could simply be the result of more 23 24 available β-catenin. We likewise observed increased binding of α-catenin to APC. α-catenin is 25 26 61 27 reported to indirectly associate with APC via β-catenin , these three proteins contact each 28 29 other at ends 62 . Cytosolic complexes of α- and β-catenin have also been 30 31 63 32 described . Thus dynamic α-catenin binding to APC is most likely due to its association with 33 34 increased levels of β-catenin. Furthermore, APC binds β-catenin not only within the destruction 35 36 37 complex but also in the nucleus to enhance β-catenin nuclear export in the non-Wnt activated 38 39 cell 64 . 40 41 42 43 Additional dynamic interactors of APC with increased binding upon Wnt3a stimulation 44 45 included ATAD3A and ATAD3B. These paralogs have been reported to be localized in 46 47 65, 66 48 mitochondria, but they expose a cytosolic AAA domain . ERBB2-interacting protein (Erbin) is 49 50 a novel interactor, which potentially links APC and ERBB2 signaling. Furthermore, there is 51 52 67, 53 evidence that Erbin binds to β-catenin and negatively regulates Wnt induced gene expression 54 55 68 . 56 57 58 59 60 [18]

ACS Paragon Plus Environment Page 19 of 43 Journal of Proteome Research

1 2 3 We identified Girdin as a novel APC interactor that exclusively binds upon Wnt3a activation 4 5 69 6 (compare Figure 2 A and B). Girdin is a known substrate and regulator of Akt signaling . 7 8 Furthermore, Girdin is a paralog of Daple, which has been reported to interact with the central 9 10 11 Wnt signal mediator DVL, through its Gly-Cys-Val C-terminal motif 69 . Interestingly, however, 12 13 Girdin does not have this motif and therefore at least it must bind DVL in a different manner. 14 15 16 17 In the nucleus APC competes with the transcription factor TCF for β-catenin and the APC- 18 19 β-catenin complex is then thought to bind CtBP2 70 . The APC-β-catenin-CtBP2 complex reduces 20 21 22 the pool of β-catenin that can bind to TCF factors and thereby represses Wnt-dependent gene 23 24 expression. The transcriptional regulator CtBP2 displays the exact opposite APC binding 25 26 27 dynamics to Girdin. Upon Wnt3a activation, this protein is released from its association with 28 29 APC. Colorectal cell lines with truncated APC have diminished binding of APC to CtBP2 70 , which 30 31 32 therefore contributes to increased expression of Wnt target genes. Our observation that the 33 34 APC-CtBP2 interaction is lost upon Wnt activation demonstrates that the truncation of APC is 35 36 37 mechanistically equivalent to stimulation by the Wnt ligand in abolishing CtBP2 binding to the 38 39 C-terminal part of APC. 40 41 42 43 Our APC interactome also contained WDR26 and MAEA, whose binding was diminished 44 45 upon pathway activation. These proteins had previously been found in an Axin-1 interaction 46 47 41 48 screen . Because APC and Axin-1 each have important roles in the destruction complex, 49 50 binders to both proteins are more likely to also have Wnt related functions. This motivated us 51 52 53 to investigate if more APC interactors might be linked to Wnt signaling in the same way. We 54 55 56 57 58 59 60 [19]

ACS Paragon Plus Environment Journal of Proteome Research Page 20 of 43

1 2 3 therefore performed dynamic interaction screens for other Wnt signaling components with the 4 5 6 aim of integrating their interactomes. 7 8 9 The dynamic Axin-1 interactome reveals shared APC and Axin-1 interactors linking them 10 11 12 to Wnt signaling . We determined dynamic Axin-1 interactors by triple SILAC pull-downs with 13 14 and without two hour Wnt3a activation (Figure 1A). Experiments were done in biological 15 16 17 duplicates with swapped SILAC labeling (Suppl. Table 1 and Suppl. Figure 7). We required a 18 19 minimum ratio of four for significant interactors, which was determined from box plot statistics. 20 21 22 Results are summarized in Figure 5A in the heat map format. In total we identified 18 specific 23 24 Axin-1 interactors that were present in both duplicates. We extensively cover β-catenin 25 26 27 destruction complex component members such as APC, CKI-α, Wtx, GSK-3β and PP2A, all of 28 29 which turned out to bind constitutively to Axin-1, in agreement with a recent report 60 . 30 31 32 Additionally, we found moderately increased Axin-1 interaction of SKP1 and β-TrCP2. Both of 33 34 these proteins are members of the ubiquitin ligase complex that targets β-catenin for 35 36 71 37 proteasomal degradation . CDK1δ is known to phosphorylate and activate DVL after Wnt3a 38 39 activation 72, 73 and our data show that its binding to Axin-1 does not depend on Wnt activation. 40 41 42 Rho GTPase activating protein 21 (RhoGAP21) was reported as a β-catenin interactor on the 43 44 basis of TAP pull-downs41 and as α-catenin interactor that is required for α-catenin recruitment 45 46 47 to adherens junctions 74 . We identify this protein as a dynamic interactor to Axin-1, whose 48 49 binding is markedly enhanced by Wnt activation. Further supporting its role in Wnt signaling is 50 51 52 the observation that RhoGAP21 is also a dynamic interactor of APC (Suppl. Table 2). However, 53 54 55 because of our stringent identification criteria it only appears in the final Axin-1 and not the 56 57 APC interactome results. 58 59 60 [20]

ACS Paragon Plus Environment Page 21 of 43 Journal of Proteome Research

1 2 3 Comparing the Axin-1 and APC interactomes revealed ten shared interactors (Figure 6). 4 5 6 Remarkably, the interaction dynamics of each shared component with either of these proteins 7 8 was highly similar, which suggests that they interact with an APC-Axin-1 complex. Among the 9 10 11 shared interactors are the known destruction complex members CKI-α and Wtx as well as the 12 13 ubiquitin ligase component SKP1. We detect enhanced binding of β-catenin to APC and Axin-1, 14 15 16 which as noted above may partly reflect β-catenin accumulation upon stimulation. Moreover a 17 18 19 complex of these three proteins has been reported to localize at the membrane after Wnt3a 20 21 activation 75 . The interaction of Axin-1 with α-catenin was greatly enhanced by Wnt3a 22 23 24 stimulation, similar to its binding to APC. Intriguingly Girdin is also a highly significant dynamic 25 26 interactor of Axin-1 and APC upon Wnt3a activation. The novel protein FAM83B interacts 27 28 29 constitutively with both Axin-1 and APC like other destruction complex members (and may 30 31 therefore be a novel member) and is very likely involved in Wnt signaling. 32 33 34 The DVL2 interactome reveals constitutive binding of Girdin independent of Wnt 35 36 37 activation. Disheveled (DVL) proteins are important Wnt signal mediators that counteract 38 39 destruction complex action upon Wnt3a stimulation. The exact mechanism by which the Wnt 40 41 42 signal is transduced - including the complex formation at the membrane, which involves 43 44 phosphorylated DVL - is still not fully understood 38 . Furthermore, DVL proteins integrate 45 46 47 different branches of Wnt signaling including the planar cell polarity (PCP) pathway. Triple SILAC 48 49 DVL2 pull-downs covered known DVL2 binder such as DVL3 and the positive Wnt regulator 50 51 52 CKI-ε 76 . The negative Wnt PCP regulator Vang-like protein 1/ Strabismus 2 77, 78 , was also a 53 54 55 significant interactor of DVL2. None of the ten identified interactions were modulated by the 56 57 Wnt signal (Figure 5B, Suppl. Figure 8). Among the newly discovered interactors, we found the 58 59 60 [21]

ACS Paragon Plus Environment Journal of Proteome Research Page 22 of 43

1 2 3 three members of the BTB/POZ domain-containing protein family KCTD10/KCTD13/TNFAIP1 79 . 4 5 6 We detect Girdin as a novel and specific interactor for DVL2. In contrast to its dynamically 7 8 increased binding to APC and Axin-1 upon Wnt3a stimulation, Girdin binds constitutively to 9 10 11 DVL2. 12 13 14 CtBP2 binds to β-catenin in a non-stimulus dependent manner. We determined the 15 16 17 dynamic CtBP2 interactome to characterize a potentially dynamic nuclear regulator of Wnt 18 19 signaling. We detected 4 specific interactors of CtBP2 but none of them with significant 20 21 22 stimulus dependent changes (Figure 5C, Suppl. Figure 8). Among these was β-catenin, which is 23 24 known to interact with CtBP2 as well as with APC to repress transcription of Wnt target genes 25 26 27 (see also above). Upon Wnt stimulation β-catenin interacts with CtBP2 and TCF to contribute to 28 29 transcriptional activation 80 . Our observation that the β-catenin–CtBP2 interaction is not 30 31 32 dynamically regulated by Wnt, agrees with suggestions that these two proteins function in both 33 34 repression 70 and activation of gene expression 80 . 35 36 37 38 CONCLUSION AND OUTLOOK 39 40 41 42 Here we have described a three-state quantitative proteomics approach to study the 43 44 dynamics of protein-protein interactions. We used the SILAC technology because of its 45 46 47 simplicity and accuracy of quantification when coupled to a high resolution mass spectrometric 48 49 readout. Proteins were expressed as GFP fusions from bacterial artificial chromosomes that had 50 51 52 been integrated into the host cell genome, ensuring close to endogenous expression levels, 53 54 correct modification state and compartmentalization of the bait proteins. This is especially 55 56 57 important for studying a signaling pathway such as Wnt, in which the regulation of protein 58 59 60 [22]

ACS Paragon Plus Environment Page 23 of 43 Journal of Proteome Research

1 2 3 amount is critical and in which the signal propagates through different cellular locations. The 4 5 6 approach described here is generic as BAC-GFP cell lines can be produced in a streamlined 7 8 procedure 27 . We employed one-dimensional gel separation using somewhat more material 9 10 11 (four 15 cm dishes per condition) than in single run analyses 23 . Analysis of the results was more 12 13 complex than double-labeling SILAC because three states are compared. However, these 14 15 16 analysis steps have now been incorporated into the freely available MaxQuant environment or 17 18 19 as R-scripts. Consequently, dynamic analysis of interaction partners is relatively streamlined 20 21 and it can now be used routinely for pathways of interest or as a follow up on initial high- 22 23 24 throughput protein interaction screens. 25 26 27 There are several obvious extensions of the workflow described here. For example, 28 29 dynamic interaction measurements can be repeated at different time points to investigate the 30 31 32 changing composition of signaling complexes over time. In this experiment the three SILAC 33 34 states can each represent different time points because the specificity of the binders has 35 36 37 already been established. 38 39 40 Here we have applied the QUBIC triple SILAC dynamic interaction screen to the challenging 41 42 43 case of Wnt signaling. We performed pull-downs on central members of the pathway from the 44 45 destruction complex and from other different pathway levels. The SILAC ratios efficiently 46 47 48 filtered out non-specific binders, reducing an initial set of about 1,000 identified proteins to a 49 50 relatively small number (10 to 50). These proteins contained many positive controls that were 51 52 53 either known interaction partners or that already had some other connection to Wnt signaling. 54 55 Among the novel interaction partners those shared by APC and Axin-1 are most likely to be 56 57 58 59 60 [23]

ACS Paragon Plus Environment Journal of Proteome Research Page 24 of 43

1 2 3 functional members of the Wnt pathway. Interestingly, almost all of these interaction partners 4 5 6 turned out to have similar interaction dynamics, consistent with a role in a shared complex with 7 8 APC and Axin-1. Other dynamic interaction partners of APC or Axin-1 are also good candidates 9 10 11 for functional roles in this pathway by virtue of their Wnt-dependent interaction modulation. 12 13 14 One example is Girdin (CCDC88A), which we separately found as a novel interactor of APC, 15 16 17 Axin-1 and Disheveled 2 (DVL2). Intriguingly, interaction with APC and Axin-1 are contingent on 18 19 Wnt-stimulation, whereas interaction with DVL2 is not. This raises the possibility that Girdin 20 21 22 and DVL2 are in a pre-formed complex, which may then recruit destruction complex members 23 24 to the membrane receptors upon Wnt pathway stimulation. In this context, we observed that 25 26 27 interaction of APC and Axin-1 with other destruction complex members did not change upon 28 29 Wnt3a activation. This finding sheds some light on the unresolved mechanism of destruction 30 31 60 32 complex dynamics at the plasma membrane. In concordance with recent observations , our 33 34 data is consistent with a potential translocation of a relatively intact destruction complex, at 35 36 37 least after two hours of Wnt stimulation. 38 39 40 In conclusion, we have described a streamlined interaction screen, which accurately 41 42 43 discriminates constitutive from dynamic, signal dependent interactions. In contrast to targeted 44 45 techniques, such as western blotting, it can both discover and characterize such dynamic 46 47 48 interactors in the same experiment. 49 50 51 52 53 54 Acknowledgments - We thank Bianca Splettstoesser and Anja Wehner for technical help and Markus Räschle and 55 56 Nina C. Hubner for critical reading of the manuscript and helpful discussions. Juergen Cox advised on the data 57 58 59 60 [24]

ACS Paragon Plus Environment Page 25 of 43 Journal of Proteome Research

1 2 3 analysis. Natalie Krahmer from the group of Organelle Architecture and Dynamics helped with the microscopy. We 4 5 also thank Ina Poser and Anthony A. Hyman from the Max Planck Institute of Molecular Cell Biology and Genetics 6 7 for kindly providing the GFP-tagged cell lines. This work was supported by Munich Center for Integrated Protein 8 9 10 Science (CIPSM) and by European Commission's 7th Framework Program PROteomics SPECification in Time and 11 12 Space (PROSPECTS, HEALTH-F4-2008-021648). 13 14 15 REFERENCES 16 17 18 1. Gingras, A. C.; Gstaiger, M.; Raught, B.; Aebersold, R., Analysis of protein complexes using mass 19 20 spectrometry. Nature reviews. Molecular cell biology 2007, 8, (8), 645-54. 21 2. Rigaut, G.; Shevchenko, A.; Rutz, B.; Wilm, M.; Mann, M.; Seraphin, B., A generic protein 22 purification method for protein complex characterization and proteome exploration. Nature 23 biotechnology 1999, 17, (10), 1030-2. 24 3. Ranish, J. A.; Yi, E. C.; Leslie, D. M.; Purvine, S. O.; Goodlett, D. R.; Eng, J.; Aebersold, R., The 25 study of macromolecular complexes by quantitative proteomics. Nature genetics 2003, 33, (3), 349-55. 26 4. Blagoev, B.; Kratchmarova, I.; Ong, S. E.; Nielsen, M.; Foster, L. J.; Mann, M., A proteomics 27 28 strategy to elucidate functional protein-protein interactions applied to EGF signaling. Nat Biotechnol 29 2003, 21, (3), 315-8. 30 5. Vermeulen, M.; Hubner, N. C.; Mann, M., High confidence determination of specific protein- 31 protein interactions using quantitative mass spectrometry. Curr Opin Biotechnol 2008, 19, (4), 331-7. 32 6. Trinkle-Mulcahy, L.; Andersen, J.; Lam, Y. W.; Moorhead, G.; Mann, M.; Lamond, A. I., Repo-Man 33 recruits PP1 gamma to chromatin and is essential for cell viability. The Journal of cell biology 2006, 172, 34 35 (5), 679-92. 36 7. Gavin, A. C.; Aloy, P.; Grandi, P.; Krause, R.; Boesche, M.; Marzioch, M.; Rau, C.; Jensen, L. J.; 37 Bastuck, S.; Dumpelfeld, B.; Edelmann, A.; Heurtier, M. A.; Hoffman, V.; Hoefert, C.; Klein, K.; Hudak, M.; 38 Michon, A. M.; Schelder, M.; Schirle, M.; Remor, M.; Rudi, T.; Hooper, S.; Bauer, A.; Bouwmeester, T.; 39 Casari, G.; Drewes, G.; Neubauer, G.; Rick, J. M.; Kuster, B.; Bork, P.; Russell, R. B.; Superti-Furga, G., 40 Proteome survey reveals modularity of the yeast cell machinery. Nature 2006, 440, (7084), 631-6. 41 8. Kocher, T.; Superti-Furga, G., Mass spectrometry-based functional proteomics: from molecular 42 43 machines to protein networks. Nature methods 2007, 4, (10), 807-15. 44 9. Sowa, M. E.; Bennett, E. J.; Gygi, S. P.; Harper, J. W., Defining the human deubiquitinating 45 enzyme interaction landscape. Cell 2009, 138, (2), 389-403. 46 10. Wepf, A.; Glatter, T.; Schmidt, A.; Aebersold, R.; Gstaiger, M., Quantitative interaction 47 proteomics using mass spectrometry. Nature methods 2009, 6, (3), 203-5. 48 11. Mak, A. B., Ni, Z., Hewel, J. A , Chen, G. I. , Zhong, G., Karamboulas, K. , Blakely, K. , Smiley, S. , 49 50 Marcon, E. , Roudeva, D. , Li, J., Olsen, J. B., Punna, T. , Isserlin, R. ,Chetyrkin, S., Gingras, A. C. , Emili, A., 51 .Greenblatt, J, and Moffat, J. , A lentiviral-based functional proteomics approach identifies chromatin 52 remodelling complexes important for the induction of pluripotency Mol Cell Proteomics 2010, In press. 53 12. Paul, F. E.; Hosp, F.; Selbach, M., Analyzing protein-protein interactions by quantitative mass 54 spectrometry. Methods (San Diego, Calif 2011, 54, (4), 387-95. 55 13. Paoletti, A. C.; Parmely, T. J.; Tomomori-Sato, C.; Sato, S.; Zhu, D.; Conaway, R. C.; Conaway, J. 56 W.; Florens, L.; Washburn, M. P., Quantitative proteomic analysis of distinct mammalian Mediator 57 58 59 60 [25]

ACS Paragon Plus Environment Journal of Proteome Research Page 26 of 43

1 2 3 complexes using normalized spectral abundance factors. Proceedings of the National Academy of 4 Sciences of the United States of America 2006, 103, (50), 18928-33. 5 6 14. Rinner, O.; Mueller, L. N.; Hubalek, M.; Muller, M.; Gstaiger, M.; Aebersold, R., An integrated 7 mass spectrometric and computational framework for the analysis of protein interaction networks. 8 Nature biotechnology 2007, 25, (3), 345-52. 9 15. Wang, T.; Gu, S.; Ronni, T.; Du, Y. C.; Chen, X., In vivo dual-tagging proteomic approach in 10 studying signaling pathways in immune response. Journal of proteome research 2005, 4, (3), 941-9. 11 16. Mousson, F.; Kolkman, A.; Pijnappel, W. W.; Timmers, H. T.; Heck, A. J., Quantitative proteomics 12 13 reveals regulation of dynamic components within TATA-binding protein (TBP) transcription complexes. 14 Mol Cell Proteomics 2008, 7, (5), 845-52. 15 17. Bantscheff, M.; Eberhard, D.; Abraham, Y.; Bastuck, S.; Boesche, M.; Hobson, S.; Mathieson, T.; 16 Perrin, J.; Raida, M.; Rau, C.; Reader, V.; Sweetman, G.; Bauer, A.; Bouwmeester, T.; Hopf, C.; Kruse, U.; 17 Neubauer, G.; Ramsden, N.; Rick, J.; Kuster, B.; Drewes, G., Quantitative chemical proteomics reveals 18 mechanisms of action of clinical ABL kinase inhibitors. Nature biotechnology 2007, 25, (9), 1035-44. 19 18. Selbach, M.; Mann, M., Protein interaction screening by quantitative immunoprecipitation 20 21 combined with knockdown (QUICK). Nature methods 2006, 3, (12), 981-3. 22 19. Selbach, M.; Paul, F. E.; Brandt, S.; Guye, P.; Daumke, O.; Backert, S.; Dehio, C.; Mann, M., Host 23 cell interactome of tyrosine-phosphorylated bacterial proteins. Cell host & microbe 2009, 5, (4), 397-403. 24 20. Vermeulen, M.; Eberl, H. C.; Matarese, F.; Marks, H.; Denissov, S.; Butter, F.; Lee, K. K.; Olsen, J. 25 V.; Hyman, A. A.; Stunnenberg, H. G.; Mann, M., Quantitative interaction proteomics and genome-wide 26 profiling of epigenetic histone marks and their readers. Cell 2010, 142, (6), 967-80. 27 28 21. Breitkreutz, A.; Choi, H.; Sharom, J. R.; Boucher, L.; Neduva, V.; Larsen, B.; Lin, Z. Y.; Breitkreutz, 29 B. J.; Stark, C.; Liu, G.; Ahn, J.; Dewar-Darch, D.; Reguly, T.; Tang, X.; Almeida, R.; Qin, Z. S.; Pawson, T.; 30 Gingras, A. C.; Nesvizhskii, A. I.; Tyers, M., A global protein kinase and phosphatase interaction network 31 in yeast. Science (New York, N.Y 2010, 328, (5981), 1043-6. 32 22. Choi, H.; Larsen, B.; Lin, Z. Y.; Breitkreutz, A.; Mellacheruvu, D.; Fermin, D.; Qin, Z. S.; Tyers, M.; 33 Gingras, A. C.; Nesvizhskii, A. I., SAINT: probabilistic scoring of affinity purification-mass spectrometry 34 data. Nature methods 2011, 8, (1), 70-3. 35 36 23. Hubner, N. C., Bird A. W. , Cox J., Splettstoesser B., Bandilla P., Poser I., Hyman A., Mann M., 37 Quantitative proteomics combined with BAC TransgeneOmics reveals in-vivo protein interactions. JCB 38 2010, in press. 39 24. Zhang, Y.; Buchholz, F.; Muyrers, J. P.; Stewart, A. F., A new logic for DNA engineering using 40 recombination in Escherichia coli. Nat Genet 1998, 20, (2), 123-8. 41 25. Kittler, R.; Pelletier, L.; Ma, C.; Poser, I.; Fischer, S.; Hyman, A. A.; Buchholz, F., RNA interference 42 rescue by bacterial artificial chromosome transgenesis in mammalian tissue culture cells. Proceedings of 43 44 the National Academy of Sciences of the United States of America 2005, 102, (7), 2396-401. 45 26. Sarov, M.; Schneider, S.; Pozniakovski, A.; Roguev, A.; Ernst, S.; Zhang, Y.; Hyman, A. A.; Stewart, 46 A. F., A recombineering pipeline for functional genomics applied to Caenorhabditis elegans. Nature 47 methods 2006, 3, (10), 839-44. 48 27. Poser, I.; Sarov, M.; Hutchins, J. R.; Heriche, J. K.; Toyoda, Y.; Pozniakovsky, A.; Weigl, D.; 49 Nitzsche, A.; Hegemann, B.; Bird, A. W.; Pelletier, L.; Kittler, R.; Hua, S.; Naumann, R.; Augsburg, M.; 50 51 Sykora, M. M.; Hofemeister, H.; Zhang, Y.; Nasmyth, K.; White, K. P.; Dietzel, S.; Mechtler, K.; Durbin, R.; 52 Stewart, A. F.; Peters, J. M.; Buchholz, F.; Hyman, A. A., BAC TransgeneOmics: a high-throughput method 53 for exploration of protein function in mammals. Nat Methods 2008, 5, (5), 409-15. 54 28. Charbonnier, S.; Gallego, O.; Gavin, A. C., The social network of a cell: recent advances in 55 interactome mapping. Biotechnology annual review 2008, 14, 1-28. 56 57 58 59 60 [26]

ACS Paragon Plus Environment Page 27 of 43 Journal of Proteome Research

1 2 3 29. Ong, S. E.; Blagoev, B.; Kratchmarova, I.; Kristensen, D. B.; Steen, H.; Pandey, A.; Mann, M., 4 Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to 5 6 expression proteomics. Mol Cell Proteomics 2002, 1, (5), 376-86. 7 30. Mann, M., Functional and quantitative proteomics using SILAC. Nat Rev Mol Cell Biol 2006, 7, 8 (12), 952-8. 9 31. Blagoev, B.; Ong, S. E.; Kratchmarova, I.; Mann, M., Temporal analysis of phosphotyrosine- 10 dependent signaling networks by quantitative proteomics. Nat Biotechnol 2004, 22, (9), 1139-45. 11 32. Andersen, J. S.; Lam, Y. W.; Leung, A. K.; Ong, S. E.; Lyon, C. E.; Lamond, A. I.; Mann, M., 12 13 Nucleolar proteome dynamics. Nature 2005, 433, (7021), 77-83. 14 33. Olsen, J. V.; Blagoev, B.; Gnad, F.; Macek, B.; Kumar, C.; Mortensen, P.; Mann, M., Global, in vivo, 15 and site-specific phosphorylation dynamics in signaling networks. Cell 2006, 127, (3), 635-48. 16 34. Trinkle-Mulcahy, L.; Boulon, S.; Lam, Y. W.; Urcia, R.; Boisvert, F. M.; Vandermoere, F.; Morrice, 17 N. A.; Swift, S.; Rothbauer, U.; Leonhardt, H.; Lamond, A., Identifying specific protein interaction 18 partners using quantitative mass spectrometry and bead proteomes. The Journal of cell biology 2008, 19 183, (2), 223-39. 20 21 35. Boulon, S.; Ahmad, Y.; Trinkle-Mulcahy, L.; Verheggen, C.; Cobley, A.; Gregor, P.; Bertrand, E.; 22 Whitehorn, M.; Lamond, A. I., Establishment of a protein frequency library and its application in the 23 reliable identification of specific protein interaction partners. Mol Cell Proteomics 2009 . 24 36. Clevers, H., Wnt/beta-catenin signaling in development and disease. Cell 2006, 127, (3), 469-80. 25 37. Klaus, A.; Birchmeier, W., Wnt signalling and its impact on development and cancer. Nat Rev 26 Cancer 2008, 8, (5), 387-98. 27 28 38. Angers, S.; Moon, R. T., Proximal events in Wnt signal transduction. Nature reviews 2009, 10, (7), 29 468-77. 30 39. MacDonald, B. T.; Tamai, K.; He, X., Wnt/beta-catenin signaling: components, mechanisms, and 31 diseases. Developmental cell 2009, 17, (1), 9-26. 32 40. van Amerongen, R.; Nusse, R., Towards an integrated view of Wnt signaling in development. 33 Development (Cambridge, England) 2009, 136, (19), 3205-14. 34 41. Major, M. B.; Camp, N. D.; Berndt, J. D.; Yi, X.; Goldenberg, S. J.; Hubbert, C.; Biechele, T. L.; 35 36 Gingras, A. C.; Zheng, N.; Maccoss, M. J.; Angers, S.; Moon, R. T., Wilms tumor suppressor WTX 37 negatively regulates WNT/beta-catenin signaling. Science (New York, N.Y 2007, 316, (5827), 1043-6. 38 42. Major, M. B.; Roberts, B. S.; Berndt, J. D.; Marine, S.; Anastas, J.; Chung, N.; Ferrer, M.; Yi, X.; 39 Stoick-Cooper, C. L.; von Haller, P. D.; Kategaya, L.; Chien, A.; Angers, S.; MacCoss, M.; Cleary, M. A.; 40 Arthur, W. T.; Moon, R. T., New regulators of Wnt/beta-catenin signaling revealed by integrative 41 molecular screening. Science signaling 2008, 1, (45), ra12. 42 43. Sue Ng, S.; Mahmoudi, T.; Li, V. S.; Hatzis, P.; Boersema, P. J.; Mohammed, S.; Heck, A. J.; 43 44 Clevers, H., MAP3K1 functionally interacts with Axin1 in the canonical Wnt signalling pathway. Biological 45 chemistry 2010, 391, (2-3), 171-80. 46 44. Shevchenko, A.; Wilm, M.; Vorm, O.; Mann, M., Mass spectrometric sequencing of proteins 47 silver-stained polyacrylamide gels. Analytical chemistry 1996, 68, (5), 850-8. 48 45. Shevchenko, A.; Tomas, H.; Havlis, J.; Olsen, J. V.; Mann, M., In-gel digestion for mass 49 spectrometric characterization of proteins and proteomes. Nat Protoc 2006, 1, (6), 2856-60. 50 51 46. Rappsilber, J.; Ishihama, Y.; Mann, M., Stop and go extraction tips for matrix-assisted laser 52 desorption/ionization, nanoelectrospray, and LC/MS sample pretreatment in proteomics. Anal Chem 53 2003, 75, (3), 663-70. 54 47. Rappsilber, J.; Mann, M.; Ishihama, Y., Protocol for micro-purification, enrichment, pre- 55 fractionation and storage of peptides for proteomics using StageTips. Nature protocols 2007, 2, (8), 56 1896-906. 57 58 59 60 [27]

ACS Paragon Plus Environment Journal of Proteome Research Page 28 of 43

1 2 3 48. Cox, J.; Mann, M., MaxQuant enables high peptide identification rates, individualized p.p.b.- 4 range mass accuracies and proteome-wide protein quantification. Nat Biotechnol 2008, 26, (12), 1367- 5 6 72. 7 49. Cox, J.; Matic, I.; Hilger, M.; Nagaraj, N.; Selbach, M.; Olsen, J. V.; Mann, M., A practical guide to 8 the MaxQuant computational platform for SILAC-based quantitative proteomics. Nature protocols 2009, 9 4, (5), 698-705. 10 50. Team, R. D. C., R: a language and environment for statistical computing . R Foundation for 11 Statistical Computing: Vienna, Austria, 2008. 12 13 51. Carbon, S.; Ireland, A.; Mungall, C. J.; Shu, S.; Marshall, B.; Lewis, S., AmiGO: online access to 14 ontology and annotation data. Bioinformatics (Oxford, England) 2009, 25, (2), 288-9. 15 52. Cline, M. S.; Smoot, M.; Cerami, E.; Kuchinsky, A.; Landys, N.; Workman, C.; Christmas, R.; Avila- 16 Campilo, I.; Creech, M.; Gross, B.; Hanspers, K.; Isserlin, R.; Kelley, R.; Killcoyne, S.; Lotia, S.; Maere, S.; 17 Morris, J.; Ono, K.; Pavlovic, V.; Pico, A. R.; Vailaya, A.; Wang, P. L.; Adler, A.; Conklin, B. R.; Hood, L.; 18 Kuiper, M.; Sander, C.; Schmulevich, I.; Schwikowski, B.; Warner, G. J.; Ideker, T.; Bader, G. D., 19 Integration of biological networks and gene expression data using Cytoscape. Nature protocols 2007, 2, 20 21 (10), 2366-82. 22 53. Shannon, P.; Markiel, A.; Ozier, O.; Baliga, N. S.; Wang, J. T.; Ramage, D.; Amin, N.; Schwikowski, 23 B.; Ideker, T., Cytoscape: a software environment for integrated models of biomolecular interaction 24 networks. Genome research 2003, 13, (11), 2498-504. 25 54. Hubner, N. C.; Bird, A. W.; Cox, J.; Splettstoesser, B.; Bandilla, P.; Poser, I.; Hyman, A.; Mann, M., 26 Quantitative proteomics combined with BAC TransgeneOmics reveals in vivo protein interactions. J Cell 27 28 Biol 2010, 189, (4), 739-54. 29 55. Wang, X.; Huang, L., Identifying dynamic interactors of protein complexes by quantitative mass 30 spectrometry. Mol Cell Proteomics 2008, 7, (1), 46-57. 31 56. Olsen, J. V.; Schwartz, J. C.; Griep-Raming, J.; Nielsen, M. L.; Damoc, E.; Denisov, E.; Lange, O.; 32 Remes, P.; Taylor, D.; Splendore, M.; Wouters, E. R.; Senko, M.; Makarov, A.; Mann, M.; Horning, S., A 33 dual pressure linear ion trap orbitrap instrument with very high sequencing speed. Molecular & cellular 34 proteomics : MCP 2009, 8, (12), 2759-69. 35 36 57. Brocardo, M.; Henderson, B. R., APC shuttling to the membrane, nucleus and beyond. Trends in 37 cell biology 2008, 18, (12), 587-96. 38 58. Guarguaglini, G.; Duncan, P. I.; Stierhof, Y. D.; Holmstrom, T.; Duensing, S.; Nigg, E. A., The 39 forkhead-associated domain protein Cep170 interacts with Polo-like kinase 1 and serves as a marker for 40 mature centrioles. Molecular biology of the cell 2005, 16, (3), 1095-107. 41 59. Ganem, N. J.; Compton, D. A., The KinI kinesin Kif2a is required for bipolar spindle assembly 42 through a functional relationship with MCAK. The Journal of cell biology 2004, 166, (4), 473-8. 43 44 60. Yokoyama, N.; Yin, D.; Malbon, C. C., Abundance, complexation, and trafficking of Wnt/beta- 45 catenin signaling elements in response to Wnt3a. Journal of molecular signaling 2007, 2, 11. 46 61. Rubinfeld, B.; Souza, B.; Albert, I.; Munemitsu, S.; Polakis, P., The APC protein and E-cadherin 47 form similar but independent complexes with alpha-catenin, beta-catenin, and . The Journal 48 of biological chemistry 1995, 270, (10), 5549-55. 49 62. Harris, T. J.; Peifer, M., Decisions, decisions: beta-catenin chooses between adhesion and 50 51 transcription. Trends in cell biology 2005, 15, (5), 234-7. 52 63. Bienz, M., beta-Catenin: a pivot between cell adhesion and Wnt signalling. Curr Biol 2005, 15, 53 (2), R64-7. 54 64. Henderson, B. R., Nuclear-cytoplasmic shuttling of APC regulates beta-catenin subcellular 55 localization and turnover. Nature cell biology 2000, 2, (9), 653-60. 56 65. Bogenhagen, D. F.; Rousseau, D.; Burke, S., The layered structure of human mitochondrial DNA 57 nucleoids. The Journal of biological chemistry 2008, 283, (6), 3665-75. 58 59 60 [28]

ACS Paragon Plus Environment Page 29 of 43 Journal of Proteome Research

1 2 3 66. He, J.; Mao, C. C.; Reyes, A.; Sembongi, H.; Di Re, M.; Granycome, C.; Clippingdale, A. B.; 4 Fearnley, I. M.; Harbour, M.; Robinson, A. J.; Reichelt, S.; Spelbrink, J. N.; Walker, J. E.; Holt, I. J., The 5 6 AAA+ protein ATAD3 has displacement loop binding properties and is involved in mitochondrial nucleoid 7 organization. The Journal of cell biology 2007, 176, (2), 141-6. 8 67. Ress, A.; Moelling, K., The PDZ protein erbin modulates beta-catenin-dependent transcription. 9 Eur Surg Res 2008, 41, (3), 284-9. 10 68. Ress, A.; Moelling, K., Interaction partners of the PDZ domain of erbin. Protein and peptide 11 letters 2006, 13, (9), 877-81. 12 13 69. Enomoto, A.; Ping, J.; Takahashi, M., Girdin, a novel -binding protein, and its family of 14 proteins possess versatile functions in the Akt and Wnt signaling pathways. Annals of the New York 15 Academy of Sciences 2006, 1086, 169-84. 16 70. Hamada, F.; Bienz, M., The APC tumor suppressor binds to C-terminal binding protein to divert 17 nuclear beta-catenin from TCF. Developmental cell 2004, 7, (5), 677-85. 18 71. Liu, C.; Kato, Y.; Zhang, Z.; Do, V. M.; Yankner, B. A.; He, X., beta-Trcp couples beta-catenin 19 phosphorylation-degradation and regulates Xenopus axis formation. Proceedings of the National 20 21 Academy of Sciences of the United States of America 1999, 96, (11), 6273-8. 22 72. Bryja, V.; Schulte, G.; Arenas, E., Wnt-3a utilizes a novel low dose and rapid pathway that does 23 not require casein kinase 1-mediated phosphorylation of Dvl to activate beta-catenin. Cellular signalling 24 2007, 19, (3), 610-6. 25 73. Peters, J. M.; McKay, R. M.; McKay, J. P.; Graff, J. M., Casein kinase I transduces Wnt signals. 26 Nature 1999, 401, (6751), 345-50. 27 28 74. Sousa, S.; Cabanes, D.; Archambaud, C.; Colland, F.; Lemichez, E.; Popoff, M.; Boisson-Dupuis, S.; 29 Gouin, E.; Lecuit, M.; Legrain, P.; Cossart, P., ARHGAP10 is necessary for alpha-catenin recruitment at 30 adherens junctions and for Listeria invasion. Nature cell biology 2005, 7, (10), 954-60. 31 75. Hendriksen, J.; Jansen, M.; Brown, C. M.; van der Velde, H.; van Ham, M.; Galjart, N.; Offerhaus, 32 G. J.; Fagotto, F.; Fornerod, M., Plasma membrane recruitment of dephosphorylated beta-catenin upon 33 activation of the Wnt pathway. Journal of cell science 2008, 121, (Pt 11), 1793-802. 34 76. Sakanaka, C.; Leong, P.; Xu, L.; Harrison, S. D.; Williams, L. T., Casein kinase iepsilon in the wnt 35 36 pathway: regulation of beta-catenin function. Proceedings of the National Academy of Sciences of the 37 United States of America 1999, 96, (22), 12548-52. 38 77. Park, M.; Moon, R. T., The planar cell-polarity gene stbm regulates cell behaviour and cell fate in 39 vertebrate embryos. Nature cell biology 2002, 4, (1), 20-5. 40 78. Angers, S.; Thorpe, C. J.; Biechele, T. L.; Goldenberg, S. J.; Zheng, N.; MacCoss, M. J.; Moon, R. T., 41 The KLHL12-Cullin-3 ubiquitin ligase negatively regulates the Wnt-beta-catenin pathway by targeting 42 Dishevelled for degradation. Nature cell biology 2006, 8, (4), 348-57. 43 44 79. Van Bogaert, P.; Azizieh, R.; Desir, J.; Aeby, A.; De Meirleir, L.; Laes, J. F.; Christiaens, F.; 45 Abramowicz, M. J., Mutation of a potassium channel-related gene in progressive myoclonic epilepsy. 46 Annals of neurology 2007, 61, (6), 579-86. 47 80. Fang, M.; Li, J.; Blauwkamp, T.; Bhambhani, C.; Campbell, N.; Cadigan, K. M., C-terminal-binding 48 protein directly activates and represses Wnt transcriptional targets in Drosophila. The EMBO journal 49 2006, 25, (12), 2735-45. 50 51 52 53 54 55 56 57 58 59 60 [29]

ACS Paragon Plus Environment Journal of Proteome Research Page 30 of 43

1 2 3 4 5 6 FIGURE LEGENDS 7 8 9 10 11 Figure 1. Analysis of interaction dynamics by QUBIC triple SILAC based quantitative mass 12 13 spectrometry. (A) Experimental workflow for triple SILAC pull-downs to determine Wnt3a 14 15 16 dependent interaction dynamics. The cell line expressing the GFP-tagged protein of interest is 17 18 light and medium SILAC labeled, the untransfected wild-type control cell line is heavy SILAC 19 20 21 labeled. Cells are lysed after two hour treatment with Wnt3a (200 ng/mL) or vehicle solution 22 23 respectively. GFP-pull-downs are performed separately for each SILAC state. Eluates are 24 25 26 combined, separated on a one-dimensional gel into eight slices and in-gel digested. Resulting 27 28 peptide mixtures are analyzed by high resolution LC-MS/MS on an LTQ-Orbitrap Velos. SILAC 29 30 31 ratios are automatically quantified by MaxQuant. (B) SILAC peptide triplets representing peak 32 33 profiles characteristic of background, constitutive and dynamic binders. The complete elution 34 35 36 profile of the isotope cluster of the peptide is shown (m/z scale in x-direction, elution time in y- 37 38 direction and MS signal in z-direction). (C) Data analysis plot of the ratio representing 39 40 41 interaction specificity versus the ratio representing stimulus specificity of the interaction. Filled 42 43 dots represent significant interactors and of these, constitutive interactors are depicted in blue. 44 45 46 Dynamic interactors with enhanced binding to the bait protein are shown in red and those with 47 48 reduced binding to the bait protein in green. 49 50 51 52 Figure 2. Dynamic APC interactome. (A) Results from a triple SILAC pull-down as described in 53 54 Figure 1 using Wnt3a as the stimulus and GFP-APC as bait protein, plotted as explained in 55 56 57 Figure 1C. Annotated filled circles represent specific interactors determined by box plot 58 59 60 [30]

ACS Paragon Plus Environment Page 31 of 43 Journal of Proteome Research

1 2 3 statistics of the fold-change distribution of unstimulated pull-down against control. Specific 4 5 6 dynamic interactors with enhanced binding to APC upon Wnt stimulation are depicted in green; 7 8 the ones with decreased binding are depicted in red. Significance thresholds for dynamic 9 10 11 changes were obtained from a box plot of fold-change distribution of stimulated pull-down 12 13 against unstimulated pull-down. Constitutive, specific interactors are shown in blue. (B) Same 14 15 16 experiment as in (A) but with the fold-change distribution of stimulated pull-down against 17 18 19 control on the x-axis. In this plot dynamic interactors move to the upper right hand quadrant as 20 21 can be seen for the proteins shown in bold. 22 23 24 Figure 3. Volcano plot to determine reproducible APC interactors. (A) Log2 ratios of the median 25 26 27 of four pull-downs of GFP-APC against control (x-axis) are plotted versus –log10 of the p-values 28 29 derived from a t-test. Proteins with a minimum four-fold change combined with a p value 30 31 32 smaller than 0.1 are considered significant (red lines). (B) Same as (A) but for simulated pull- 33 34 downs against controls. 35 36 37 38 Figure 4. Dynamic APC interactome visualized by a heat map and one-way hierarchical 39 40 clustering. The three ratios of the triple SILAC pull-down (median of four experiments) are used 41 42 43 to cluster the reproducible APC interactors (determined in Figure 3) by one-way hierarchical 44 45 clustering. A green color value signifies specific binding to APC without Wnt stimulation (first 46 47 48 column) or with Wnt stimulation (second column) in the heat map. The third column depicts 49 50 the SILAC ratio of stimulated against unstimulated bait pull-down. In this column a green color 51 52 53 value represents enhanced binding to APC upon Wnt activation and a red color represents 54 55 reduced binding. Constitutive interactors have no significant ratio and therefore appear in 56 57 58 59 60 [31]

ACS Paragon Plus Environment Journal of Proteome Research Page 32 of 43

1 2 3 black. All ratio intensities are shown in log scale. Note that only those parts of the cluster that 4 5 6 contain significant binders are depicted. Refer to Suppl. Figure 6 for the complete clustering. 7 8 Additionally, t-test results for the ratio reproducibility (Figure 3) were visualized in the right 9 10 11 panel after the clustering process. Proteins with reproducible ratios (p-value < 0.1) are in blue 12 13 and those above the threshold in yellow. KIFC3 does not meet the significance criteria for 14 15 16 specific and reproducible APC binding (minimum ratio of 4 and p-value < 0.1) both without and 17 18 19 with stimulus and is therefore greyed out in the figure. 20 21 22 Figure 5. Dynamic Axin-1 (A), CtBP2 (B) and DVL2 (C) interactomes. Heat map and one-way 23 24 hierarchical clustering of SILAC ratios from biological duplicates of triple SILAC experiments 25 26 27 performed with label swap (FWD and REV experiment). Only those parts of the cluster with 28 29 significant ratio intensities are depicted. Proteins that do not fulfilling the significance criteria of 30 31 32 reproducible detection with a minimum ratio of 4 in the forward and reverse pulldown for at 33 34 least one stimulus state were greyed out. For color coding of the heat map see Figure 4. All 35 36 37 ratio intensities are shown in log scale. 38 39 40 Figure 6. Overlap of APC and Axin-1 interactomes. Protein-protein interactions were drawn in 41 42 43 Cytoscape, after importing the pull-down data from Figure 4 and Figure 5A. Baits are depicted 44 45 in yellow, shared APC and Axin-1 interactors in blue and unique interactors for APC and Axin-1 46 47 48 in purple and pink, respectively. Lines represent detected interactions. Green lines indicate 49 50 enhanced interaction upon Wnt3a activation while red lines indicate reduced interactions upon 51 52 53 Wnt3a activation. Line width reflects the SILAC ratio intensity for the dynamic interactors. 54 55 56 57 58 59 60 [32]

ACS Paragon Plus Environment Page 33 of 43 Journal of Proteome Research

1 2 3 4 SUPPORTING INFORMATION 5 6 7 Suppl. Figure 1. Activation of the Wnt pathway after incubation with Wnt3a for 2 h. Western 8 9 10 Blot of input material for the affinity purifications from HeLa cells stably expressing BACs with 11 12 human GFP-tagged APC, Axin-1, DVL2 or CtBP2 incubated with Wnt3a or vehicle for 2 h. Wnt 13 14 15 activation is evaluated by β-catenin increase; β-tubulin is used as loading control. 16 17 18 Suppl. Figure 2. Fold-change distribution of the background binders for the GFP-APC triple 19 20 21 SILAC pulldown shown in Figure 2. Density plots of the stimulus specificity ratio (left panel), 22 23 normalized L/M (dark red), and the interaction specificity ratios (right panel), normalized M/H 24 25 26 (orange) and normalized L/H (red). 27 28 29 Suppl. Figure 3. Box plot statistics to define outlier significance for the GFP-APC triple SILAC 30 31 32 pulldown shown in Figure 2. Box plots and their corresponding statistical results are displayed. 33 34 35 Suppl. Figure 4. Filtering procedure that underlies the data analysis plots of the GFP-APC triple 36 37 SILAC pulldown shown in Figure 2. Contaminant proteins that were not assigned as such in the 38 39 40 database were discovered and discarded in the following way. We determined their normalized 41 42 M/L (GFP-APC AP / control AP) and H/L (GFP-APC AP + Wnt3a / control AP) ratios of the 43 44 45 experiments with swaped heavy and light labels and used a cut-off ratio of 0.5 (in orange in 46 47 panel A and B). 48 49 50 51 Suppl. Figure 5. Co-localization of FAM73A with mitochondria. Fluorescence microscopy of GFP- 52 53 FAM73A and mitochondria stained by MitoTracker Red CM-H2XRos. 54 55 56 57 58 59 60 [33]

ACS Paragon Plus Environment Journal of Proteome Research Page 34 of 43

1 2 3 Suppl. Figure 6. Complete heat map visualizing the median of four replicates of GFP-APC triple 4 5 6 SILAC pull-downs. One-sample t-test using the three median SILAC ratios of all identified 7 8 proteins from four biological replicates. 9 10 11 12 Suppl. Figure 7. Biological duplicates of GFP-Axin-1 triple SILAC pull-downs. Ratios of GFP- 13 14 Axin-1 against control of FWD and REV pull-downs are plotted against each other (left panel). 15 16 17 Similarly the ratios of GFP-Axin-1 with Wnt3a against control of FWD and REV pull-downs are 18 19 plotted against each other (right panel). 20 21 22 23 Suppl. Figure 8. Biological duplicates ofGFP-DVL2 and GFP-CtBP2 triple SILAC pull-downs. Ratios 24 25 of GFP-DVL2 with Wnt3a against control (left panel) and GFP-CtBP2 with Wnt3a against control 26 27 28 (right panel) of FWD and REV pull-downs are plotted against each other. 29 30 31 Suppl. Table 1. Specific interactors visualized in the APC, Axin-1, DVL2 and CtBP2 interactome 32 33 34 heat maps. For the Axin REV pull-down we used the ‘width adjustment’ option during the 35 36 clustering process in Perseus. Cell localizations of all interactors are listed. STRING analysis was 37 38 39 performed for the individual interactomes and a statistic on known interactors is shown. 40 41 42 Suppl. Table 2. Protein group files of APC, Axin-1, DVL2 and CtBP2 protein pull-downs. 43 44 45 MaxQuant output files containing all protein information. Files were uploaded to TRANCHE. 46 47 48 Link to the data at TRANCHE: https://proteomecommons.org/tranche/, using the following 49 50 hash: 51 52 53 lGMv8OB/574FanvAfNNlAaB7Iuksi91OAYRdyGfV6bqLqfUD/v6/ZGiiG/GniA7caNksWmUwR7PJrv 54 55 e6Yd9xo9pmwzgAAAAAAAADLg== and the passphrase: MaxiTriple SILAC 56 57 58 59 60 [34]

ACS Paragon Plus Environment Page 35 of 43 Journal of Proteome Research

1 2 3 4 TOC 5 6 7 Protein interactions are often regulated depending on cellular context but a generally studied 8 9 10 in a static format. Here we develop a triple-SILAC based dynamic interaction screen and apply 11 12 it to the Wnt signaling pathway. Our screens identify many known Wnt signaling complex 13 14 15 components and link novel candidates to Wnt signaling, including FAM83B and Girdin, which 16 17 we found as interactors to multiple Wnt pathway players. 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 [35]

ACS Paragon Plus Environment Journal of Proteome Research Page 36 of 43 FIGURES

1 2 Figure 1, Hilger et al. 3 4 5 A. 6 7 Light Medium Heavy 8 9 Arg0, Lys0Arg6, LysD4 Arg10, Lys8 10 11 12 13 14 15 Cell line expressing Cell line expressing Wild-type control 16 GFP-tagged protein GFP-tagged protein cell line 17 18 + Wnt3a (2h) +Vehicle (2h) +Vehicle (2h) 19 20 21 Cell lysis and AP 22 23 24 25 + 26 27 28 29 30 - 31 32 33 34 35 36 37 SDS buffer elution 38 39 40 41 42 43 1D-SDS-gelelectrophoresis 44 45 46 47 48 49 50 51 52 53 54 55 56 In-gel digestion 57 58 59 LC-MS/MS 60 ACS Paragon Plus Environment

Data analysis with MaxQuant Page 37 of 43 Journal of Proteome Research Figure 1, Hilger et al.

1 B. 2 3 Unspecific background binders Bait protein and specific interactors 4 5 Constitutive interactions Stimulus dependent interactions 6 7 8 Reduced binding Enhanced binding 9 10 11 12 13 14 15 16 17 relative intensity relative intensity relative intensity 18 relative intensity 19 20 21 m/z m/z m/z m/z 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 ACS Paragon Plus Environment 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Proteome Research Page 38 of 43 Figure 1, Hilger et al.

1 2 C. 3 GFP-bait + stimulus / GFP-bait 4 GFP-bait + stimulus / control 5 Unspecific Interaction specificity

6 )) background binder GFP-bait / control 7 8 9 10 Dynamic interactors: 11 enhanced binding 12 + stimulus 13 14 15 16 17 18 Specific constitutive 19 binders 20

21 AP + stimulus / GFP-bait AP GFP-bait 22 23 Stimulus specificity

24 (Ratio ( 25 2 Dynamic interactors: 26 log reduced binding 27 + stimulus 28 -4 -2 0 2 4 6 29 -4 -2 0 2 4 6 30 31 log (Ratio (GFP-bait AP / control AP)) 32 2 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment Page 39 of 43 Journal of Proteome Research Figure 2, Hilger et al.

1 2 3 4 5 6 A. B. 7 8 α-Catenin 9 α-Catenin β-Catenin β-Catenin 10 KHDRBS1 KHDRBS1 MYH2 11 Girdin PLEKHA7 DSC1 DSC1 Girdin ATAD3B CCDC88C 12 ATAD3B 13 ATAD3A ATAD3A NCKAP1 NCKAP1 CYFIP1 Juntion plakoglobin Erbin CYFIP1 Juntion plakoglobin 14 Erbin SKP1 β-TrCP2 SKP1 15 CLTC CLTC Plakophilin-4 Plakophilin-4 CLTA KLHL13 KLHL21 PKP3 CLTA KLHL21 16 LIN7C TPR LIN7C IQ GAP 3 IQ GAP 3 Plakophilin-2 FAM83B Plakophilin-2 FAM83B 17 PHLDB2DVL2 CK1α Wtx PHLDB2 CK1α Wtx 18 BCCIPCMP-NeuNAc synthetase APC DVL2 CMP-NeuNAc synthetase APC CHD2 FAM73A PRKCDBP FAM73A KIFC3 Axin-1 PEO1 PRKCDBP Axin-1 KIF2A KIF2A RNF41 19 H3F3A RNF41CEP170 KIFC3 HERC2 PEO1 CEP170 HERC2 APC2 NEURL4 PTRF APC2 NEURL4 TRIM27 PTRF PECI HIST1H1C PECI SSBP1 SSBP1 20 MYCBP2 MAP7D1 MYCBP2 MAP7D1 21 H1FX NAV1 NAV1 CtBP2 WDR26 WDR26 22 AP)) + Wnt3a / GFP-bait AP GFP-bait MAEA AP)) + Wnt3a / GFP-bait AP GFP-bait MAEA 23 24

25 (Ratio ( (Ratio (

26 2 2 -3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3 27 log RRP1B log 28 -2 0 2 4 6 -20246 29 log (Ratio (GFP-bait AP / control AP)) log (Ratio (GFP-bait AP +Wnt 3a/ control AP)) 30 2 2 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 ACS Paragon Plus Environment 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Proteome Research Page 40 of 43 Figure 3, Hilger et al.

1 2 A. B. 3 4 5 6 7 8 9 10 11 12 13 APC

14 HERC2 15 16 17 Axin-1 18 Axin-1 19 CMP-NeuNAc synthetase NEURL4 HERC2 Wtx SSBP1 SSBP1 20 FA M83B APC ATAD3B β-Catenin 21 CKI-α CMP-NeuNAc synthetase Wtx 22 ATAD3A PTRF (p value t-test) (p α-Catenin MA EA value t-test) (p ATAD3A 10

23 10 24 - log - log Juntion plakoglobin CKI-α CLTC 25 WDR26 FA M83B MA P7D1 Juntion plakoglobin 26 PTRF NEURL4 KIF2A Erbin SKP1 KIF2A 27 CEP170 28 FAM73A Plakophilin-2 CtBP2 FA M73A 29 Girdin 30 CEP170 Hsp60 31 32 33 34 35

36 01234 01234 37 38 -4 -2 0 2 4 -4 -2 0 2 4 39 40 log 2 (median ratio (GFP -A P C AP w/o Wnt3a / control AP)) log 2 (median ratio GFP -A P C AP +W nt3a / control AP) 41 42 43 44 45 46 ACS Paragon Plus Environment 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 41 of 43 Journal of Proteome Research Figure 4, Hilger et al.

1 2 3 4 5 6 7 8 9 Intensity ratio Intensity p-value 10 11 GFP-APC + Wnt3a / GFP-APC GFP-APC + Wnt3a / GFP-APC 12 GFP-APC + Wnt3a / control GFP-APC + Wnt3a / control 13 14 15 GFP-APC / control GFP-APC / control 16 17 18 19 20 21 CtBP2 22 MAEA 23 24 WDR26

25 HERC2 + Wnt3a

26 APC - APC binding

27 Wtx 28 Axin-1 29 FAM73A 30 31 SSBP1 32 CMAS 33 KIF2A 34 KIFC3 35 PTRF 36 NEURL4 37 MAP7D1 Static APC interactors 38 - 39 CKI- α 40 FAM83B 41 CEP170 42 PKP2 43 SKP1 44 JUP 45 ATAD3A 46

47 α-Catenin 48 β-Catenin + Wnt3a 49 ATAD3B + APC binding 50 CLTC 51 Girdin 52 Hsp60 53 54 ErbinCLTC 55 no value no value

56 p-value 57 58 59 60 ACS Paragon Plus Environment Journal of Proteome Research Page 42 of 43 Figure 5, Hilger et al.

1 2 3 A. B. C. 4 5

6 GFP-Axin-1 + Wnt3a / GFP-Axin-1GFP-Axin-1 +FWD Wnt3a / GFP-Axin-1 REV GFP-CtBP2 + Wnt3a / GFP-CtBP2GFP-CtBP2 FWD + Wnt3a / GFP-CtBP2 REV GFP-DVL2 + Wnt3a / GFP-DVL2GFP-DVL2 FWD + Wnt3a / GFP-DVL2 REV 7 GFP-Axin-1 + Wnt3a / control FWD GFP-Axin-1 + Wnt3a / control REV GFP-DVL2 + Wnt3a / controlGFP-DVL2 FWD + Wnt3a / control REV GFP-CtBP2+ Wnt3a / controlGFP-CtBP2 FWD + Wnt3a / control REV 8

9 GFP-Axin-1 / control FWD GFP-CtBP2 / control FWD GFP-Axin-1 / control REV GFP-DVL2 / control FWDGFP-DVL2 / control REV GFP-CtBP2 / control REV 10 11 12 13 14 15 16 SSBP1 DVL2 CtBP2 17 GAPVD1 TNFAIP1 CtBP1 18 CKI-δ/ε KCTD10-1 LSD1 19 SNX24 KCTD13 Hic-2 DVL-3 LCORL 20 Axin-1 Strabismus 2 C10orf112 21 Wtx 22 Girdin β-Catenin CKI-α 23 EPB41L2 CoREST CKI-α 24 KCTD10-2 Mark3 GSK-3α 25 CKI-ε Wiz APC 26 ZNF217 GSK-3β 27 H3-K9-HMTase5 28 PP2A-Aα 29 FAM83B 30 ATAD3A 31 β-TrCP2 32 RhoGAP21 33 Girdin 34 α-Catenin 35 β-Catenin 36 SKP1 37 JUP 38 39 40 41 42 43 44 45 46 ACS Paragon Plus Environment 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 43 of 43 Journal of Proteome Research Figure 6, Hilger et al.

1 2 3 4 5 6 7 8 9 10 Erbin 11 Hsp60 12 13 Plakophilin 14 15 16 CLTC 17 18 19 20 CKI-α 21 22 Wtx 23 GSK-3 β 24 25 26 β-Catenin α-Catenin 27 RhoGAP21 28 29 30 31 GSK-3 α 32 Axin-1 33 34 CMP-NeuNAc synthetase 35 β-TrCP2 36 37 Girdin 38 39 CKi-δ 40 41 Juntion 42 plakoglobin PP2A-Aα 43 44 45 46 47 48 49 50 51 52 53 54 CtBP2 55 56 57 58 59 60 ACS Paragon Plus Environment