Mammalian secretome profiling to identify adaptive and maladaptive signaling in homeostasis and diet induced obesity

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Mammalian Secretome Profiling to Identify Adaptive and Maladaptive Signaling in Homeostasis and Diet Induced Obesity by Amanda Stevens Meyer A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (MOLECULAR BIOLOGY) December 2022 Copyright 2022 Amanda Stevens Meyer ii Epigraph “Fulfillment comes from striving to succeed, to survive by your own wits and strength. Such things make each of us who we are.” - Michael J. Sullivan iii Acknowledgements Science requires phenomenal teamwork and dedicated, long-term training by many to be successful. The passion, intensity, and kindness I have experienced from my colleagues and mentors is one of the many, yet highly important reasons to why I love what I do. I am extremely grateful and fortunate to been able to work with and learn from so many incredible individuals over the course of my PhD. Thank you to everyone I have met and to those who have helped shape me as a scientist over these years. Before joining a lab for your PhD, some of the most common and arguably best advice is to choose the advisor, not the research. As anyone who’s spent years on a project knows, you will grow to love your project and, at times, hate it. When you are able to work well with your advisor the science, discussions, and mentoring greatly enhance your PhD. I chose to work with Andy due to the rigor and scientific excellent of his and his lab’s work. Before I started graduate school, I had the pleasure of seeing Andy present at our guest seminar series. Seeing the level of quality and thoroughness of the work was something I aspired to. So, I would like to thank Andy for being exactly the mentor I needed. Throughout my PhD Andy has allowed me to be highly independent and to follow my own path while guiding me along the way. From Andy, I have learned how pursue new veins of research, gained a critical eye for data, and became an independent scientist. Thank you, Andy, for being so supportive and for preparing me well for my career. It has been an absolute pleasure. I would also like to thank my collaborator and mentor, Shingo Kajimura PhD. Shingo is an expert in adipocyte biology and was highly recommended to me by Andy as a mentor for my adipose tissue work. Shingo has been incredibly helpful and provided keen insight into my work on adipose tissue, which I could not have gotten to this point iv without. He also has provided me support and encouragement that gave me the confidence to pursue my adipose tissue work. Thank you so much Shingo for all of your help and support. I am grateful to my thesis committee for their support, curiosity, and interest in my work. It has been a pleasure to learn from so many bright and inspiring scientists. Your feedback has been instrumental in my work. Additionally, being a graduate student in the Department of Molecular and Computational Biology has been highly rewarding. I am extremely happy and satisfied by how much MCB cares and supports its students in their scientific training and in who they are as people. I could not have done my PhD without the support of many friends and family. In particular, I would like to thank my fellow labmate, climber, and friend, Riana Parvez (Feng). Working with and being in the same lab as Riana has made my PhD so enjoyable. I could not have asked for a better partner in crime. Without Riana, I don’t know I would have ever gotten through the infamous “third year slump” or survived the mornings. I have enjoyed our morning (and afternoon.. and evening) teatimes, discussing our projects, and getting through grad school together. I have loved learning so much about the collecting duct and intercalated cells and will greatly miss climbing with you. Your morning shift notes left on my desk helped me get through the night shift during covid along with our zoom “teatimes”. Our PhD years have been nothing other than memorable and I am so happy to have shared them with you. Riana, thank you for everything. I also want to thank my dear friend Chai (Shoujie Chai). Meeting Chai on the first day of grad school was one of the best things that happened. We quickly became friends, and I am so lucky in that regard. Chai has given me the encouragement and believed in my even when I have not. I have been inspired by Chai’s work ethic, dedication, and v passion. Chai has always been someone I have looked up to and sought guidance from. Thank you, Chai, for all of our fond memories and for always being an amazingly kind friend. Thank you to my dear friend, Yichen Li. Yichen is an incredible scientist whom I’ve always looked up to as a role model. Discussing science, philosophy, and existence in general with Yichen has been one of my favorite parts of my time here. I have been inspired by our conversations over the years and have always gained key insight into my own work from our discussions. I would also like to thank Yichen for being an incredible friend. A lot has happened over the course of my PhD and Yichen has always been someone I am able to count on. Thank you for all the loving memories and insightful discussions. Lastly, I would like to thank my parents and siblings, Alison, Natalie, Andrew, Matthew, and Thomas, for their love and support. I could not have succeeded without all of you. I am so thankful for the visits, many facetime calls, and care packages sent over the years. I am looking forward to spending more time together (and the seasons) back in New England. Graciously, Amanda Stevens Meyer vi TABLE OF CONTENTS Epigraph…………………………………………………………………………..….…………ii Acknowledgments………………………………………………………..…………………...iii List of Tables…….………………………………………………………..…………………...vii List of Figures..….………………………………………………………..…………………..viii Abstract…….………………………………………………………..………………...………...x Chapter 1: Introduction: Overview of biotin-based mammalian in vivo proximity labeling to study cell-type specific proteomes and subcellular compartment mapping………………………………………………………………………………………....1 Chapter 2: Proteomics of protein trafficking by in vivo tissue-specific labeling………...19 Chapter 3: A genetic model system for in vivo proximity labeling of the mammalian secretome………..………………………………………………………………60 Chapter 4: In vivo proximity labeling identifies adipocyte-specific signaling changes in diet induced obesity……………………………………..……………………..125 Chapter 5: Conclusions & Discussion……………………………………………………...172 Chapter 6: Perspective on proximity labeling to map subcellular compartments…….175 References…………………………………………………………………………………….178 vii List of Tables 1.1 Proximity labeling biotin ligases………………………………………………………3 2.1 Oligonucleotides used in this study………………………………………….………58 2.2 BirA*G3-ER teratoma streptavidin enriched (Enrichment score≥1) and non- enriched (Enrichment score=0) proteome lineage markers………………………..59 3.1 Primary antibodies information…………………………………………………….123 3.2 Secondary antibodies information………………………………………………….124 4.1 Primary antibodies information…………………………………………………….170 4.2 Secondary antibodies information………………………………………………….171 viii List of Figures 1.1 Biotin ligase proximity labeling……………………………………………………….……2 1.2 Overview of mouse genetic proximity labeling models……………………………..…...9 2.1 Rosa26 (R26) BirA*G3–ER and R26 BirA*G3–ER;Cre 3A mouse embryonic stem cell (mESCs) characterization…….…………………………………………………….….44 2.2 Characterization of in vitro biotinylation in the BirA*G3–ER mouse embryonic stem cell (mESC) line……..…………………………………………………………….…..45 2.3 Analysis of biotin labeling of proteins in vivo in a mouse embryonic stem cell (mESC)–derived teratoma model…………………………………………………….…....46 2.4 Identification of mouse teratoma-derived proteins in serum…………………….....….48 2.5 Validation of mass spectrometry proteomics-based predictions in tumor and serum samples…………………………………………...…………………………….…....49 S2.1 mESC streptavidin and BirA expression characterization and positive control, Alpl, secretion analysis (Fig. 2.4 supplement)……. ……………………………...…….51 S2.2 Teratoma characterization (Fig. 2.3 supplement)……………………………………….53 S2.3 Teratoma and serum streptavidin and BirA*G3-myc western blots (Fig. 2.3 supplement)……………………………………………………………………………...…54 S2.4 Thresholding and signal peptide analyses of biotinylated serum and teratoma proteins identified using tandem mass-tag (TMT) mass spectrometry (MS) using BirA*G3-ER labeling (Fig. 2.4 supplement)……………………………………………...55 S2.5 Additional and complementary analyses of teratoma and serum datasets (Fig. 2.4-2.5 supplement)………….. ……………………………………………………...57 3.1 Generation and characterization of Sox2-BirA*G3 mice………………………………...95 3.2 Analysis of biotinylated proteins in Sox2-BirA*G3 and control mice………………….97 3.3 Identification of biotinylated proteins in Sox2-BirA*G3 liver, brain, and kidney tissues by mass spectrometry…...…………………………………………………………98 3.4 Analysis of biotinylated proteins secreted to peripheral blood………………………100 3.5 Generation and characterization of Alb-BirA*G3 mice………………………………...102 3.6 Analysis of hepatocyte-secreted serum proteins in Alb-BirA*G3 mice………………103 S3.1 BirA*G3 expression and colocalization with ER marker Calnexin in Sox2-BirA*G3 mice……...………………………………………………………………...105 S3.2 Native GFP and mKate2 expression in kidney and brain sections………………….107 S3.3 Total protein stains used for western blot normalization and quantification……..108 S3.4 Detection of biotin and biotinylated proteins in Sox2-BirA*G3 and control tissues……………………………………………………………………………………...109 S3.5 Partial BirA silencing observed in Sox2-BirA*G3 mice was affirmed by comparison with Sox2-Cre; TdTomato mice………..…………………………………...110 S3.6 Homogenous mKate2 and BirA*G3 expression was detected in Sox2-BirA*G3 and control mouse pups at 10 days post birth and in CRE-induced liver cells from CAGG-BirA*G3 mice…………………..…………………………………………...111 S3.7 No detrimental effects were observed in Sox2-BirA*G3 and control mice…………112 ix S3.8 Streptavidin purification of biotinylated proteins for LC-MS/MS from Sox2-BirA*G3 mice……...………………………………………………………………...113 S3.9 Quantitative LC-MS/MS data analysis for three tissues…………………………….114 S3.10 Identification of enriched proteins in Sox2-BirA*G3 tissues………………………..116 S3.11 Tissue-specific analysis for tissue-specific enriched proteins……………………...117 S3.12 Data analysis for serum mass spectrometry results………………………………...118 S3.13 Quantitative MS data analysis and western blot validation for serum…………...119 S3.14 Characterization of Alb-BirA*G3 and control mice………………………………….120 S3.15 Characterization of protein biotinylation in Alb-BirA*G3 and control mice……...121 S3.16 Analysis of Alb-BirA*G3 and control serum enriched proteins…………………….122 4.1 Adipocyte secretory compartment protein biotinylation in mice…………………….153 4.2 Analysis of adipocyte derived biotinylated proteins in SD or HFD treated mice…..154 4.3 Identification of diet induced biotinylated proteins in subcutaneous and visceral white adipose tissue……………………………………………………………………….156 4.4 Identification of diet induced circulating adipocyte-derived biotinylated proteins..158 4.5 Identification of adipocyte-derived circulating proteins………………………………159 S4.1 Adipocyte-specific BirA*G3 expression in Adipoq-BirA*G3 mice……………………160 S4.2 Adipocyte-derived biotinylated serum protein enrichment…………………………162 S4.3 Quantitative MS data analysis of subcutaneous and visceral adipose tissue………164 S4.4 MS analysis of adipose tissue diet enriched proteins and single-nuclear. RNA-seq validation……...……………………………………………………………….166 S4.5 Adipose tissue MS enriched protein validation……………………………………….167 S4.6 Quantitative MS data analysis of adipocyte-derived circulating biotinylated proteins……………………………………………………………………………..……..168 S4.7 Adipocyte expression of adipocyte-derived circulating serum biotinylated proteins by single-nuclear RNA-seq validation………………………………………169 x Abstract Identifying cell-type specific circulating proteins remains a key challenge. Here, we adapted biotin-based proximity labeling to in vivo mammalian systems. We demonstrated in vivo proximity labeling “proof-of-principle” by identifying teratoma- derived circulating proteins in mice. We then generated a genetic proximity labeling mouse where a promiscuous biotin ligase, BirA*G3, is localized to the endoplasmic reticulum allowing for biotinylation and subsequent identification of secreted proteins. To profile secretion in a cell-type specific manner, we used a floxed BirA*G3 allele when crossed to a Cre-driver excises a GFP-stop cassette resulting in BirA*G3 expression. Extensive characterization of this mouse shows the ability to label and identify secreted proteins from numerous tissues including liver (ALB, AGT), kidney (UMOD), immune cells (CXCL7), muscle (MSTN), fat (ADIPOQ, Cfd), gut (FGF15), and brain (OXT). We then sought to apply this technique to identify diet induced obesity adipocyte secretion changes in mice. High fat diet treatment for 12 weeks resulted in significant weight gain and altered biotinylated protein profiles. Proteomic analysis of these samples identified diet specific protein changes. These studies demonstrate proximity labeling detection of cell-type specific secretomes in homeostasis and in disease. This mouse model provides a new tool for the research community to apply to numerous biological systems. By using this mouse model to investigate adipocyte secretion we have identified new insight in adipocyte-derived circulating factors and provided strong evidence of proximity labeling approaches to identify disease signatures. Together, this body of work has contributed to the advancement and development of in vivo proximity labeling techniques. 1 Chapter 1: Introduction Overview of biotin-based mammalian in vivo proximity labeling to study cell-type specific proteomes and subcellular compartment mapping Introduction to Proximity Labeling Cells are regulated by complex and dynamic machinery. Mechanisms by which this is done have proven difficult to parse out. In particular, protein mapping, identifying subcellular compartment specific proteomes, has long been a challenging subject of study. Eukaryotic cells are divided into numerous compartments, organelles, and phase- dependent condensates. Efforts to dissect these specific components have required harsh methods that need high input, specialized tools, and are more amenable to certain cell types than others. As a result, there have been limitations on the knowledge produced from these studies, particularly by the challenges of cell-type specific morphology and disease complexity. One new way to address this problem is proximity labeling. Proximity labeling focuses on identifying interaction networks and pathways of molecules and mapping of subcellular compartments of interest that have been difficult to decipher using classic biochemical approaches (Fig. 1.1) 1,2 . It does this by localizing a promiscuous ligase, for example, within a specific region of the cell or by attaching it to a specific protein-of-interest, resulting in ligand tagging of region or protein-associated targets through localized enzymatic activity. 2 Figure 1.1: Biotin ligase proximity labeling. Biotin ligases biotinylate proximal proteins in the presence of biotin which are subsequently affinity purified (AP) using avidins or antibody-based approaches. Affinity purified proteins are then identified and quantified by mass spectrometry (MS). To date numerous proximity labeling enzymes have been identified or created and characterized; the majority of which are biotin ligases (Table 1.1) 1 . APEX and BioID were amongst the first generation of proximity labeling enzymes. APEX and BioID utilize an engineered ascorbate peroxidase and E. coli promiscuous biotin ligase BirA*R118G respectively, to add biotin to lysine residues on proteins of interest allowing for efficient subsequent streptavidin-based affinity purification 3,4,5 . APEX allows for fast labeling time (< 1 minute) but can cause toxicity dues to its use of H 2O 2, while BioID is non-toxic, with slow labeling times (> 18 hours) 6 . While both of these methods allowed for exciting and novel research looking at cellular compartment proteomes and protein-protein interaction (PPI), they could be improved. To overcome the limitations of the toxicity of APEX, TurboID, a fast, relatively non-toxic promiscuous biotin ligase was generated 6 . As a result, TurboID has been widely adapted into a multitude of model systems, including bacteria, yeast, plants, worms, flies, and mammals, both in PPI studies and subcellular compartment mapping (Table 1.1) 7 . To date, a number of comprehensive reviews have 3 been written comparing various proximity labeling approaches, enzymes, and capabilities 1,2,8,9 . Thus, this review will focus on in vivo biotin ligase proximity labeling in mice due to the recently published genetic models, the various approaches used, and the results they produce 10,11,12 . This review will also provide a brief overview of TurboID and its predecessor’s since they have been utilized in almost all of the mammalian in vivo systems 6 . Table 1.1: Proximity labeling biotin ligases. Enzyme Category WT Enzyme Species of Origin Mutantions Size (kDa) Labeling Time Self-Biotinylating Model organisms AirID Biotin ligase Biotin ligase Ancestral reconstruction R118S 35 3 h No mammalian cell culture BioID Biotin ligase BirA Escheria coli R118G 27 15-24 h Yes mammalian cell culture, yeast, flies, plants, mice BioID2 Biotin ligase Biotin ligase Aquifex aeolicus R40G 27 15-24 h Yes mammalian cell culture, mice BASU Biotin ligase Biotin ligase Bacillus subtillis ∆aa1-65, R124G, E323S, G325R 29 18 h Yes mammalian cell culture G3 Biotin ligase BirA Escheria coli Q65P, S150G, L151P, I87V, R118S, E140K, Q141R, V160A, T192A, M209V, I305V 35 10 min Yes mammalian cell culture, flies, mice ∆G3 Biotin ligase BirA Escheria coli *all G3 + ∆aa1-63 35 10 min Yes mammalian cell culture microID Biotin ligase Biotin ligase Aquifex aeolicus BioID2 aa2-171 (NBioID2) 19.7 10 min Yes mammalian cell culture microID2 Biotin ligase Biotin ligase Aquifex aeolicus R40G, K36R, L41S, K44R, L46F, K102R + ∆aa63 19 10 min No mammalian cell culture lbMicroID2 Biotin ligase Biotin ligase Aquifex aeolicus R40G, K36R, L41S, K44R, L46F, K102R, I120A, N124A + ∆aa63 19 10 min* No mammalian cell culture miniTurbo Biotin ligase BirA Escheria coli *all ∆G3 + K191I 28 10 min Yes mammalian cell culture, bacteria, yeast, flies, worms, plants Split-BioID* Biotin ligase BirA Escheria coli R118G, E256/G257 35 15-24 h Yes mammalian cell culture Split-BioID Biotin ligase BirA Escheria coli R118G, E140/Q141 35 15-24 h Yes mammalian cell culture Contact-ID Biotin ligase BirA Escheria coli R118G, G78/G79 35 15-24 h Yes mammalian cell culture Split-TurboID Biotin ligase BirA Escheria coli *all TurboID, L73/G74 35 10 min Yes mammalian cell culture (HEK293T) TurboID Biotin ligase BirA Escheria coli *all G3 + K194I, M241T, S263P 35 10 min Yes mammalian cell culture, bacteria, yeast, flies, worms, plants, fish, mice ultraID Biotin ligase Biotin ligase Aquifex aeolicus BioID2 (R40G)+L41P aa2-171 (NBioID2) 19.7 10 min Yes mammalian cell culture 4 Biotin Ligase Proximity Labeling with TurboID Overview of TurboID TurboID was derived from wild type E. coli biotin ligase (BirA) using directed evolution. Mutating the R118 residue of wild type E. coli biotin ligase (BirA) is known to increase promiscuity 5 . Therefore, Brannon et. al, created 7 different R118 mutants and benchmarked them against the BioID BirA*R118G mutant. A R118S mutation was shown to have a two-fold increase in labeling activity compared to the R118G and was selected for further mutagenesis 6 . Error-prone PCR was coupled with a previously established yeast surface expression system to screen for the most promiscuous mutants. This system allows for the amount of surface BirA* to be compared to the amount of biotinylation. Enzymes with low BirA* and high biotinylation were selected for further mutagenesis. This resulted in generations of mutant BirA* (BirA*G1, BirA*G2, BirA*G3) with the final mutant being TurboID 6 . These mutants were then characterized in HEK293T cells compared to existing proximity labeling enzymes (Table 1.1). TurboID showed similar labeling activity after 10 minutes compared to 18 hours of labeling by biotin ligases: BioID, BioID2, and BASU 6 . Limitations of TurboID TurboID is now the most widely used proximity labeling enzyme due to its fast and high labeling activity. The generating lab has published additional variants of TurboID for cellular compartment localization, reduced size (miniTurbo), and direct PPI (splitTurbo). TurboID and its relatives have been used in a number of systems from in vitro systems to in vivo worm, fly, mammals, yeast, bacteria, and plants 1,7 (Table 1.1). Despite its popularity, TurboID has been shown to have limitations for PPI studies 5 due to some off-target labeling. To overcome this, efforts have been made to engineer more stringent biotin ligases. One such ligase, AirID (ancestral BirA), was shown to have similar biotinylation capabilities to TurboID with reduced off-target labeling and does not self-biotinylate allowing for more stringent identification of PPI networks 13 . Additionally, AirID has lower potential long-term toxicity, but low catalytic activity may limit in vivo applications 1,13 . Similar enzymes with low catalytic activity in vivo have been used to successfully identify PPI networks or map subcellular compartments, however, their datasets, particularly for protein mapping, are much more limited compared to TurboID and its relatives 1,10,11,12,13,23 . Taken together, enzymes such as AirID or BioID(2) may provide more stringent data for PPI networks, reduce any potential long-term in vivo toxicity, but may produce limited results for subcellular compartment protein mapping. Although TurboID improved activity of proximity labeling, there are still additional limitations in PPI studies due to the size of BioID (321aa), TurboID, and APEX on fusion proteins that can lead to a null protein of interest (POI), loss of function, and/or detrimental effects particularly in vivo. To overcome this, smaller proximity labeling enzymes, miniTurbo and BioID2 were generated 6,14,15,16 . Although miniTurbo has 1.5-2 fold less activity than TurboID, it uses less endogenous biotin allowing for potentially cleaner temporal studies 6 . After failed attempts to reduce the size of BioID (E. coli BirA) by N-terminal deletion of the DNA binding domain resulted in loss of enzymatic activity, BioID2 was created by identifying the smallest known biotin ligase, a ligase from Aquifex aeolicus (233aa; Table 1.1) 16 . Further, recent work has generated next-generation small promiscuous biotin ligases, ultraID, microID (BioID aa2-171), and microID2, all of which were derived from C-terminal truncation of BioID2 with an R40G mutation and additional mutations in ultraID (Table 1.1) 14,15 . They have about a 7 kDa size decrease compared to miniTurbo making them the smallest (~19.7 kDa) engineered biotin ligases 6 thus far. These new small biotin ligases exhibit increased labeling efficiency compared to BioID2 and only slightly decreased labeling efficiency compared to TurboID and decreased background biotinylation providing a next generation set of highly efficient, small biotin ligases 14,15 . Proximity Labeling to Decipher Protein-Protein interaction Networks Proximity labeling has been successful in identifying protein-protein interaction networks. These studies have not only identified highly similar PPI networks to Co-IP studies but have also identified more transient interacting partners for proteins of interest 17-21 . However, the need to generate a fusion-protein (ligase-protein of interest) has limited the approach in various systems and for proteins that lose function when fused 16 . An interesting new approach to this problem, is using antibody-localized ligases. Here, a biotin ligase is fused to an antibody against the protein of interest or against a tagged version of the protein of interest 222 . This allows antibody-based localization of TurboID to the POI without the need to generate a TurboID-POI fusion protein. Additionally, this has been shown in vivo highlighting the power of this approach for genetic systems against tagged POIs (e.g., HA, Flag) for rapid integration with existing genetic models 222 . Lastly, to increase stringency and contact-dependent PPI network mapping, split- proximity labeling (e.g., Split-BioID, Split-TurboID, Split-APEX) methods were generated, where each half of the biotin ligase is fused to one POI 23-27 . When both POI interact, the biotin ligase forms initiating labeling of proximal proteins. Proximity Labeling for Subcellular Compartment Mapping Although initially broadly used to study protein-protein interaction networks, proximity labeling has become an exciting tool to map proteomes at the subcellular levels. Numerous groups have applied proximity labeling to map organelles and subcellular structures such as molecular condensates in flies, worms, yeast, plants, and mammals 28- 7 49 . There has been particular focus in mammals to map cell-type specific secretomes and proteomes 48-51 . In zebrafish, worms, and flies cell-type specific mapping has included mitochondria, membrane proteins, P granules, secretomes, and general proteomes. These methods have taken advantage of signal sequences (KDEL, MTS, CAAX, NLS, NES) and fusion proteins to localize TurboID to the compartment of interest 10,11,12,28,29,43,48,49,51 . There has been evidence that fusions proteins may provide more specific localization, particularly for ER localized proteins 48 . However, fusion proteins require additional testing to ensure normal function is not disrupted. Using smaller biotin ligases for fusion proteins (miniTurbo, BioID2) reduces the chance for tertiary structure issues increasing the success of fusion protein based approaches 6,16 . In vivo Proximity Labeling in Mouse Non-Genetic Approaches to in vivo Proximity Labeling Proximity labeling has been quickly adapted in model systems such as fly and worm, and later adapted in zebrafish. These in vivo systems have allowed for identification of new protein-protein interaction networks in homeostasis and regeneration and subcellular compartment mapping and have provided novel biological insight 6,22,35,42,43 . Generation of genetic mammalian models is technically challenging and time consuming and not required for easily transduced tissues. Thus, multiple groups used transduction methods and grafts to show “proof-of-principle” and identify new biological insights in mice using proximity labeling. Multiple groups used viral AAV transduction of the liver with ER-localized TurboID to profile liver secreted proteins both known and unknown 48,49,51 . Further, a mouse teratoma model with and ER-localized BirA*G3 showed the feasibility to identify grafted cell-secreted proteins in blood 50 . 8 Although many of these groups successfully identified well-known proteins in their system, few have shown strong novel biological insight with the exception of a recent nuclear proximity labeling paper that harnessed proximity labeling and computational biology to identify microproteins and unannotated proteins 51 . This method has been termed microID 51 , not to be confused with the relatively new microID enzymes 14,15 . These studies highlight the potential of proximity labeling to produce robust and exciting data but lack the data supporting feasibility to identify striking novel biology, particularly in disease conditions. Genetic Mouse Models of Proximity Labeling Proximity labeling has been used in numerous applications, however, in vivo studies have been limited by the availability of transgenic model systems and limitations of transduction methods. This is a particular limitation in mammalian systems primarily due to the slow nature of generating rodent models and the expertise required to do so. To overcome this, multiple groups have focused on generating transgenic proximity labeling mice compatible with the Cre/loxP system to allow for broad use by future groups enabling studies of cell-types of interest to them. In the past two years, three genetic proximity labeling biotin ligase mouse models (two secretome (ER localized biotin ligase) and one cytoplasmic) have been published, including the mouse model from our group (Figure 1.2) 10,11,12 . Each of these mouse lines involved using a promiscuous biotin ligase, either localized to a subcellular compartment or not. 9 Figure 1.2: Overview of mouse genetic proximity labeling models. Schematic depicts mouse alleles, cell-types profiled, number of enriched proteins identified compared to controls or other cell-types (n’s), and cell-type specific proteins identified using this method. These mouse models focus on “proof-of-principal” work for in vivo mammalian proximity labeling. Each model successfully recombines and expresses a biotin ligase in a cell-type specific manner by using a Cre-driver. The BioID2/secretome mouse used a (VE)-cadherin and muscle creatine kinase (MCK) Cre’s to profile endothelial and muscle secretion respectively 10 . Here endothelial and muscle specific secreted proteins were identified, including well known cell-type specific secreted proteins. Additionally, in an exercise study the BioID2 mouse was able to capture exercise induced muscle secretion changes in serum supporting the use of this strategy in physiological studies 10 . However, the power of this mouse model’s ability to be used in temporal studies is limited by the slow labeling time of BioID2 and the low catalytic activity in vivo 1 . The BirA*G3/secretome mouse used a ubiquitous Cre and Albumin-Cre to profile labeling in 10 major organs and then specifically in liver 11 . Here, liver, brain, kidney, and serum were used to show that ER-localized proximity labeling can capture and identify numerous cell-type specific secreted proteins. This is further confirmed in the liver-specific mouse, where numerous liver circulating proteins such as AGT, APOE, ALB, and IGFBP2 were identified 11 . Both of these secretome mouse models show strong evidence for proximity labeling to provide powerful data on cell-type specific secretomes. Lastly, the TurboID mouse, which has TurboID expressed throughout the cell with the exception of the nucleus (NES) profiled neuronal and astrocyte proteomes from multiple brain through cell-restricted activity of Camk2a-Cre ERT2 and Aldh1l1-Cre ERT2 12 . The study identified brain-region and cell-type specific proteome signatures 12 . This model is particularly beneficial for cell-types that are difficult to isolate and are prone to state changes during isolations and processing. Toxicity of in vivo Proximity labeling To date, there has been little evidence of proximity labeling induced toxicity in either in vitro or in vivo systems. The most prominent evidence was from ubiquitous- TurboID flies and TurboID-NES and miniTurboID-NES HEK293T proximity labeling 6 . Here, it was shown that when TurboID was expressed ubiquitously and constitutively in flies there was a decrease in fly viability and size in the absence of biotin. This is rescued by biotin supplementation during development, suggesting that excessive biotin sequestration without additional biotin supplementation results in decreased growth and toxicity 6 . In HEK293T constitutively expressing TurboID-NES and miniTurboID-NES cells (either stable lines or lipofected), there was a decrease in cell proliferation compared to WT HEK293T cells and between biotin treated and untreated counterparts 6 . However, extensive characterization in mouse models has suggested that this toxicity may not be grossly relevant. In all three mouse strains enabling proximity labeling (BioID2, BirA*G3, 11 and TurboID) no pathological affects have been reported in the recombined mice 10,11,12 . Notably, the TurboID mouse utilized TurboID expression constitutively in mouse neurons and astrocytes for two-four weeks with two weeks of constant biotin supplementation and did not see any observable adverse effects 12 . Further, the BirA*G3 mouse model extensively investigated toxicity through histology, bulk RNA sequencing, and ER stress and unfolded protein response (UPR) resulting in no differences between BirA*G3 expressing and non-expressing mice 11 . Additionally, viability and fertility of these mice were normal 11 . Lastly, though unpublished, our group was unable to generate cytoplasmic BirA*G3 mice using our Sox2-Cre driver but have been successful using tissue-specific Cre-drivers. Sox2 is a transcription factor expressed in the epiblast, when used as a Cre- driver it results in floxed allele recombination early on in development in all lineages. Coupled with our published data showing CAGG-Cre or Alb-Cre activation of BirA*G3 rescued hepatocyte BirA*G3 mosaicism suggests that developmental timing of proximity labeling activation and tissue-restriction are key determinants of toxicity and allele regulation. Although, there has been variable evidence on biotin ligase proximity labeling toxicity in vivo and almost none in mice, it is still an important aspect of the system. Disruption in biotin homeostasis may be one issue, as seen in flies, however, it is still unclear how widespread levels of biotinylation affect normal protein function, localization, and signaling. Since biotinylation mainly occurs on lysine residues in biotin ligase-based proximity labeling, there is the potential for post-translational modification disruptions such as ubiquitinoylation, which also occurs on lysine residues 1 . Further studies are required to better understand how biotinylation of proteins affects their normal state and function. 12 Biotin supplementation Numerous groups have utilized in vivo biotin based proximity labeling employing various strategies to supplement animals with biotin to induce labeling. Here, we discuss the various biotin supplementation strategies utilized in mice and their efficacy. The two primary methods to increase in vivo biotin levels are oral supplementation (chow/water) and injection. Most groups to date have used longer-term biotin supplementation (>5 days, up to two weeks) via chow, water, and injection 10,11,12,48,49,50,51 . These methods have resulted in strong biotinylation of proteins allowing for greater enrichment of proteins. Biotin chow and water offer extremely easy and efficient methods for supplementing mice. When used for multiple consecutive days this leads to a massive increase in systemic biotin levels. This results in biotinylation of a large swath of proteins overtime capturing more global changes. This strategy, however, may not be appropriate for physiologically responsive proteins which undergo significant abundance changes in short timeframes (e.g., minutes, hours), such as insulin or leptin’s sharp increase after eating. For physiologically responsive and temporally regulated proteins, biotin injection may offer improved labeling and enrichment. Our group has shown that after just 1 hour of biotin injection (coupled with ad libitum biotin water at time of injection) biotinylated proteins can be detected in the serum from constitutively active BirA*G3 mice 11 . Notably, BirA*G3 has a higher affinity for biotin than TurboID allowing it to use trace amounts of biotin present, thus base labeling (no biotin) must be determined for each model 6 . TurboID may be an improved choice for temporally and physiologically focused studies. Thus, it is critical to determine the optimal biotin supplementation amount and timing. 13 Sample Enrichment, Processing, and Analysis Biotinylated Protein Enrichment Strategies A key aspect of proximity labeling is the actual protein enrichment. The enrichment methods selected whether at the protein or peptide level can heavily influence the outcome and type of data generated. Biotinylated proteins can be enriched either as proteins in their native state or after being digested into peptides. Peptide digestion followed by biotinylated peptide enrichment has been shown to lead to more stringent data 48,52 . In this method any non-biotinylated peptides, even if from a biotinylated protein are lost. This results in very high confidence of proteins identified but comes with a large loss of potential data. Protein enrichment, on the other hand, allows for peptide digestion post-enrichment. This results in all peptides from an enriched biotinylated protein being sequence. Comparison of liver secretome datasets from multiple groups supports this difference in data, with the protein enriched datasets capturing many more liver secreted proteins than the peptide enriched ones 11,48,49 . However, peptide enrichment has the additional benefit of being able to stringently identify biotinylated residues on peptides providing more direct evidence that proximity labeling has occured 52 . Additionally, the peptide enrichment reduces co-enrichment of proteins that may be bound to biotinylated proteins. Further, the affinity-binding method can lead to differences in proteomic outcomes. Biotinylated protein enrichment is most commonly done using avidin family protein-conjugated beads. Avidin-conjugated beads come in a variety of resins and avidins. Generally, the agarose resin beads have increased binding capacity compared to magnetic beads. This can be an important in optimizing the amount of beads to protein and for balancing the amount of avidin protein in the final sample. Within in the avidin 14 protein family, streptavidin is often used due to its extremely high affinity for biotin (~K d = 10 -14 M) 52 . Additionally, streptavidin’s tetrameric form allows for increased binding capacity. Neutravidin is another tetrameric avidin that has a similar biotin affinity to streptavidin but lacks any glycosylation. Neutravidin has a more neutral isoelectric point and less non-specific binding properties. Both streptavidin and neutravidin require harsh elution methods (8M guanidine-HCl or boiling) that strips the resins of the avidins adding a large amount of avidin protein to the enriched proteins. Monomeric avidin has a reduced affinity for biotin (~K d = 10 -8 M) allowing for elution by gentler methods and avoids stripping the beads of the avidins 54,55,56 . Lastly, antibody-based enrichment methods have the benefit of gentler elution methods and increased binding stringency. Elution methods further come into play when preparing samples for proteomics. On-bead digestions are a popular method for removing and generating peptides from enriched proteins. However, this method will also digest the avidins on the beads. Too much avidin peptides can make lead to difficulty in sequencing the peptides of interest. To avoid this, methods have been developed to chemically modify streptavidin rendering it resistant to trypsin digestion 57 . Others have used antibody-based peptide enrichment or monomeric avidin followed by gel digestion to generate avidin-free MS samples. Overall, the enrichment method is critically important to the system being used and question being asked. Although, standard approaches are needed, proximity labeling experiments do require a degree of individual tailoring. In many cases, method comparison and testing will be beneficial for each experimental system and biological question. Analysis pipelines and discovering new biology Unlike the field of transcriptomics, affinity purification proteomics suffers from a lack of well-defined, readily accessible pipelines. This is primarily due to the very specific 15 nature and individuality of proteomic based experiments. However, since proximity labeling methods are growing, the field would greatly benefit from analytical pipelines to help discover novel biology from these datasets. Certain tools exist and the standard practices of proteomics data analysis are applied, however, many of these tools have limitations such as operating systems dependencies (e.g., Windows applications only), lack of readily available analysis tutorials, and niche software that is not always maintained. Further, statistical approaches have varied for determining enriched proteins. Particularly for subcellular compartment mapping, the field would greatly benefit from the standardization of analyses and comparisons used to harness the full power of proximity labeling based approaches. Proximity labeling can interrogate cell-type specific proteomes, map protein- protein interaction networks, and distinguish subcellular compartments. The enzymes used and methods employed for proximity labeling based work require optimization and tailoring to each system but have the capability to provide interesting new biological insights that complement existing techniques. The generation of mouse genetic models for proximity labeling will enable broad use of proximity labeling to study homeostatic and disease conditions in mammals in vivo 10,11,12 . By utilizing the Cre-loxP systems these mice can be rapidly and broadly implemented by researchers across the globe. This will only benefit the field of proximity labeling by generating better approaches, analytical pipelines, and standardizing in vivo approaches. Proximity Labeling to Study Adipocyte Signaling in Obesity Proximity labeling has been proven to be highly capable of mapping subcellular compartment proteomes and secretomes in a variety of tissues. However, it has yet been 16 applied to disease areas in mammals. Obesity is a global disease affecting over 1 billion people in the world 58,59,60 . Obesity affects individuals at any or all life stages; from childhood to adulthood. It is a comorbidity and/or risk factor for most major common diseases and is marked by key endocrine changes 60 . Obesity drives adipocyte dysregulation and dysfunction leading to a multitude of cellular and endocrine level changes. Adipokines are circulating proteins released by adipocytes that have a broad range of functions and altered levels in obesity 60 . However, approaches such as blood proteomics are limited by lack of ability to identify a protein’s tissue of origin and predominantly identify highly abundant proteins such as albumin. Further, proteins secreted to blood by multiple tissues (e.g., FGF21, HP, or APOE) have required conditional knockout models or overexpression systems to interrogate an individual tissue’s role in these proteins circulating levels 59, 65, 66 . Adipocytes well characterized secreted proteins makes them an excellent cell-type to show both “proof-of-principle” and discover new biological insights via in vivo proximity labeling. Further, adipose tissue is a key endocrine tissue required for homeostasis and energy regulation. In obesity, white adipocytes, the lipid energy storing cells, undergo hyperplasia and hypertrophy to accommodate the additional energy storage requirements. However, there is a point at which this compensation breaks down and adipose tissue becomes unable to bear the excess energy storage requirements. Adipocyte death increases in conjunction with hypoxia, which drives infiltration of proinflammatory macrophages. This macrophage proinflammatory shift is thought to contribute to systemic low-grade inflammation 61,62,64 . Adipose tissue regulates physiological conditions through signaling hormones (adipokines), which target specific organs, such as the brain, skeletal muscle, liver, pancreas, and kidney to maintain homeostasis in response to environmental conditions 58,60,64,65,66,67 . Adipokine serum levels 17 are well characterized and change in response to environmental cues such as positive (e.g., high fat diet, obesity) and negative energy balance (e.g., fasting), and cold exposure 60,61,68,69,70,71 . In addition to environmental changes in adipokines, there are well documented and well correlated major organ responses to adipose tissue dynamics 58,60,61,68,72 . Despite extensive research on adipose tissue and obesity, the mechanisms by which adipose tissue regulates other organs remains unclear. Here, we apply a genetic proximity labeling mouse model to identify adaptive and maladaptive adipocyte secretion changes in diet induced obesity. Goals of my Thesis In 2018, the field of proximity labeling had a prominent advancement: TurboID. Generated by Alice Ting, Tess Brannon, and Ting’s lab, TurboID was characterized in fly and mammalian cells with the help of Justin Bosch, Norbert Perrimon, and Perrimon’s lab 6 . Due to its success, the interest in proximity labeling genetic model systems grew. Norbert Perrimon, a colleague and friend of Andy McMahon, invited him to join their proximity labeling group to generate a mouse model system. From this my thesis began. When I joined Andy’s lab, I initiated proximity labeling work and characterization in mouse systems in the lab with the help of Jill McMahon and Jinjin Guo. What started out as a rotation project to show that BirA*G3 (biotin ligase proximity labeling) mouse embryonic stem cells indeed labeled secreted proteins quickly evolved into an extensive characterization of using proximity labeling to map secreted proteins in mice. The goals of my thesis were to characterize and publish our genetic secretome mouse and subsequently use it to study adipocyte secretory changes in diet induced obesity. I am delighted to share that with the collaboration with Rui Yang, we were able to extensively 18 characterize our mouse model and through work of my own, shown the ability to use proximity labeling to identify obesity specific adipocyte secretory changes. Through all of these experiences I have grown into a capable, independent scientist. I have been provided with high quality training, expectations, and support by my mentor, Andy. This has well-prepared me for a career in scientific research and helped me become the scientist I am today. 19 Chapter 2 Proteomics of protein trafficking by in vivo tissue-specific labeling My work on the mammalian BirA*G3 studies in this chapter has been published in Nature Communications (PMID 33888706) together with independent studies on Drosophila from our collaborators in the Perrimon laboratory. The mouse component of the published study is presented in this chapter. This work was led by Ilia Droujinine (Figures 4 and Supplemental Figures 1Y-DD, 3C-D, 4, 5A-I, 5M-N) and myself (Figures 1B-K, 5, 3A, 3P-Q, 5; Supplemental Figures 1A-X, 1EE, 3A-B, 5O) with data collection and manuscript preparation contribution by Dan Wang, David Rocco, Jill A. McMahon (1A, 3B-C; Supplemental Figures 2), Rui Yang, Jinjin Guo, Luye Mu, Dominque K. Carrey, Tanya Svinkina, Rebecca Zeng, Tess Branon, Areya Tabatabi, Justin A. Bosch, John M. Asara, and Alice Y. Ting, and Tracy Tran (Figure 3D-O). Steven A. Carr and Namrata Udeshi supervised proteomic experiments and analysis. Yanhui Hu conducted proteomic analysis. Andrew P. McMahon, Ilia Droujinine, and Norbert Perrimon conceptualized, supervised, and advised on experimental design, data analysis, and manuscript preparation and review. INTRODUCTION Local tissue homeostasis is becoming increasingly well-understood. However, the physiological importance and presence of secreted blood and inter-organ communication factors is only beginning to be documented from experiments in Drosophila and vertebrates. Secreted factors acting directly or indirectly between organs encode key 20 regulators of systemic homeostasis 73 . These factors traffic, or translocate, intracellularly from their production sites within cells 74 to distal organs through the blood circulation 1 . For example, adipokines such as leptin and adiponectin encode adipose tissue-derived systemic metabolic regulators that act on the brain and other organs 73 . In addition, myokines such as interleukin-6 are secreted by muscles and control metabolism in adipose tissue 73 . Despite their importance, the identification of inter-organ communication factors is technically challenging, and a number of published results were later determined to be irreproducible or controversial 73,75-77 . Also, origins and/or destinations of known factors including glucagon-like peptide 1 (GLP-1), ghrelin, leptin, cholecystokinin (CCK), and growth differentiation factor 11 (GDF-11) remain to be clarified 73,77 . Moreover, because large-scale screening methods are lacking, the long- standing biological question of how many proteins are transferred between any two given organs has yet to be addressed 73 . While a number of factors have been recognized in mammals, many physiologically important factors likely remain to be identified 73 . In addition, defining the secretome of a given cell type directly in the in vivo setting is an important unmet need. Liquid chromatography tandem mass spectrometry (LC-MS/MS) when used in a targeted manner is a powerful approach to identify secreted factors in blood 78 . While LC- MS/MS is a sensitive method, unprocessed blood samples are exceedingly complex with large dynamic ranges of protein concentrations – the majority of the protein mass consists of a small number of protein species which dominate the MS signal, making identification of lower-abundance proteins challenging. Furthermore, LC-MS/MS of blood does not identify the origins and destinations of secreted proteins. Also, MS proteomics of cell culture supernatants is not physiological (because it is performed in vitro and, frequently, in serum-free media 79 ) and cannot identify destinations of factors. 21 In this work, to overcome these limitations, we develop a method whereby all secreted proteins from a specific tissue are labeled by biotin in vivo using the engineered promiscuous biotin ligase BirA*G3 (a relative of TurboID 6 ), then collected by affinity- enrichment from distal organs and identified by quantitative LC-MS/MS. Using this approach, we simplify the proteome under study to the most relevant protein candidates involved in inter-organ trafficking and determine their origins and destinations. To examine the potential of this approach in mammalian systems, a conditional BirA*G3 allele was generated through targeting of mouse embryo stem cells (ESCs), and biotinylation was analyzed in ESC-derived teratomas and host serum samples. Quantitative tandem mass tag mass spectrometry identifies biotin-dependent labelling of secreted proteins shared between the tumor and serum samples. Among 291 streptavidin-enriched blood serum proteins from BirA*G3 tumors are several low- abundance proteins with hormonal properties. Our findings indicate that the communication network of secreted proteins is extensive, and we provide a resource for candidate inter-organ communication factors. The BirA*G3 approach has broad potential across different model systems to identify cell secretomes and mediators of inter-organ communication in healthy or diseased states. METHODS Targeted ES Cells Generation of R26 BirA*G3 Mouse embryonic stem cells (ESCs) The ES cell line B6(Cg)-Tyr<c-2J>/J (https://www.jax.org/strain/000058) was used to target the Rosa26 locus with a Cre-inducible mouse codon optimized BirA*G3-ER vector including the elements: CAGGS-GFP/BirA*G3-ER-myc-IRES mKate2 (R26 BirA*G3): 22 Clones A11, A2, B1, C2, D8. The sequence of the inserted elements is available upon request from the corresponding authors. Generation of Cre activated (R26 BirA*G3-ER;Cre) ES cells R26 BirA*G3-ER Clone B1 ES cells were expanded on mitotically inactivated (Mitomycin C Sigma M0503) DR4 derived (Jackson lab Dnmt1 tm3Jae Hprt b-m3 Tg(pPWL512hyg)1Ems/J) mouse embryo fibroblasts (MEF) in 4500mg/l Dulbecco’s Modified Eagle’s Medium (ThermoFisher 11965-118) supplemented with 15% ES cell qualified Fetal Bovine Serum, 0.1 mM β-Mercaptoethanol, 2 mM L-Glutamine (ThermoFisher 25030-081), 0.1 mM MEM Non Essential Amino Acids (ThermoFisher 11140-050), 1 mM Sodium Pyruvate (Sigma S8636), 50 units of Penicillin, 50 µg Streptomycin (ThermoFisher 15070-063) and 2×10 4 Leukemia Inhibitory Factor (Sigma ESG1107). Upon confluence a 6 cm dish was dispersed with trypsin and 4.7×10 6 ESC were resuspended in cold PBS (Mg/Ca free) mixed with 40µg pCaggs-Cre plasmid and electroporated (Biorad Gene Pulser: 240 V, 500 uFD); 2.9 X 10 6 ES cells were plated to a 10 cm dish of MEF feeder cells. Growing colonies were examined for the loss of GFP fluorescence and the activation of mKate2. Colonies that were RFP+/GFP- were picked, dispersed and serially diluted to generate single cell plating in 96 well MEF feeder plates. Candidate clones were expanded and monitored for RFP+/GFP- fluorescence, resulting in four R26 BirA*G3-ER;Cre clonal lines (3A, 4A, 6A, 7A). There was no evidence of integration of the pCaggs-Cre plasmid into the DNA of the mKate2+ clones as tested by PCR. (Note: pCaggs-Cre was a gift from Connie Cepko (Addgene plasmid 13775; http://n2t.net/addgene_13775). Teratoma Generation The procedure was adapted from Solter 99,100 . All surgeries and animal work was 23 carried out according to federal and institutional guidelines, animal protocols covering the work in APM’s laboratory were approved by the University of Southern California’s IACUC committee. The mouse vivaria are on a 14-10 hour light-dark cycle. The lights turn on at 5 am and turn off at 7 pm. The ambient temperature is kept between 70-75°F and the ambient humidity is 30-70%. R26 BirA*G3-ER and R26 BirA*G3-ER;Cre 3A ES cells were grown to confluence on mitotically inactivated MEFs in 6 cm tissue culture dishes. ESCs were trypsinized, split to 1-2×10 6 cells/tube, pelleted at 300×g, and left on ice while the recipient mice were prepared for surgery. Prior to injection, the cell pellets were placed in 40 µL growth media or 50% Matrigel (Corning BD354277) maintaining clumps. No difference in tumor formation was observed between Cre+ and Cre- R26 BirA*G3-ER clones. Week 8–12 C57Bl/6N male mice (Charles River) were anesthetized with Ketamine/Xylazine. The surgical site, on the dorsal flank was shaved and wiped with Proviodine and alcohol. An 8-10 mm incision was made, the fascia was incised and the left kidney was externalized. The kidney capsule was kept moistened with sterile saline during the procedure. A small incision was made in the outer membrane of the renal capsule at the caudal end, using a sharp 24 gage needle and the sub capsular space was flushed with 0.5-1 mL of sterile saline using a sterile, blunted 30 gauge needle (B30-50, Strategic Applications, Inc.) attached to a 1 ml syringe. A 20 gage or 24 gage indwelling needle (SURFLO® PTFE I.V. Catheter needle, TESR-OX2025CA, VWR) attached to a 1 mL syringe was flushed with ES cell growth media (-LIF) and 1-2×10 6 cells of the pelleted ES cells were drawn into the tip of the needle and injected into the sub capsular space. The capsule incision was briefly cauterized, and the kidney was replaced into the retroperitoneum. Subsequently, the muscle layer was sutured (Ethicon J494G Coated Vicryl Suture, e.sutures.com), and the skin was closed with wound clips. 24 Teratoma Treatment and Collection Mice receiving ES cell injections were maintained on regular chow and water for 3 weeks followed by a 7-day period of biotin supplementation: 2000 ppm in the chow (LabDiet, 5WLP) and 5 mM in the water (Sigma, B4639). Mice were euthanized 4 weeks post ESC injection, urine and blood were collected, and the mice were perfused with cold Phosphate Buffered Saline (PBS), prior to collection of tissue samples. The blood was allowed to clot at room temperature for 30 minutes, then samples were centrifuged at 2000×g for 10 minutes at 4°C, the serum collected and re-spun for an additional 10 minutes, then samples were aliquoted, flash frozen in liquid nitrogen and stored prior to use at -80°C. Tissue samples for histology or immunohistochemistry were fixed for 2 hours in 4% paraformaldehyde in PBS at 4°C, washed three times in PBS for 15 min. and then left in 30% sucrose overnight at 4°C. The samples were transferred into OCT (Tissue-Tek* O.C.T. Compound, VWR, 25608-930), frozen in a bath of ethanol and dry ice then stored at -80°C. Sections were cut by cryostat at a thickness of 10-12 µm onto superfrost microscope slides (VWR, 48311-703) and stored at -80°C prior to use. In vitro Assays R26 BirA*G3-ER and R26 BirA*G3-ER;Cre 3A ES cells were grown in 6 cm tissue culture plates on mitotically inactivated MEF to confluence. The cells were trypsinized, split to 0.5×10 6 cells/well and allowed to grow for 2 days before experimental treatment. Three days post-split, cells were treated with 50 M biotin for 24 hours for general characterization. For the biotin dose study, R26 BirA*G3-ER;Cre 3A ES cells were on feeder free 10 cm tissue culture dishes to spontaneously differentiate to fibroblast-like cells for multiple passages. Cells were then plated onto 6 cm tissue culture at 0.5×10 6 25 cells/well. Two days post plating, cells were treated with biotin in varying concentration (diluted in DMEM from a 50mM stock solution in dH 2 0, pH 7.4): 0, 5, 10, 12.5, 25, 50, 100, or 250 M biotin. For the biotin time exposure study, cells were plated as above and treated with 50 M biotin for 1, 5, 10, 15, 30, 60, 120, 240, 480, 720 minutes. Protein Lysate Preparation Protein lysates were prepared using established protocols, with modifications 28,112 . The lysis buffer was RIPA buffer (made in the lab, 50 mM Tris, pH=8.0, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X-100; or commercial (Pierce)) supplemented with 1 mM benzamidine hydrochloride (VWR), 4 µM pepstatin (VWR), 100 µM PMSF (Sigma-Aldrich), and one cOmplete ULTRA Mini EDTA-free protease inhibitor tablet (Roche). Tissues in lysis buffer were combined with zirconium oxide beads (NextAdvance) and homogenized several times for 4 minutes on setting 9 using the Bullet Blender (NextAdvance), with additions of extra lysis buffer where necessary, and brief centrifugations in between. Additional lysis buffer was added where necessary, and samples were left on ice for 30 min. Next, samples were centrifuged for 15 min at 16,000×g at 4°C and supernatants were transferred to low protein binding tubes (Eppendorf). Total protein concentrations were measured using the BCA protein assay kit (Pierce), according to manufacturer’s instructions. Protein samples were diluted to equal concentrations in one experiment using lysis buffer. Protein lysates were flash frozen and stored at -80°C. Mouse embryonic stem cells (mESC) or mESC derived fibroblasts were collected in 1mL PBS via scraping and then pelleted at 300×g for 5 minutes. Pellets were resuspended in 40 L lysis buffer (M-PER (ThermoFisher, 78501) with 1X protease inhibitor cocktail (CellSignaling, 5871S)). Cells were lysed at room temperature for 10 26 minutes and tapped to shake at the 5-minute mark. Teratomas were homogenized in 500 L RIPA complete lysis buffer (RIPA buffer (ThermoFisher, 89901) with 1× cOmplete mini EDTA-free protease inhibitor cocktail (Sigma, 11836170001), 1mM benzamidine hydrochloride (VWR, TCB0013-100G), 4 M pepstatin (Sigma, EI10), 100 M PMSF (Sigma, 11359061001)) and bead homogenized using stainless steel beads (NextAdvance, SSB14B- RNA) for 5 minutes at setting 10, Bullet Blender Storm (NextAdvance, BT24M). Both cell and tissue lysed samples were centrifuged at 14,000×g for 15 minutes at 4°C to remove cellular debris and the supernatant was collected. Protein lysates were quantified by Pierce BCA (ThermoFisher, 23227) microplate assay or QuBit 2.0 Fluorometer Protein assay (ThermoFisher, 33211). Protein lysates were flash frozen and stored at -80C. For the analysis of Alpl secretion, media (serum free) from cell cultures were collected and flash frozen in liquid nitrogen until needed. Centrifugal filter units (MilliporeSigma UFC503096, 30 kDa cutoff) were used to concentrate protein from media per manufacturer’s instructions (500 μL media input, 30 min spin, 14,000×g). Concentrated media (1-2 μL/lane) was then used for western blot analysis. Streptavidin Bead Pulldowns Streptavidin pulldowns were performed as part of this study 28 . Streptavidin magnetic beads (Pierce 88817) were washed (using a magnetic stand for separation) twice in lysis buffer (see above) and resuspended in lysis buffer. For mouse teratomas, 9 mg of total protein per sample was combined with 0.9 mL of beads. For mouse serum, 2.5 mg of total protein per sample was combined with 0.25 mL of beads. Samples were incubated overnight at 4°C with end-over-end mixing. On the next morning, beads were washed twice in lysis buffer, once in 2 M urea in 10 mM Tris (pH=8.0), and twice in lysis buffer. For mouse sample western blots, streptavidin magnetic beads (Pierce, 88817) were 27 washed once and resuspended in RIPA complete lysis buffer (above). Pulldown reactions were setup as follows: 400 𝜇L RIPA complete lysis buffer, 10 𝜇L beads/100 𝜇g protein. Pulldown reactions either used 100 or 200 𝜇g of input protein. Reactions were incubated at 4°C overnight, rotating. The following day, beads were washed on a magnetic rack twice with 500 𝜇L of RIPA complete lysis buffer, followed by 500 𝜇L 2M Urea in 10 mM Tris, and twice again with 500 L RIPA complete lysis buffer. Beads were resuspended in 12 𝜇L Li-Cor loading buffer (Li-Cor, 928-40004) with 1.43 M 𝛽-mercaptoethanol. See the Western blotting section and Quantitative mass spectrometry section for further bead processing. Western Blot (fluorescent) Analysis Following pulldowns, beads were resuspended in 12 𝜇L Li-Cor loading buffer (Li- Cor, 928-40004) with 1.43 M 𝛽-mercaptoethanol. Beads were boiled at 95C for 5 minutes to elute proteins. Equal amounts of total protein were mixed with 1X Li-Cor loading buffer (above) and boiled at 95C for 5 minutes. Samples were briefly spun down and placed on ice before loading. Total protein or pulldown protein elutes were loaded on 10% SDS acrylamide gels and ran using standard 1X SDS-Running buffer with Li-Cor 4 𝜇L one-color molecular marker (Li-Cor, 928-40000). Samples on the gel were transferred to methanol activated PVDF 0.45 m membranes using Biorad’s wet tank mini-protean system for 1-3 hours at 300 constant mA in a sample dependent context. After transfer, membranes were dried at 37°C for 5 minutes and then re-activated with methanol. Blots were stained with Li-Cor’s Revert-700 Total Protein Stain (Li-Cor, 926-11010) for normalization and imaged using a Li-Cor Odyssey Clx. Blots were then de-stained per kit instructions and put in block (Li-Cor Intercept block, 927-60001) for 1 hour, room 28 temperature, shaking. Blots were then transferred to primary antibody (block with 0.1% Tween20) overnight at 4°C, shaking. Primary antibodies: Rb Adiponectin (Abcam, ab181699, 1:1000), Rb C7 (Abcam, ab126786, 1:1000), Rb Apoa1 (Abcam, ab20453, 1:1000), Rb Glucosidase sub-unit beta (Abcam, ab134071, 1:1000), Rb Calr (Abcam, ab92516, 1:1000), Rb myc-tag (Abcam, ab9106, 1:1000), Gt Alpl (mouse alkaline phosphatase; R&D systems, AF2910, 1:1000). The following day, blots were washed four times in TBS-T for 5 minutes each at room temperature, shaking, and then incubated in secondary antibody (1:10,000) in block with 0.1% Tween20 and 0.1% SDS, and/or streptavidin conjugate (1:5,000; 680 or 800, Li-Cor, 926-68079, 926-32230) if visualizing biotinylated proteins, for 1 hour at room temperature, shaking. Blots were then washed twice with TBS-T for 5 minutes each, room temperature, shaking, followed by two 5-minute TBS washes at room temperature, shaking. Blots were imaged on a Li-Cor Odyssey Clx. After imaging blots were dried at 37C for 5 minutes, then stored. Western Blot (fluorescent) Quantification Biotinylation levels were quantified via western blot using Li-Cor’s fluorescent western blot Emperia Studio (Version 1.3.0.83) and ImageStudio (Version 5.2.5) analysis software and protocols. Total protein stain images of each blot were used to normalize biotinylation (streptavidin) signal intensity in R by determining the lane normalization factor (Li-Cor) for each blot, which was then used to normalize protein of interest signals. ggplot2 and GraphPad Prism 8.0 were used to visualize normalized biotinylated protein signal. Quantitative Tandem Mass-Tag (TMT) MS of BirA* Mouse Samples Protein lysates and pulldowns were performed as described above (Protein lysate 29 preparation, and Streptavidin beads pulldowns sections). The different MS samples and their associated TMT labels are presented in Supplementary Figures 2.1a, 2.4a. i. On-bead digestion. Samples collected and enriched with streptavidin magnetic beads from mouse serum/teratomas were washed twice with 200 µL of 50mM Tris-HCl buffer (pH 7.5), transferred into new 1.5 mL Eppendorf tubes, and washed 2 more times with 2 M urea/50 mM Tris (pH 7.5) buffer. Samples were incubated in 0.4 µg trypsin in 80 µL of 2 M urea/50 mM Tris buffer with 1 mM DTT, for 1 hour at room temperature while shaking at 1000×g. Following pre-digestion, 80 µL of each supernatant was transferred into new tubes. Beads were washed twice with 60 µL of 2 M urea/50 mM Tris buffer, and these washes were combined with the supernatant. The eluates were spun down at 5000×g for 1 min and the supernatant was transferred to a new tube. Samples were reduced with 4 mM DTT for 30 min at room temperature, with shaking. Following reduction, samples were alkylated with 10 mM iodoacetamide for 45 min in the dark at room temperature. An additional 0.5 µg of trypsin was added and samples were digested overnight at room temperature, while shaking at 700×g. Following overnight digestion, samples were acidified (pH<3) with neat formic acid (FA), to a final concentration of 1% FA. Samples were spun down and de-salted on C18 StageTips as previously described 54 . Eluted peptides were dried to completion and stored at -80°C. ii. TMT labeling of peptides. Desalted peptides were labeled with TMT (10-plex) reagents (Thermo Fisher Scientific). Each 0.8 mg vial of TMT reagent was resuspended in 41 µL of MeCN. Peptides were resuspended in 100 µL of 50 mM HEPES and labeled with the TMT reagents as described in Supplementary Figures 2.1a, 2.4a. Samples were incubated at RT for 1 hour with shaking at 1000×g. TMT reaction was quenched with 8 µL of 5% hydroxylamine at 30 room temperature for 15 min with shaking. TMT labeled samples were combined, dried to completion, reconstituted in 100 µL of 0.1% FA, and desalted on StageTips. iii. SCX stage tip fractionation of peptides. Single shot analyses were performed on 50% of each sample. Desalted peptides were re-suspended in 9 µL of 3% MeCN/0.1% FA and 4 µL was injected on a HF-X (mouse samples) mass spectrometer (Thermo Fisher Scientific, see methods below). In addition, flow throughs of combined samples were also analyzed via single shot and fractionation. To increase depth-of-coverage, 50% of the TMT labeled peptide sample was fractionated by strong cation exchange (SCX) StageTips packed with 3 disks of SCX (3M Empore) material below 2 disks of C18 material. StageTips were conditioned with 100 µL of 100% MeOH, followed by 100 µL of 0.5% acetic acid/80% MeCN. Next, StageTips were equilibrated with 100 µL of 0.5% acetic acid, followed by 100 µL of 0.5% Acetic Acid/ 500mM NH4AcO/ 20% MeCN, and finally with 100 µL of 0.5% acetic acid. Peptide samples were resuspended in 250 µL of 0.5% acetic acid and loaded onto the StageTips, washed 2x with 100 µL of 0.5% acetic acid, and transeluted from C18 material onto the SCX material with 100 µL of 0.5% acetic acid/80% MeCN. Three step-wise elutions from the C18 disks were completed as follows: the first fraction was eluted with 50 µL of 50 mM NH 4 HCO 3 : 20% MeCN (pH 5.15, adjusted with acetic acid), the second with 50 µL of 50 mM NH 4 HCO 3 : 20% MeCN (pH 8.25, adjusted with acetic acid), the third with 50 µL 50 mM NH 4 OH: 20% MeCN (pH 10.3, adjusted with acetic acid). Each eluate was collected separately and 200 µL of 0.5% acetic acid was added to each. Each fraction was desalted on C18 StageTips as described above, and samples were dried to completion. iv. Liquid chromatography and mass spectrometry. Desalted peptides were re-suspended in 9 µL of 3% MeCN/0.1% FA and analyzed by online nanoflow liquid chromatography tandem mass spectrometry (LC-MS/MS) 31 using a Q Exactive HF-X or an Q Exactive Plus mass spectrometer (QE+, Thermo Fisher Scientific) coupled on-line to a Proxeon Easy-nLC 1200 (Thermo Fisher Scientific) as previously described 29 . Briefly, 4 µL of each sample was loaded at onto a microcapillary column (360 μm outer diameter × 75 μm inner diameter) containing an integrated electrospray emitter tip (10 μm), packed to approximately 24 cm with ReproSil-Pur C18- AQ 1.9 μm beads (Dr. Maisch GmbH) and heated to 50°C. SCX fractionated samples were analyzed twice using two 4 µL injections using a 110 min LC-MS method using a Q Exactive HF-X (mouse original samples and first injection of flow through samples) or a Q Exactive Plus (second mouse flow through injections) mass spectrometer. Mobile phase flow rate was 200 nL/min, comprised of 3% acetonitrile/0.1% formic acid (Solvent A) and 90% acetonitrile /0.1% formic acid (Solvent B). The 110-minute LC-MS/MS method used the following gradient profile: (min:%B) 0:2; 1:6; 85:30; 94:60; 95:90; 100:90; 101:50; 110:50 (the last two steps at 500 nL/min flow rate). Non-fractionated samples (single shots) were analyzed using a 260 min LC-MS/MS method on a Q Exactive HF-X (mouse samples) mass spectrometer. This method had the following gradient profile: (min:%B) 0:2; 1:6; 235:30; 244:60; 245:90; 250:90; 251:50; 260:50 (the last two steps at 500 nL/min flow rate). The Q Exactive HF-X or Q Exactive Plus were operated in the data-dependent mode acquiring HCD MS/MS scans (r = 15,000 for HF-X, 17,500 for QE+) after each MS1 scan (r = 60,000 for HF-X, 70,000 for QE+) on the top 12 (QE+) or top 20 (HF-X) most abundant ions using an MS1 target of 3 × 10 6 and an MS2 target of 5 × 10 4 . The maximum ion time utilized for MS/MS scans was 120 ms; the HCD-normalized collision energy was set to 31 (HF-X) or 28 (QE+); the dynamic exclusion time was set to 20 s, and the peptide match and isotope exclusion functions were enabled. Charge exclusion was enabled for charge states that were unassigned, 1 and >7. 32 BirA* MS Data Analysis All protein trafficking MS data were analyzed using Spectrum Mill software package v 7.00 (mouse data). Similar MS/MS spectra acquired on the same precursor m/z within +/- 60 s were merged. MS/MS spectra were excluded from searching if they were not within the precursor MH+ range of -600-6000 Da (mouse data) or if they failed the quality filter by not having a sequence tag length >0. These are standard exclusion criteria, and the precursors outside this mass range are typically not peptides. MS/MS spectra were searched against all mouse proteins annotated at UniProt database 53 containing 46,519 (mouse) and 264 (mouse) common contaminants. All spectra were allowed +/- 20 ppm mass tolerance for precursor and product ions, 40% (mouse) minimum matched peak intensity, and “trypsin allow P” enzyme specificity with up to 2 missed cleavages. The fixed modifications were carbamidomethylation at cysteine, and TMT6 at N-termini and internal lysine residues. Variable modifications included oxidized methionine and N-terminal protein acetylation and deamination (mouse only). Individual spectra were automatically designated as confidently assigned using the Spectrum Mill autovalidation module. Specifically, a target-decoy based false-discovery rate (FDR) scoring threshold criteria via a two-step auto threshold strategy at the spectral and protein levels was used. First, peptide mode was set to allow automatic variable range precursor mass filtering with score thresholds optimized to yield a spectral level FDR of <1.2%. A protein polishing autovalidation was applied to further filter the peptide spectrum matches using a target protein-level FDR threshold of 0. Following autovalidation a protein-protein comparison table was generated, which contained experimental over control TMT ratios. For all experiments, non-mouse contaminants and reverse hits were removed. Furthermore, data was median normalized. Next, we established threshold TMT ratios for hit-calling using positive control 33 (PC) secreted/receptor and negative control (NC) intracellular lists. For the PC list, we used fly receptor and secretome lists 123 , as well as fly orthologs (using DIOPT 91 ) of human receptome 124 , human secreted proteins annotated at UniProt 53 and blood plasma MS (highly confident cumulative multiple-dataset PeptideAtlas list 125 , and a list from humans with trauma 126 ). PC proteins were also checked for presence of a signal peptide 127 and a transmembrane (TM) domains 128 . For the NC list, we used high-confidence mitochondrial proteins 118 , as well as cytoskeletal and high confidence transcription factors (nuclear) 123 . Proteins identified by our BirA*G3-ER MS were compared to the PC and NC lists and assigned to as being a PC or NC protein. For each BirA*G3-ER comparison, we plotted the fraction above each log 2 BirA*/control TMT ratio, and determined a threshold TMT ratio at which was generally greater than 0.9. Enrichment score (E-S) was defined as the number of comparisons in which a specific protein’s TMT ratio exceeds the threshold TMT ratio. The PC list were UniProt annotated secreted proteins, while the NC list was the UniProt overlapping list of transcription factors and nuclear proteins, and cytoskeletal genes 118 . Note that the NC list was compared with secreted, receptors, ER proteins, and overlapping genes were removed. The threshold TMT ratio was chosen at which (FPR=0.1 for teratomas and FPR=0 for serum). Proteins with E-S of 0 are background, proteins with E-S of 1 are lower-confidence hits, and proteins with E-S of 9 are highest confidence hits. Note that in Fig. 4a-b and g, and Supplementary Figures 2.4c- h and k to l, 2.5g, k-l, means (and where applicable, SEMs) were calculated from log 2 values of TMT-ratios. To obtain human ortholog information, we used DIOPT, versions 5 and 6 91 . MS- identified proteins at different scores were examined for the presence of putative signal peptides, using SignalP program 127 (SignalP4., SignalP5.0 129 ). In addition, based on UniProt annotation 53 , we determined whether the identified proteins could be ER 34 resident. Using the SecretomeP program 130 , we examined whether some of the proteins could be unconventionally/non-classically secreted. In addition, protein abundance information was from integrated entire organism PAX database for mouse 122 . We compared the MS-identified proteins to fly and mouse orthologs (using DIOPT version 5.3 91 ) of mammalian adipocyte 79,135-143 and myocyte 144-151 secretomes. For higher confidence, we required a protein to be identified in at least 2 adipocyte or myocyte secretome datasets to be considered an adipocyte or myocyte secreted protein for our analysis in Supplementary Figure 2.5i-j. Mouse serum hits were also compared to mammalian whole blood (cells removed) proteomics datasets (highly confident cumulative multiple-dataset PeptideAtlas list 125 , and a list from humans with trauma 126 ). Mouse Immunofluorescent Analyses Frozen sectioned tissues were thawed at room temp for 10 min. The slides were rinsed in PBS for 5 min then blocked with 1.5% Seablock (ThermoFisher) in PBS + 0.25% TritonX block buffer for 1 hour at room temperature. The primary antibody mixture (diluted in block buffer) was added and the slides were incubated at 4°C overnight. Primary antibodies used in the study were as follows: GFP (Aves labs, GFP-1020, 1:500), RFP Tag (Invitrogen, MA5-15257, 1:500), BirA 6C4c7 (Abcam, ab232732, 1:300), Alexa Fluor™ Plus 647 Phalloidin (ThermoFisher, A30107, 1:1000). The secondary antibody used were Goat anti-Mouse IgG1 Cross-Adsorbed Secondary Antibody, Alexa Fluor 555 (ThermoFisher, A-21127, 1:1000) Donkey anti-Mouse IgG, Alexa Fluor 594, Donkey anti- Chicken Alexa Fluor 488), and streptavidin DyLight649 conjugate (Vector labs, SA-5649- 1, 1:1000). Slides were incubated with 1 mg/ml Hoechst 33342 (Molecular Probes) in PBS for 5 min to stain the nuclei prior to being mounted in ProLong Gold Antifade Reagent (Life technologies) and imaged at 10× or 63× using the Leica SP8 confocal microscope. 35 mESCs were cultured on glass coverslips for three days and then fixed in 4% PFA for 15 minutes, room temperature. Coverslips were washed three times with 1×PBS for 5 minutes each and then permeabilized in 0.25% Trition-X100 for 5 min. Coverslips were blocked in 1×PBS + 0.25% Triton-X100 + 2% Seablock (ThermoFisher, 37527) for 1 hour at room temperature. Primary antibody was diluted in block (above) and incubated for 16 hours at 4C. Primary antibody: mouse anti-BirA (Abcam, ab232732, 1:500). The following day, coverslips were washed with 1XPBS + 0.25% Trition-X100, four times for 5 min each. Secondary antibody and streptavidin conjugate were diluted in block (above) and incubated at room temperature for 1 hour. Secondary antibodies: Goat anti-Mouse IgG1 Cross-Adsorbed Secondary Antibody, Alexa Fluor 555 (ThermoFisher, A-21127, 1:1000), and streptavidin DyLight649 conjugated (Vector labs, SA-5649-1, 1:1000). Coverslips were then washed with 1×PBS + 0.25% Trition-X100, four times for 5 min each. Coverslips were incubated with 1 mg/mL Hoechst 33342 (Molecular Probes) in 1×PBS for 10 min to stain the nuclei prior to being mounted onto a slide with Fluoromount G (ThermoFisher, 00-4958-02) and imaged at 40× or 63× using the Leica SP8 confocal microscope. Histology Frozen sectioned tissues were thawed at room temperature for 10 min and rinsed in PBS for 5 min followed by a brief rinse in water. The sections were stained for 30-45 sec in hematoxylin then for 15-30 s in eosin and finally counterstained with fast red for 1 min. Samples were scanned at high resolution (10×) on a Zeiss Axio Scan.Z1 Slide Scanner to generate high-resolution tiled image files of the tissue section. Data Analysis and Statistics Data was analyzed using ZEN2012 software (Carl Zeiss), FIJI, Adobe Photoshop 36 CC2019, Microsoft Excel, Graphpad Prism 7, OriginPro 2017 or 2020, Emperia Studio analysis software (LiCor; Version 1.3.0.83) and ImageStudio (Version 5.2.5) for western blot analysis, R (version 4.0.0 (2020-04-24), Platform: x86_64-apple-darwin17.0 (64-bit); RStudio Version 1.3.959), Biorad Chemidoc MP acquisition software (ImageLab Touch 2.4), Nanodrop acquisition software (ND8000 version 2.2.0), and Specramax Paradigm acquisition software (Molecular Devices; Softmax Pro 6.2). Additionally, SignalP4.1, SignalP5, TMHMM (v 2.0), SecretomeP2.0 were used. Data is shown as mean ± standard error of the mean (SEM), unless otherwise indicated in the figure legend. Biological replicates are shown, unless otherwise indicated. In Fig. 2.4c-d, h-i; Supplementary Figures 2.4i-j, m-o, 2.5i-j and h using GraphPad Prism, two-sided Fisher’s exact tests were done. In Fig. 2.4e-f, using GraphPad Prism, a Kruskal-Wallis test with Benjamini, Krieger, Yekutieli Linear Two-Stage Step-Up FDR was done (serum: p=0.0386, Kruskal-Wallis statistic=16.27; teratoma: p<0.0001, Kruskal-Wallis statistic=125.1). Data Availability Statement The original mass spectra for all experiments, and the protein sequence databases used for searches have been deposited in the public proteomics repository MassIVE (https://massive.ucsd.edu) and are accessible at [ftp://massive.ucsd.edu/MSV000086664/] (mouse and fly BirA* datasets), and [ftp://massive.ucsd.edu/MSV000086291/] (fly hemolymph datasets). The following publicly-available datasets/databases were used: UniProt database mouse, and human; https://www.uniprot.org/), DIOPT (https://www.flyrnai.org/cgi- bin/DRSC_orthologs.pl), GLAD (https://www.flyrnai.org/tools/glad/web/), Signaling Receptome (DOI: 10.1126/stke.2003.187.re9), human plasma proteome datasets (DOI: 10.1074/mcp.M110.006353 (PeptideAtlas) and 10.1074/mcp.M600068-MCP200), 37 SignalP database (http://www.cbs.dtu.dk/services/SignalP/ and http://www.cbs.dtu.dk/services/SignalP-4.1/), Drosophila mitochondrial proteome (DOI: 10.1073/pnas.1515623112), TMHMM (http://www.cbs.dtu.dk/services/TMHMM/), SecretomeP (http://www.cbs.dtu.dk/services/SecretomeP/), FlyAtlas microarray (http://flyatlas.org/atlas.cgi), Drosophila RNAseq (DOI: https://doi.org/10.1038/nature12962), PAXdb (https://pax-db.org/), NCBI Gene (https://www.ncbi.nlm.nih.gov/gene/), FlyBase (https://flybase.org/), mammalian adipocyte secretomes (DOI: 10.1021/pr7006945, 10.1074/mcp.M600265-MCP200, 10.1021/pr049772a, 10.1074/mcp.M111.010504, 10.1021/pr800650r, 10.1021/pr100621g, 10.1016/j.molmet.2014.02.005, 10.1021/pr800650r, 10.1093/abbs/gmp085, 10.1074/mcp.M600217-MCP200), mammalian myocyte secretomes (DOI: 10.1002/prot.20803, 10.1074/mcp.M110.004804, 10.1021/acs.jproteome.5b00720, 10.1016/j.bbapap.2013.08.004, 10.2337/db08-0943, 10.1152/ajpendo.00326.2011, 10.1371/journal.pone.0062008, 10.1074/mcp.M114.039651). The main text and supplementary information shows all of the data collected as part of this work. The source data are provided with this paper. The corresponding authors will provide original data on reasonable request. RESULTS Secretion of biotinylated proteins from BirA*G3-ER-expressing teratomas into serum in mice To transition the BirA* approach to mammals, we generated mouse embryonic stem cells (mESCs) with a Cre-inducible mouse codon optimized BirA*G3-ER inserted 38 into the Rosa26 “safe-harbor” locus (CAGGS-GFP/BirA*G3-ER-mKate2(R26); Fig. 2.1a). These cells show efficient Cre-dependent biotinylated protein secretion in vitro, including a positive control, Alpl, as well as Cre dependent BirA*G3-ER expression (Fig. 2.1b-k; Supplementary Figure 2.1). To further characterize BirA*G3-ER;Cre cells, mESCs or mESC derived fibroblast-like cells were cultured in varying concentrations of biotin (0-250 M biotin; Fig. 2.2a-b) or in 50 M biotin for different lengths of time (1 minute up to 12 hours; Fig. 2.2c-d). Optimal labelling was observed incubating cells in 50-100M biotin for 4-8 hours, but significant levels of biotinylation were observed within 5 minutes of biotin addition and 1 hour of labelling. We further used a fluorescent western blot to show that secretion of Alpl from mESCs is not significantly affected by BirA*G3-ER expression and activity (Supplementary Figure 2.1ee). To show the applicability of this approach in vivo, we injected 1-2×10 6 BirA*G3-ER (negative control without Cre) or BirA*G3-ER;Cre mESCs under the kidney capsule to generate teratomas 99,100 . Biotin was administered at 2000 ppm in the chow and 5 mM in the water for the final 7 days before tissue collection 4 weeks post transplantation (Fig. 2.3a). Teratomas were evaluated histologically (hematoxylin and eosin staining) for an expected multi-germ layer complex tissue morphology characteristic to this class of pluripotent cell derived tumor (Fig. 2.3b-c; Supplementary Figure 2.2a-j). Immunostaining of teratomas showed the expected outcomes: GFP was restricted to BirA*G3-ER mESC-derived tumors, and mKate2 and BirA exclusively to BirA*G3-ER;Cre mESC derived tumors (Fig. 2.3d-o; Supplementary Figure 2.2k-z). BirA*G3-ER-mKate2 expression co-localized with streptavidin in BirA*G3-ER;Cre teratomas suggesting Cre dependent biotinylation (Supplementary Figure 2.2s-z). Western blotting of tumor tissues demonstrated Cre dependent biotinylation at the total protein level (Supplementary Figure 2.3a) and after streptavidin pull-down (Fig. 2.3p). Consistent with 39 secretion of biotinylated proteins from the tumors, streptavidin pull-down of serum detected a strong biotinylated protein signature specifically in animals with BirA*G3- ER;Cre teratomas (Fig. 2.3q; Supplementary Figure 2.3b). As expected, BirA*G3 with ER retention signal was detected in teratomas but not in serum (Supplementary Figure 2.3c- d). To evaluate and quantify biotinylated proteins in teratomas and serum samples further, we used BirA*G3-ER and LC-MS/MS with TMT mass tags (Supplementary Figure Supplementary 2.4a-b). Using the analysis pipeline developed, we calculated the enrichment score (E-S) for each teratoma or serum protein (details in Supplementary Fig. 2.4c-f; see Methods). Using an E-S of at least 1, we detected 1641 labeled proteins present in the tumor and 291 in serum samples, including known hormones, growth factors, cytokines and proteins with poorly characterized functions. Proteins with higher average TMT-ratios have higher E-S, highest-confidence hits have E-S of 9, background proteins (non-hits) have E-S of 0 (Fig. 2.4a-b; Supplementary Figure 2.4g-h), and as expected, hits are enriched for PC proteins (Supplementary Figure 2.4i-j). The teratoma or serum biological replicates show good agreement in TMT-ratio signals (Fig. 2.4a-b). As expected and in agreement with the western blot data (see Supplementary Figure 2.3c-d), we identified 53 unique peptides for the BirA*G3 enzyme itself in the BirA*G3-ER;Cre teratomas (total intensity=5.56×10 12 , average log 2 TMT ratio=2.17±0.09, and score=9), but none in the serum. Our analysis of the teratoma proteome revealed that a higher E-S correlated significantly with more signal peptide-containing, ER-resident, and transmembrane proteins, suggesting that many of our hits are ER proteins (Supplementary Figure 2.4k; Fig. 2.4c; Supplementary Figure 2.4m-o; see Methods). Specifically, of the top 381 ranked predictions with an E-S of at least 8, 61% have a signal peptide, 36% are ER-resident, and 40 38% have transmembrane domains. In the teratoma-secreted proteome of the serum, a higher E-S correlated significantly with more signal peptide-containing proteins, suggesting that our hits are secreted (Supplementary Figure 2.4l; Fig. 2.4d; see Methods). Of the top 38 ranked predictions with an E-S of at least 8, 92% have a signal peptide. As expected, known ER-resident proteins were not enriched in labeled serum proteins (Supplementary Figure 2.4n; see Methods). We estimated protein abundances of the teratoma and serum candidates using the PAX database (see Methods) and determined that increased E-S showed enrichment for lower whole-body abundance proteins (Fig. 2.4e-f). These results are consistent with the enrichment of the most relevant proteins away from abundant contaminants using biotinylation by BirA*G3. We next compared the teratoma and serum proteomics datasets. We first showed that the top expressed (average reading cutoff = 0.5) control unlabeled background proteins from teratoma and serum show no significant enrichment for UniProt secretion annotation (Supplementary Figure 2.5a-b). The overlap between control serum and teratoma does not include annotated secreted proteins and may instead include most abundant background proteins in both samples that non-specifically bind to streptavidin beads in the enrichment protocol (Supplementary Figure 2.5c). By contrast, BirA*G3- ER;Cre-biotinylated serum and teratoma hits are enriched for secreted proteins and show a highly statistically-significant overlap with each other (p=3.78×10 -21 ; Supplementary Figure 2.5d-f). In addition, as the E-S increases, the fraction of teratoma hits that were identified in the serum increases; however, not all teratoma hits were identified in the serum, suggesting that only a fraction of proteins are secreted, or the steady state levels of some secreted proteins in the serum are below current levels of detection (Supplementary Figure 2.5g-h). Moreover, as the E-S increases, the fraction of serum hits that were identified in the teratoma increases (Fig. 24-h). Comparing the teratoma serum 41 proteome with that of previously-published whole serum plasma (see Methods) showed a distinct profile (Fig. 2.4i). Thus, the serum dataset is enriched for proteins present in the teratoma but not those generally present in the blood. Further analysis of teratoma streptavidin-enriched ER proteins and non-enriched proteins revealed lineage markers for mesoderm, endoderm, and ectoderm (Supplementary Table 2.2; see Methods). Tumors show strong muscle phenotypes histologically, and consistently, analysis of the highest confidence hits in serum showed an enrichment for myocyte, as well as adipocyte, secretomes (Supplementary Figure 2.5i- j; see Methods). Finally, using a complementary p-value-based approach for identifying hits (Supplementary Figure 2.5k-l) showed strong correlation with our E-S/thresholding- based method Supplementary Figure 2.5m-n). Using streptavidin pulldowns and western blotting, proteins observed to be highly enriched MS hits in BirA*G3-ER;Cre teratoma and serum were validated based on high expression and antibody availability. Multi-median mouse log 2 BirA*/wt ratios were used to find the most enriched proteins in serum of BirA*G3-ER;Cre over BirA*G3-ER (control) samples. Apolipoprotein A1 (Apoa1) and complement component 7 (C7) showed log 2 BirA*/wt ratios greater than 1.0 and adjusted p-value < 0.06. Western blots of streptavidin pulldowns of serum samples demonstrated specific enrichment in serum from BirA*G3-ER;Cre tumor harboring mice validating MS predictions (Fig. 2.5; Supplementary Figure 2.5o). For predicted teratoma hits, the ER proteins glucosidase sub-unit beta and calreticulin were present only in BirA*G3-ER;Cre streptavidin pulldowns, and Apoa1, C7, and complement C1q tumor necrosis factor-related protein 3 (C1qtnf3; an adiponectin paralog) were all significantly enriched in BirA*G3-ER;Cre streptavidin pulldowns (Fig. 2.5b). Lower Apoa1 and C7 in the BirA*G3-ER streptavidin pulldowns represents non-specific binding of these abundant serum proteins to 42 streptavidin beads as western blot analysis shows streptavidin (biotinylation) and Apoa1 or C7 co-localization only in BirA*G3-ER;Cre streptavidin pulldowns (Fig. 2.5b). DISCUSSION We have developed and applied a platform, using promiscuous BirA*, to investigate secreted protein trafficking from subcellular-compartments of specific cells to distal-organs in vivo (Fig. 2.3p-q, 2.5a-c). Our platform provides biochemical evidence suggestive of proteins trafficking to other organs or distal body parts and allows differentiation between ER- and non-conventionally-secreted proteins through the use of ER- or cytosolically-localized BirA*, respectively. Interestingly, our results indicate that the inter-organ communication network of secreted proteins is more extensive than previously thought. To transition the BirA* approach to mammals, we generated BirA*G3-ER- expressing kidney teratomas, and identified several known hormone or signaling proteins in the teratoma-derived serum proteomics. Although the amount of their production from the teratoma is unclear, these factors have been reported circulate in wild type serum on the order of ng/mL (Prl2c, resistin, insulin growth factor 2, angiopoietin, C1qtnf3, Postn, IGFBPs, PPBP), and pg/mL (colony stimulating factor 1). In our mouse serum, we identified proteins previously found in whole mammalian blood (see Methods), however, only some serum proteins were biotinylated, suggesting of specificity to teratoma-labeling (Supplementary Figure 2.6d). For example, albumin, the most abundant protein in whole blood was not significantly enriched in teratoma- derived biotinylated serum proteome (average log 2 TMT ratio=-0.51±0.17 for serum and - 0.27±0.13 for teratoma). 43 It is possible that biotinylation of proteins in the ER may affect the modification and sorting properties of labeled proteins. Future studies may also systematically examine if the secretome of a cell type under study is affected by biotinylation. Although we do not observe obvious leakage of BirA* to the circulation due to tissue damage, it is possible that biotinylation could lead to ER stress or modify the localization or activity of signaling components. Characterizing potential ER stress due to the BirA* system in mice is potentially confounded by teratoma-to-teratoma variability and diseased states (e.g., hypoxia) inherent to teratomas. Future studies of BirA* activity in normal mouse tissues will provide additional insight and facilitate the development of labeling strategies to minimize unwanted outcomes (e.g. by shortening the labeling time window). Our platform is widely applicable to research in inter-organ, local (within a tissue), or intracellular (e.g., within neuronal projections) protein trafficking, unconventional secretion, protein trafficking in co-culture systems, and to mouse or other organisms in vivo. In addition, transiently-secreted proteins or those released due to perturbations (such as diet, stress, tumors) 73 or circadian rhythms may be studied by inducing BirA* biotinylation using biotin for a limited time. This may be achieved using time-restricted feeding, systemic injections, or local delivery of biotin. With the proof-of principle tumor studies here, we expect mouse strains enabling Cre-dependent biotinylation in the secretory pathway in the organism will enable insights into cell/tissue/organ function and interaction under healthy or diseased conditions. MAIN FIGURES AND TABLE 44 Fig. 2.1 Rosa26 (R26) BirA*G3–ER and R26 BirA*G3–ER;Cre 3A mouse embryonic stem cell (mESCs) characterization. Schematic showing Cre-induced recombination of the R26 BirA*G3-ER allele to generate the BirA*G3- ER;Cre expressing cell line. GFP is in blue, BirA*G3-ER-myc is in black, mKate2 is in red, homologous regions to the R26 locus are in blue, and other elements are colored in gray (A). Native immunofluorescence in BirA*G3-ER parental (GFP+ (green)/mKate2−) or Cre-recombined (GFP−/mKate2+(red)) ESC colonies. DAPI (nuclei) is in blue. Representative images of four repeats. Scale bar: 100 μm (B-I). Western blot analysis of mESC whole cell lysate with or without Cre recombination (J-K). Total protein stain (red) (J). Western blot probed with streptavidin (cyan) or an anti-myc (red) antibody to detect myc-tagged BirA*G3 (K). For each image, all lanes are from the same blot (see Source Data for uncropped blots). Representative results of three western blots. See also: Supplementary Fig. 2.1. Source data are provided as a Source Data file. 45 Fig. 2.2 Characterization of in vitro biotinylation in the BirA*G3–ER mouse embryonic stem cell (mESC) line. Western analysis of streptavidin binding to biotinylated proteins following 24 h of labeling at the indicated concentrations of biotin (A). All lanes are from the same blot (see Source Data for uncropped blots). Quantification of biotin labeling relative to whole protein in A (B). Western analysis of streptavidin binding to biotinylated proteins labeled in 50 μM biotin for the indicated times. All lanes are from the same blot (see Source Data for uncropped blots) (C). Quantification of biotin labeling relative to total protein in c. In this figure, representative results are shown from two western blots each (D). Source data are provided as a Source Data file. 46 Fig. 2.3 Analysis of biotin labeling of proteins in vivo in a mouse embryonic stem cell (mESC)–derived teratoma model. Schematic representation of the teratoma study. GFP+ BirA*G3-ER mESC are shown in green, and mKate2+ BirA*G3-ER;Cre mESC are in red (A). Hematoxylin and eosin staining of cryosectioned teratoma tissue (B- 47 C). High magnification images. Representative results of three replicate images. Scale: 100 μm (B’-C’). Analysis of reporter gene activity and BirA*G3 in cryo-sections of teratoma tissue. Scale bar: 50 μm. Green is GFP, red is mKate2, cyan is BirA, magenta is phalloidin, and white is DAPI. Representative images from 2 biological replicates with 2 sections per replicate. Another set of images was also acquired in= testing the binding of antibodies (D-O). Streptavidin labeling of biotinylated proteins from streptavidin pulldown of tumor (P) and serum samples (Q). Note: the stronger signal in sample 2 in q reflects the significantly larger tumor mass in this individual. Representative results from six (P) and four (Q) western blots (P-Q). For each image, all lanes are from the same blot (see Source Data for uncropped western blots). See also: Supplementary Figs. 12 and 13. Source data are provided as a Source Data file. 48 49 Fig. 2.4 Identification of mouse teratoma-derived proteins in serum. Teratoma and serum log2(BirA*G3-ER/wt) TMT ratios in three replicates. Each point is n = 3 comparisons, mean ± SEM log2TMT ratio. The points are colored by the enrichment score (E-S): the number of comparisons (from 9) in which TMT-ratio > threshold (score 9 is for most confident hits [red] and 0 is background [black]) (A, B). As the E-S increases, the fraction of proteins with putative signal peptides increases. Two-sided Fisher’s exact test. Teratoma p-values (****) from left to right (c): 5.11 · 10−198, 7.86 · 10−221, 1.03 · 10−232, 5.11 · 10−241, 2.05 · 10−230, 1.66 · 10−216, 4.73 · 10−194, 3.10 · 10−167. Serum p-values (****) from left to right (d): 1.00 · 10−30, 3.61 · 10−33, 1.82 · 10−35, 2.48 · 10−35, 1.23 · 10−36, 1.37 · 10−31, 1.63 · 10−24, 8.42 · 10−18 (C, D). BirA*G3-ER;Cre teratoma and serum hits were enriched for lower-abundance proteins. Protein abundance information was from an integrated entire organism PAX database for mouse67. Frequency vs log10 protein abundance plots. For representation, serum and teratoma data are shown as histograms with equal bin sizes with B-spline smooth fits (calculated using OriginPro 2020). Statistics were performed on original data. Kruskal–Wallis test and Benjamini, Krieger, Yekutieli Linear Two-Stage Step-Up FDR (two-tailed p-value). Teratoma p-values (****) from top to bottom (E): 0.000071, 0.000025, 0.000009, 6.72 · 10−12, 4.26 · 10−11, 4.26 · 10−11, 1.02 · 10−8, 6.79 · 10−7. Serum pvalues (****) from top to bottom (F): 0.0004, 0.0005, 0.0005, 0.0005, 0.0005, 0.0009, 0.0041, 0.0204 (E, F). As the E-S increases, the fraction of serum hits that were identified in the teratoma increases (G–H colored as red in G). Proteins with increased scores are not enriched for proteins found in previous whole blood (cells removed) proteomics (I) (see “Methods” section). Two-sided Fisher’s exact test. p-values from left to right (H): 0.0044, 0.0002, 0.000025, 0.000005, 0.000012, 0.000005, 0.000001, 0.0001. p-values from left to right (I; N.S. means non-significant): 0.9161, 0.9136, 0.8243, 0.9093, >0.9999, 0.7987, 0.0850, 0.2001 (G-I). See also: Supplementary Figs. 2.4 and 2.5a–n, Supplementary Table 2.2. Source data are provided as a Source Data file. Fig. 2.5 Validation of mass spectrometry proteomics-based predictions in tumor and serum samples. LC–MS/MS log fold change (FC) of multimedian normalized data of BirA*G3-ER;Cre samples over BirA*G3-ER samples with adjusted (Adj.) p-value from two-sided two-sample t tests for selected targets. 50 Note: Adipoq has only 1 unique peptide from the MS data (A). Western blot analysis of target hits in streptavidin pulldown of tumor lysates (B) and serum samples (C). No serum was available from sample 01 for the BirA*G3-ER;Cre analysis in c. Each slice is from a different blot, all lanes on a slice are from the same western blot (see Source Data for uncropped blots) (B, C). Serum Apoa1 western blot was done twice (Supplementary Fig. 2.5o), while other western blots were done once, but using three (B) or two (C) biological replicates. IB means immunoblot. See also: Supplementary Fig. 2.5o. Source data are provided as a Source Data file. 51 Supplementary Figure 2.1: mESC streptavidin and BirA expression characterization, and positive control Alpl secretion analysis (Fig. 2.1 supplement). 52 Immunofluorescent staining of mESC colonies from BirA*G3-ER (B1 parent line) or BirA*G3-ER;Cre (Cre recombined 3A line) allele reporter expression of mKate2 (red) or GFP (green) with BirA (white) and streptavidin (cyan) immunostaining. Representative results from four repeats. Scale bar: 50 μm (A-X). Additional mESC clones show efficient Cre-dependent biotinylated protein secretion in vitro, including a positive control, Alpl. In (DD) asterisk (*) points to a non-specific band, whereas the arrow is the secreted Alpl band. For each slice, all lanes are from the same western blot (see Source Data for uncropped blots), IB = immunoblot (BB-DD). Is a representative image (one was taken), and (Z-AA) are representative images from two acquired (Y). The western blots in (BB-DD) were performed once, but similar results were obtained with another mESC clone in Fig. 2.1j-k. Alpl secretion is unaffected upon BirA*G3-ER expression in mouse embryonic stem cells (mESCs) (EE). Alkaline phosphatase (Alpl) secretion into media was compared by fluorescent western blot analysis31, varying BirA*G3-ER and biotin labeling (50 μM supplementation to media for 12 hours) as indicated. This experiment was performed in serum-free OPTI- MEM medium. Note: Streptavidin (680 channel) and ALPL (800 channel) were probed on the same western blot in separate channels. Slice views are from the same blot, but separated for individual channels. Secreted proteins in media were concentrated by centrifugal filters with a 30 kDa cutoff. Quantification of Alpl western blot normalized to total protein using Li-Cor Empiria Studios software and normalization protocols is shown on the bottom of the figure. rAlpl is recombinant Alpl. This result is a representative of three biologically independent repeats (each repeat was run on the western blot twice) (Y-DD). Source data are provided as a Source Data file. 53 54 Supplementary Figure 2.2: Teratoma characterization (Fig. 2.6 supplement). Images of teratomas harvested from mice after 4 weeks of growth under the kidney capsule compared to normal kidney (A-E). Hematoxylin and eosin staining of cryosectioned teratomas generated from BirA*G3- ER or BirA*G3-ER;Cre mESCs (F-J). Immunofluorescence of cryosectioned teratomas generated from BirA*G3-ER or BirA*G3-ER;Cre mESCs showing native fluorescence of reporters (GFP in green and mKate2 in red), streptavidin (cyan), and DAPI (blue) (K-Z). Teratoma experiments are representative results from n=4-5 per BirA*G3-ER or BirA*G3-ER;Cre teratoma generated. The immunofluorescence results are representative from two repeats. (F-J, N, R, V, Z) Scale: 100 μm. Supplementary Figure 2.3: Teratoma and serum streptavidin and BirA*G3-ER-myc western blots (Fig. 2.6 supplement). Streptavidin western blot of teratoma protein lysate (25 μg, a) and total serum (5 μg, b) from mice with BirA*G3-ER or BirA*G3-ER;Cre teratomas. All lanes are from the same western blot (see Source Data for uncropped blots) (A). All lanes are from the same western blot (see Source Data for uncropped blots) (B). Are representative western blots from five repeats each (A-B). BirA*G3-ER-myc was detected in teratomas but not in serum (C-D). The BirA*G3-ER (C) and BirA*G3- ER;Cre (D, red text) samples were run on different gels at the same time, and were exposed for the same length of time. BirA*G3-ER expressing cells in culture were used as controls. All lanes are from the same western blot (see Source Data for uncropped blots) (C). All lanes are from the same western blot (see Source Data for uncropped blots) (D). This western blot was done once, however, BirA*G3 peptides were also searched for in the mass spectrometry data. Source data are provided as a Source Data file. 55 Supplementary Figure 2.4: Thresholding and signal peptide analyses of biotinylated serum and teratoma proteins identified using tandem mass-tag (TMT) mass spectrometry (MS) using BirA*G3-ER labeling (Fig. 2.4 supplement). Experimental setups for the identification of serum and teratoma biotinylated proteins. The different TMT 56 state signals were compared to generate TMT ratios (right of the arrows). There were 9 TMT ratio comparisons for teratomas and 9 for serum (A-B). Teratoma and serum log2(BirA*G3-ER / wt) TMT-ratios in three replicates, with each point is n=3 comparisons, mean±SEM log2TMT ratio. Proteins identified with MS were compared to positive control (PC) secreted protein/receptor (red points) and negative control (NC, intracellular; black points) (C-D). Representative #NC or #PC versus BirA*G3-ER/wt TMT ratio graph (out of nine). A threshold TMT ratio was chosen for which %NC/%PC ≤ FPR (false positive rate; FPR=0.1 for teratomas and FPR=0 for serum) (E-F). Increased TMT-ratios are associated with higher Enrichment Scores (E-S). Each point is a mean±SEM log2TMT ratio for each identified protein. Linear regression results of a two-tailed F-test are presented (G-H). Hits (Enrichment Score ≥ 1) have higher #PC/(#PC+#NC). Statistics: Two-sided Fisher’s exact test. Teratoma p-values (****) from left to right (i): 5.92•10-99, 3.80•10-103, 1.49•10-99, 3.22•10-91, 8.01•10-88, 3.66•10-80, 1.50•10-69, 8.40•10-56. Serum p- values (****) from left to right (j): 4.35•10-17, 4.72•10-16, 1.77•10-15, 1.28•10-14, 8.60•10-14, 5.53•10-12, 4.14•10-8, 0.000002 (I-J). Proteins with an identified signal peptide25 (red) were mapped onto the log2(BirA*G3-ER/wt) TMT-ratios in three replicates graph. Each point is n=3 comparisons, mean±SEM (K- L). Teratoma hits are enriched for ER-resident proteins, while serum hits are not enriched for ER-resident proteins. Statistics: Two-sided Fisher’s exact test. Teratoma p-values (****) from left to right (M): 1.75•10- 30, 8.23•10-30, 2.55•10-31, 8.57•10-36, 7.66•10-34, 9.17•10-34, 1.20•10-34, 1.78•10-33. p-values from left to right in (N) (N.S. means not significant): 0.5692, 0.3852, 0.3749, 0.2306, 0.3422, 0.2416, 0.6609, 0.3153 (M-N). Teratoma hits are enriched for transmembrane-domain (TM) containing proteins. Two-sided Fisher’s exact test. Teratoma p-values (****) from left to right: 5.69•10-77, 6.93•10-78, 1.19•10-77, 7.11•10-82, 4.94•10-70, 3.01•10-61, 1.06•10-53, 5.94•10-45 (O). Source data are provided as a Source Data file. 57 Supplementary Figure 2.5: Additional and complementary analyses of teratoma and serum datasets. Complementary secretory pathway analysis: the top expressed (average reading cutoff = 0.5) control unlabeled background proteins from teratoma and serum show no significant enrichment for UniProt secretion annotation (A-B). The overlap between control serum and teratoma does not include annotated 58 secreted proteins, and may instead include most abundant background proteins in both samples that non- specifically bind to streptavidin (C). By contrast, BirA*G3-ER;Cre-biotinylated serum and teratoma hits (Enrichment Score (E-S) ≥ 5) are enriched for secreted proteins and show highly statistically-significant overlap with each other (p=3.78×10-21; D-F). Statistics: hypergeometric test (one-sided p-values) (A-F). As E-S increases, the fraction of teratoma hits that were identified in the serum increases; however, not all teratoma hits were identified in the serum, suggesting that only a fraction of proteins is secreted. Teratoma p-values (****) from left to right: 4.63•10-32, 6.06•10-38, 3.86•10-40, 2.34•10-36, 1.51•10-34, 6.85•10-34, 4.00•10-32, 1.39•10-30 (G-H). Serum hits are enriched for mammalian adipocyte (I) and myocyte (J) secretome orthologs (see Methods). Two-tailed Fisher’s exact test. p-values (I; left to right): 0.0624, 0.0416, 0.0636, 0.0234, 0.0348, 0.0460, 0.0003, 0.0033. p-values (J; left to right): 0.4278, 0.2941, 0.4016, 0.2237, 0.1637, 0.1737, 0.0075, 0.0110 (I-J). A p-value-based alternative approach for identifying hits. Volcano plots of average log2(BirA*G3-ER / wt) TMT-ratios (mean ± standard error of the mean; n=9 comparisons). Hits (red) were defined as having p<0.1 (serum) and p<0.01 (teratoma) and fold-change>1. p-values were calculated by comparing BirA*G3-ER/mean of all samples and wt/mean of all samples TMT signal ratios (n=3 each) using two-sample t-tests (K-L). The p-value-based approach for identifying hits (K-L) showed strong correlation with the score/thresholding-based method (M-N). Western blot analysis of target hits in streptavidin pulldown of serum samples. This is an analysis of extended set of samples from Fig. 2.5c for which sample 01 from BirA*G3-ER;Cre was available. This is a representative result from two western blots (the other one is in Fig. 25c). All lanes are from the same blot (see Source Data for uncropped blots) (O). Source data are provided as a Source Data file. Supplementary Table 2.1: Oligonucleotides used in this study 59 Supplementary Table 2.2: BirA*G3-ER teratoma streptavidin enriched (Enrichment score≥1) and non- enriched (Enrichment score=0) proteome lineage markers 60 Chapter 3 A genetic model for in vivo proximity labelling of the mammalian secretome This work has been published in Open Biology (PMID 35946312). This work was led by Rui Yang, Ilia Droujinine, and myself with data collection contribution by Jinjin Guo, Jill A. McMahon, Yanhui Hu, Dominque K. Carrey, Charles Xu, David Rocco. Primary experimental design, data collection, analysis, and manuscript preparation was carried out by Rui Yang and myself. Steven A. Carr, Namrata Udeshi, Jihui Sha, and James Wohlschlegel supervised proteomic experiments and analysis. Qiao Fang and Shishang Qin contributed to script generation for proteomic analysis. Alice Y. Ting provided the BirA*G3 construct, critical insight on proximity labeling based methods, and advised on experimental design, and manuscript preparation and review. Andrew P. McMahon and Norbert Perrimon conceptualized, supervised, and advised on experimental design, data analysis, and manuscript preparation and review. INTRODUCTION Protein secretion plays a critical role in coordinating local and systemic cellular responses in development, homeostasis, and disease 169-172 . Multi-organ failure suggests an aberrant organ–organ crosstalk resulting in linked organ pathology such as pulmonary–renal syndromes 173,174 , with an increased risk of sequential organ failure and morbidity 175 . Secreted proteins may be identified using liquid chromatography-tandem mass spectrometry (LC-MS/MS) proteomics of serum 78 . However, it is challenging to 61 identify low abundance proteins and difficult to track the organs of origin and ultimate destination of protein interactions 176,177 . Analysis of cell secretomes has benefited from enzyme-catalyzed proximity labeling approaches such as BioID 17 and TurboID 6 . In these, the activity of a promiscuous biotin-ligase in the cellular secretory pathway biotinylates resident and secreted proteins, which can then be detected by affinity enrichment and quantitative MS 6,17 . These labeling methods allow sensitive and stable detection of endogenous secreted proteome in live cells, including fly (Drosophila melanogaster) 6,50 and worm (Caenorhabditis elegans) models 6 , mouse tumor transplants 50 and specific mammalian target tissues through viral directed gene delivery 40,49,178 . These studies have provided new insight into tissue secretomes and inter-organ communication 10,39,48,49,178,179 . To overcome the limitations of viral-mediated approaches for systematic temporal and spatial analysis of mammalian cell secretomes in vivo, we generated and validated a mouse model system. In the secretome reporter strain, DNA sequences encoding an endoplasmic reticulum directed promiscuous biotin ligase, BirA*G3 6 , were inserted into the ubiquitously expressed Rosa26 locus 50 . BirA*G3 is a precursor to TurboID, generated in the directed evolution of E. coli BirA, that has a higher affinity for biotin and may be able to catalyze biotinylation prior to addition of exogenous biotin 6 . Conditional (CRE recombinase- and exogenous biotin-dependent) BirA*G3 activity resulted in rapid biotinylation of proteins trafficking through the secretory pathway and permitted the analysis of cellular secretomes through streptavidin affinity purification and quantitative mass spectrometry proteomics 6,50 . METHODS 62 Animal studies Institutional Animal Care and Use Committees (IACUC) at the University of Southern California reviewed and approved all animal work as performed in this study. All work adhered to institutional guidelines. mESCs (B6(Cg)-Tyr<c-2J>/J, Stock No.: 000058, The Jackson Laboratory) 214 carrying the loxP-flanked BirA*G3 were aggregated with 8-cell C57bl/6J embryos to obtain chimeras. The resulting chimeric mice were bred with R26PhiC31 (B6.129S4-Gt(ROSA)26Sortm3(phiC31*)Sor/J, Stock No.: 007743, The Jackson Laboratory, backcrossed to C57bl/6J for 13 generations) 215 females to remove attB-neoR-attP cassette. The resulting mice carry the BirA*G3 allele, namely BirA*G3 mice. The Sox2-Cre mice (B6.Cg-Edil3Tg(Sox2-cre)1Amc/J, stock no.: 008454, The Jackson Laboratory) 183 , the TdTomato mice (B6.Cg-Gt(ROSA)26Sortm14(CAG- tdTomato)Hze/J, Stock No.: 007914, The Jackson Laboratory)(24), the CAGGCre-ERTM mice(FVB.Cg-Tg(CAG-cre/Esr1*)5Amc/J, stock no.: 017595, The Jackson Laboratory) 216 , and the Alb-Cre mice (B6.Cg-Speer6-ps1Tg(Alb-cre)21Mgn/J, stock no.: 003574) 217 were used as described previously. In vivo Assays For all in vivo assays, tissues were collected as follows. Mice were euthanized at 8-32 weeks, blood was then collected from the inferior vena cava, followed by perfusion with 1×cold DPBS (Dulbecco’s Phosphate Buffered Saline). The blood was allowed to clot at room temperature for 30 minutes and then spun down at 2,000 x g for 15 minutes at 4°C. The serum was collected and spun again at 2,000 x g for 15 minutes at 4°C, then removed to a fresh tube and flash frozen (in liquid nitrogen) before being stored at -80°C until being used. After perfusion, tissues were collected and rinsed in 1×cold DPBS before being minced with a razor blade and aliquoted into tubes. Tissues were then flash frozen 63 and stored at -80°C until being used. For the biotin administration method study, Sox2-BirA*G3 and control mice at 8- 12 weeks were given biotin via water (n=2) (5mM, pH 7.4; Sigma B4639-5G), chow (n=2) (2,000ppm; LabDiet, SWLP), or both (n=2) for 7 days. After 7 days, serum was collected and stored as described above. For the biotin and Cre dependence labeling study, Sox2-BirA*G3 and control mice at 8-12 weeks were given biotin chow (2,000ppm) or normal diet (LabDiet 5053) for 7 days. After 7 days, serum, liver, brain, and kidney were collected as described above. For the BirA*G3 temporal labeling assay, Sox2-BirA*G3 (n=3/timepoint) and control mice (n=3) at 8-12 weeks were given biotin by subcutaneous injection (100μL 180mM sterile biotin water, pH 7.4) and water (5mM, pH 7.4) after the injection until collection. Serum, liver, kidney, and brain were collected from mice at 1, 2, 4, 6, 10, and 12 hours post-injection. Serum was collected as described above. For all other mouse studies, Sox2-BirA*G3 and control mice at 8-32 weeks were given biotin chow (2,000ppm) for 5 days. Tissues were then collected as described above. For ERT2 mice, tamoxifen (2mg/40g body weight in corn oil; Sigma cat. T5748-1G) was administered by abdominal injection twice, 3 days apart prior to biotin studies. In vitro Assays To generate endoplasmic reticulum stress positive controls 80% confluent mouse 3T3 cells (NIH/3T3 ATCC cat. CRL-1658) were treated with 5ug/mL Tunicamycin (R&Dsystems 590507) in DMSO for 5 hours. Cells were then briefly rinsed with 1X DPBS and then scrapped off the plate in fresh DPBS. Cells were pelleted by centrifugation at 300 x g for 10 minutes. Supernatant was then removed, and dry cell pellets were snap frozen on dry ice and then stored at -80C until used. For high-resolution calnexin-BirA 64 staining, mouse embryonic fibroblasts (MEFs) isolated from Sox2-BirA*G3 (n=1) and control (Sox2-Cre) (n=1) mice were cultured on coverslips coated with 0.1% gelatin. Cells at 70% confluence were briefly rinsed in 1X DPBS and then fixed in 4% PFA for 10 minutes followed by 3 washes in 1X DPBS. Coverslips with cells were stored at 4°C in 1X DPBS in the dark until used. Hematoxylin and Eosin staining Tissues (n=3/genotype) were collected and fixed in 10% phosphate-buffered formalin overnight at 4°C. Paraffin sections were prepared using standard procedures. The tissue sections were deparaffinized by immersing in xylene and rehydrated through graded alcohol series, dyed with hematoxylin and eosin (H&E) and then rinsed with water. All slices were dyed with hematoxylin and eosin (H&E) and then rinsed with water. Each slide was dehydrated through graded alcohols. Tissue sections were finally soaked in xylene twice. The sections were examined under a light microscope for evaluation of pathological changes. Frozen Tissue Preparation and Sectioning Briefly, tissues were harvested from DPBS-perfused mice and then fixed in 4% paraformaldehyde for 2 hours at 4°C. Tissues (except brain: fixation overnight and 30% sucrose for 48 hours) were then washed 3 times in 1X DPBS with calcium and magnesium before being incubated in 30% sucrose overnight. The following day, tissues were washed in OCT 3 times to remove excess sucrose and then embedded in OCT (VWR, 25608-930) and frozen in a dry-ice ethanol bath before being stored at -80°C. Tissue blocks were thawed to -20°C overnight and then cryosectioned at 10-16μm at -20°C and placed on glass slides. Slides were then stored at -80°C until immunostained. 65 Immunofluorescent Staining and Confocal Microscopy Frozen sectioned tissues were thawed at room temp for 10 minutes. Slides were then rinsed in 1× DPBS with calcium and magnesium for 10 minutes. Coverslips with cells were removed from 4°C 1X DPBS and treated as slides. Slides were permeabilized in 0.25% Triton-X100 (Sigma X100-500ML) for 5 minutes, then incubated in blocking buffer (2.0% Sea Block (Thermo, 37527) + 0.125% Triton-X100 in 1× DPBS with calcium and magnesium) for 1 hour at room temperature. Slides were then incubated in primary antibody (Supplementary Table 3.1) diluted in blocking buffer, overnight at 4°C. The following day, primary antibodies were removed, and slides were washed in blocking buffer four times for 5 minutes each. Slides were then incubated in secondary antibody diluted in blocking buffer for 1 hour at room temperature. Secondary antibody (Supplementary Table 3.2) was removed, and slides were washed in blocking buffer four times for 5 minutes each. Slides were then incubated in 1 mg/mL Hoechst 33342 (Thermo, H3570) in 1× DPBS with calcium and magnesium for 10 minutes at room temperature. Slides were then washed twice in 1× DPBS with calcium and magnesium for 5 minutes each. Slides were mounted in Immu-Mount (Thermo, 9990402) and imaged at 40× or 63× using the Leica SP8 confocal microscope. Brain images are median intensity projections from z-stacks of 0.5-1.0μm steps except the whole brain tiles scans. All images presented represent at least three images per tissue and 2 biological samples per genotype. Whole tissue section scans were imaged using the Zeiss AxioScan Z1 Slide Scanner at 20× to generate high-resolution tiled images of tissue sections. Hepatocyte Isolation and qPCR We used the classic two-step collagenase perfusion technique to isolate primary 66 mouse hepatocytes. Briefly, The Sox2-BirA*G3 and control livers were perfused by perfusion medium (GIBCO #17701-038) through Inferior vena cava. Then, the livers were dissociated by liver digested medium (GIBCO 17703-034), which is incubated 30-60 min in 37°C before use. mKate2low and mKatehigh hepatocytes were sorted by flow cytometry based on mKate2 levels. Total RNA was extracted using RNeasy mini kit (Qiagen, 74104). cDNA was synthesized using SuperScript™ IV VILO™ Master Mix (Thermo #11766050). qPCR was performed with Luna Universal qPCR Master Mix Protocol (New England Biolab #M3003) on a Roche LightCycler 96 System. Primers used in RT-qPCR are listed as follows: BirA*G3: Forward: CTCCCCGTGGTTGACTCTAC Reverse: CTCCCCGTGGTTGACTCTAC Alb: Forward: GTCTTAGTGAGGTGGAGCATGACAC Reverse: GCAAGTCTCAGCAACAGGGATACAG Gapdh: Forward: CATGGCCTTCCGTGTTCCTA Reverse: CCTGCTTCACCACCTTCTTGAT The delta-delta Ct method, also known as the 2–∆∆Ct method, was used to calculate the relative fold gene expression. RNA Sequencing and Analysis 67 Mouse samples were collected for RNA as described above. After collecting samples were immediately stored in RNALater (Sigma 90901-100ML) at 4°C overnight before being stored at -80°C until processing. Before RNA extraction samples were removed from RNALater and briefly rinsed in RNase free dH2O. Total RNA was then prepared from Sox2-BirA*G3 and control liver, kidney, and whole brain (n=2/genotype, all female) using RNeasy mini kit (Qiagen 74104) according to kit instructions with the following modifications. For whole brain, brains were homogenized in 1400μL RLT buffer. Brain samples were then further diluted 1:25 in fresh RLT buffer (total final volume 350μL) to avoid overloading the columns. Liver and kidney tissue samples were processed exactly following kit instructions. RNA integrity was determined using Agilent TapeStation 4200. Samples were then sequenced by Washington University in St. Louis School of Medicine Genome Technology Access Center (GTAC) using the following methods. Total RNA integrity was determined using Agilent BioAnalyzer. Library preparation was performed with 10ng of total RNA with Bioanalyzer RIN score of greater than 8.0. ds-cDNA was prepared using the SMARTer Ultra Low RNA kit from Illumina Sequencing (Takar-Clonetech) per manufacturer’s protocol. cDNA was fragmented using a Covaris E220 sonicator using peak incident power 18, duty factor 20%, cycles per burst 50 for 120 seconds. cDNA was blunt ended, had A base added to the 3’ ends, and then had Illumina sequencing adapters ligated to the ends. Ligated fragments were then amplified for 12-15 cycles using primers incorporating unique dual index tags. Fragments were sequenced on an Illumina NovaSeq-6000 using paired end reads extending 150 bases. Raw paired end read files were first trimmed using Trimmomatic v0.38 and then aligned to GRCm39 using STAR v2.7.0e with default options. Read counts were quantified using RSubread_2.2.6 (featureCounts) without multimapping. Read counts were then analyzed using DESeq2 v1.28.1 in R (v4.0.0) using 68 standard approaches with cutoffs of log2 fold change > 2.0 and adjusted p-value < 0.05. Protein Lysate Preparation Protein lysates were prepared as described previously 50 with the following modifications. homogenized in 500μL RIPA complete lysis buffer (RIPA buffer (ThermoFisher, 89901) with 1× cOmplete EDTA-free protease inhibitor cocktail (Sigma, 11873580001), 1mM benzamidine hydrochloride (VWR, TCB0013-100G), 4μM pepstatin A (Sigma, EI10), 100μM PMSF (Sigma, 11359061001)) and bead homogenized using stainless steel beads (NextAdvance, SSB14B-RNA) for 5 minutes at setting 10, Bullet Blender Storm (NextAdvance, BT24M). Samples were then centrifuged at 14,000×g for 15 minutes at 4°C. Supernatants were transferred to protein loBind (Eppendorf) tubes. Protein lysate concentrations were determined using Pierce BCA (Thermo cat. 23227) microplate assay per manufacturer’s instructions. Lysates were then stored at -80°C. Streptavidin Bead Pulldowns Streptavidin pulldowns were performed as described previously 50 with modifications. Streptavidin magnetic beads (Thermo cat. 88817) were resuspended in lysis buffer (above) by magnetic separation (BioRad 1614916). We tested a series of volumes of beads and washing conditions and found that 5μL beads per 100μg protein together with the following washing conditions are sufficient (results available upon request). Pulldown reactions were set up in 450μL lysis buffer with 5μL beads per 100μg protein. Pulldowns were then incubated overnight at 4°C in a wheel rotator. The following day, pulldown reactions were washed 2× in lysis buffer, then 1× in 2M Urea in 10mM Tris, and finally 2× lysis buffer. After the final wash, lysis buffer was removed and beads were either boiled in 12μL 1× loading buffer (Li-Cor 928-40004) with 1.43 M β- 69 mercaptoethanol or resuspended in 100μL lysis buffer and flash frozen, for western blotting and mass spectrometry respectively. Silver Stain Analysis Silver staining was done using Richard J. Simpson’s protocol from Cold Spring Harbor (CSH) or using the Silver Stain PlusTM kit (BioRad 1610449). Gels were fixed in a 50% methanol (VWR BDH1135-4LG), 5% glacial acetic acid (VWR 97064-482) solution, gently shaking at room temperature for 20 minutes, Gels were then incubated in 50% methanol for 10 minutes, gently shaking at room temperature, followed by a 10 minute incubation in dH 2 O. Gels were then soaked in 0.02% sodium thiosulfate (Sigma 72049) for 1 minute and then in dH 2 O for 1 minute, twice. Gels were then incubated in chilled 0.1% silver nitrate (Sigma 209139) for 20 minutes, gently shaking at 4C in the dark. Gels were then rinsed twice in dH 2 O for 1 minute each. Gels were developed in a 2% sodium carbonate (Sigma 222321) and 0.04% formaldehyde (Thermo 28906) until desired intensity was reached. Developing was stopped with a 5% glacial acetic solution and gels were stored in 1% glacial acetic acid until being discarded. Fluorescent Western Blot Analysis Western blots were performed with standard protocols and the following modifications. Equal amounts of total protein lysate were loaded per sample per reaction with 1X Li-Cor loading buffer (Li-Cor, 928-40004) with 1.43 M β-mercaptoethanol. For streptavidin pulldowns, beads were resuspended in 12μL 1X Li-Cor loading buffer (Li- Cor, 928-40004) with 1.43 M β-mercaptoethanol. All samples were then boiled at 95°C for 5 minutes to elute, then briefly spun down and kept on ice prior to loading. Total protein samples and pulldown elutes were loaded on 10% SDS acrylamide gels and ran in 70 standard 1× SDS-Running buffer with Li-Cor 5μL one-color molecular marker (Li-Cor, 928-40000) at 60V for 30 minutes, followed by 120V for ~50 minutes or until loading dye ran off. Samples on the gel were transferred to methanol activated PVDF 0.45μm membranes using BioRad’s wet tank mini-protean system for 1-3 hours at 250-300 constant mA in a sample dependent context. After transfer, membranes were dried at 37°C for 5 minutes and then re-activated with methanol. Blots were stained with Li-Cor’s Revert-700 Total Protein Stain (Li-Cor, 926-11010) for normalization and imaged using a Li-Cor Odyssey Clx. Blots were then de-stained per kit instructions and put in block (Li- Cor Intercept block, 927-60001) for 1 hour, room temperature, shaking. Blots were then transferred to primary antibody (Supplementary Table 3.1) (block with 0.2% Tween20) overnight at 4°C, shaking. The following day, blots were washed four times in TBS-T for 5 minutes each at room temperature, shaking, and then incubated in secondary antibody (Supplementary Table 3.2) in block with 0.2% Tween20 and 0.1% SDS, and/or streptavidin conjugate (1:5,000; 680 or 800, Li-Cor, 926-68079, 926-32230) if visualizing biotinylated proteins, for 1 hour at room temperature, shaking. Blots were then washed twice with TBS-T for 5 minutes each, room temperature, shaking, followed by two 5- minute TBS washes at room temperature, shaking. Blots were imaged on a Li-Cor Odyssey Clx using Li-Cor’s ImageStudio (Version 5.2.5). After imaging blots were dried at 37°C for 5 minutes, then stored. For phosphorylated proteins (EIF2α), proteins were blotted for the phosphorylated state as described above. After imaging, phospho-blots were stripped is 10mL 1X Restore Fluorescent Western Blot stripping buffer (Thermo cat. 62300) for 20 minutes at room temperature, shaking. Blots were then briefly rinsed with dH2O twice, re-blocked for 30 minutes at room temperature shaking, before primary incubation with the total protein antibody and secondary as described above. All western blot images were exported from Li-Cor, pseudo-colored and converted to RGB tiffs in 71 ImageJ (v1.51S) for figures. For specific proteins, bands were selected based on molecular weight from antibody manufacturer information and literature. Note that all western blot experiments were repeated at least twice with different biological samples and produced consistent results. Fluorescent Western Blot Quantification Biotinylation levels and proteins of interest were quantified via western blot using Li-Cor’s fluorescent western blot ImageStudio (Version 5.2.5) and Emperia Studio (Version 1.3.0.83) analysis software and protocols. Total protein stain images of each blot were used to normalize biotinylation (streptavidin) or protein of interest signal intensity in RStudio (Version 1.3.959, R Version 4.0.0) by determining the lane normalization factor (Li-Cor protocol) for each blot per manufacturer’s instructions. ggplot2 (Version 3.3.5) and GraphPad Prism 9.0 were used to visualize normalized biotinylated protein signal. Sox2-BirA*G3 Analysis by MS (corresponding to Figures 3-4 and Extended Figures 6-8) After streptavidin beads pulldowns, Sox2-BirA*G3 and control samples were sent to The Broad Institute of Harvard and MIT for MS. i. On-bead Digestion Samples collected and enriched with streptavidin magnetic beads were washed twice with 200 μL of 50mM Tris-HCl buffer (pH 7.5), transferred into new 1.5 mL Eppendorf tubes, and washed 2 more times with 200 μL of 50mM Tris (pH 7.5) buffer. Samples were incubated in 0.4 μg trypsin in 80 μL of 2M urea/50mM Tris buffer with 1 mM DTT, for 1 h at room temperature while shaking at 1000 rpm. Following pre- digestion, 80 μL of each supernatant was transferred into new tubes. Beads were then incubated in 80 uL of the same digestion buffer for 30 min while shaking at 1000rpm. 72 Supernatant was transferred to the tube containing the previous elution. Beads were washed twice with 60 μL of 2M urea/50mM Tris buffer, and these washes were combined with the supernatant. The eluates were spun down at 5000 × g for 1 min and the supernatant was transferred to a new tube. Samples were reduced with 4 mM DTT for 30 min at room temperature, with shaking. Following reduction, samples were alkylated with 10mM iodoacetamide for 45 min in the dark at room temperature. An additional 0.5 μg of trypsin was added and samples were digested overnight at room temperature while shaking at 700 × g. Following overnight digestion, samples were acidified (pH < 3) with neat formic acid (FA), to a final concentration of 1% FA. Samples were spun down and desalted on C18 StageTips as previously described56. Eluted peptides were dried to completion and stored at −80 °C. ii. TMT Labeling of Peptides Desalted peptides were labeled with TMT (6-plex) reagents (ThermoFisher Scientific). Peptides were resuspended in 80 μL of 50 mM HEPES and labeled with 20 uL 20mg/mL TMT6 reagents in ACN. Samples were incubated at RT for 1 h with shaking at 1000 × rpm. TMT reaction was quenched with 4 μL of 5% hydroxylamine at room temperature for 15min with shaking. TMT labeled samples were combined, dried to completion, reconstituted in 100 μL of 0.1% FA, and desalted on StageTips. iii. bRP Stage Tip Fractionation of Peptides 50% of the TMT labeled peptide sample was fractionated by basic reverse phase (bRP) fractionation. StageTips packed with 3 disks of SDB-RPS (Empore) material. StageTips were conditioned with 100 μL of 100% MeOH, followed by 100 μL 50% MeCN/0.1% FA and two washes with 100 μL 0.1% FA. Peptide samples were resuspended in 200 μL 1% FA (pH<3) and loaded onto StageTips. 6 step-wise elutions were carried out in 100 μL 20 mM ammonium formate buffer with increasing 73 concentration of 5%, 10%, 15%, 20%, 25%, and 45% MeCN. Eluted fractions were dried to completion. iv. Liquid Chromatography and Mass Spectrometry iv. Liquid chromatography and mass spectrometry. Single-shot LC-MS/MS analyses were performed on 50% of each sample. The remaining 50% of each sample was fractionated using bRP StageTip fractionation. For single shot and all fractionated samples, desalted peptides were resuspended in 9 μL of 3% MeCN/0.1% FA and 4 μL was injected. For serum samples, an Orbitrap Fusion Lumos Tribrid Mass Spectrometer (ThermoFisher Scientific) was used. For all other plexes, an Orbitrap Exploris 480 (ThermoFisher Scientific) was used. Mass spectrometers were coupled online to a Proxeon Easy-nLC 1200 (ThermoFisher Scientific) as previously described56. Briefly, 4 μL of each sample was loaded at onto a microcapillary column (360 μm outer diameter × 75 μm inner diameter) containing an integrated electrospray emitter tip (10 μm), packed to approximately 24 cm with ReproSil-Pur C18-AQ 1.9 μm beads (Dr. Maisch GmbH) and heated to 50 °C. bRP fractionated samples were analyzed using a 110 min LC–MS. Mobile phase flow rate was 200 nL/min, comprises 3% acetonitrile/0.1% formic acid (Solvent A) and 90% acetonitrile /0.1% formic acid (Solvent B). The 110-min LC–MS/MS method used the following gradient profile: (min:%B) 0:2; 1:6; 85:30; 94:60; 95:90; 100:90; 101:50; 110:50 (the last two steps at 500 nL/min flow rate). Data acquisition was done in the data-dependent mode acquiring HCD MS/MS scans (r = 15,000) after each MS1 scan (r = 60,000) on the top 12 most abundant ions using an MS1 AGC target of 4 x 105 and an MS2 AGC target of 5 × 104. The maximum ion time utilized for MS/MS scans was 120 ms; the HCD-normalized collision energy was set to 36 (Fusion Lumos) or 28 (Exploris 480); the dynamic exclusion time was set to 20 s, and the peptide match and isotope exclusion functions were enabled. Charge exclusion was enabled for 74 charge states that were unassigned, 1 and >7. MS data analysis All protein trafficking MS data were analyzed using Spectrum Mill software package v 7.07 (proteomics.broadinstitute.org)). Similar MS/MS spectra acquired on the same precursor m/z within ±60 s were merged. MS/MS spectra were excluded from searching if they were not within the precursor MH+ range of 600–6000 Da or if they failed the quality filter by not having a sequence tag length >0. MS/MS spectra were searched against a UniProt mouse database with a release date of December 28, 2017 containing 46,519 proteins and 264 common contaminants modified to include GFP, mKate2 and BirA*G3-ER. All spectra were allowed ±20 ppm mass tolerance for precursor and product ions, 40% minimum matched peak intensity, and “trypsin allow P” enzyme specificity with up to 2 missed cleavages. The fixed modifications were carbamidomethylation at cysteine, and TMT6 at N-termini. The variable modifications used were oxidized methionine and N-terminal protein acetylation. Individual spectra were automatically designated as confidently assigned using the Spectrum Mill autovalidation module. Specifically, a target-decoy-based false-discovery rate (FDR) scoring threshold criteria via a two-step auto threshold strategy at the spectral and protein levels was used. First, peptide mode was set to allow automatic variable range precursor mass filtering with score thresholds optimized to yield a spectral level FDR of <1.2%. A protein polishing autovalidation was applied to further filter the peptide spectrum matches using a target protein level FDR threshold of 0. Following autovalidation, a protein–protein comparison table was generated, which contained experimental over control TMT ratios. For all experiments, non-mouse contaminants and reverse hits were removed. Furthermore, the data were median normalized. For serum 75 data, we performed a moderated T-test (limma R package v4.1) to identify proteins significantly enriched in the experimental conditions compared to controls. We corrected for multiple hypotheses (Benjamini–Hochberg procedure). Any protein with an adjusted p-value of less than 0.05 and a log2 fold change greater than 1 was considered statistically enriched. For tissue data, we used the ES method described below (MS hit analysis) to identify enriched proteins. MS Hit Analysis To identify enriched proteins from MS data, we established threshold TMT ratios for hit-calling using positive control (PC) and negative control (NC) protein lists. For the PC list, we used UniProt annotated secreted proteins, while the NC list was the UniProt overlapping list of transcription factors and nuclear proteins, and cytoskeletal genes. Note that the NC list was compared with secreted, receptors, ER proteins, and overlapping genes were removed. Proteins identified by MS were compared to the PC and NC lists and assigned to as being a PC or NC protein. For each experiment (liver, brain, and kidney), there were 9 TMT ratio comparisons: we calculate the TMT ratios of every Sox2-BirA*G3 sample over every control samples (Sox2-BirA*G3-1/ control-1, Sox2-BirA*G3-1/ control-2, Sox2- BirA*G3-1/ control-3, Sox2-BirA*G3-2/ control-1, Sox2-BirA*G3-2/ control-2, Sox2- BirA*G3-2/ control-3, Sox2-BirA*G3-3/ control-1, Sox2-BirA*G3-3/ control-2, and Sox2- BirA*G3-3/ control-3). The false positive rate (FPR) is calculated using the equation below 29 , FPR=(P(TMT ratio| false positive))/(P(TMT ratio| secreted positive) ) The denominator is the conditional probability of finding a known secreted protein in this range, which is calculated as the percentage of proteins on the PC list in 76 this range over all proteins identified on the PC list. The numerator is the conditional probability of finding a false positive protein in a particular TMT ratio range. The result calculated using this equation represents the percentage of false positive proteins in this TMT ratio range over the total false positive proteins identified. We plotted FPR over TMT ratio range (Supplementary Fig. 3.10E; other plots available upon request) and selected the TMT ratio cutoff based on an FPR of 0.1 (Supplementary Fig. 3.10D; other plots available upon request), which means that a protein is 10 times more likely to be a true secreted protein than a false positive. Enrichment score (ES) of a specific protein was defined as the number of TMT ratios that exceeds the TMT ratio cutoff. Thus, proteins with ES of 0 are background, proteins with an ES of 1 are lower confidence hits, and proteins with ES of 9 are highest confidence hits. Alb-BirA*G3 Serum Analysis by MS (corresponding to Figures 5-6 and Extended Figures 10) After streptavidin beads pulldowns, Alb-BirA*G3 and control samples were sent to the UCLA proteomics core, Department of Biological Chemistry, Geffen School of Medicine at UCLA for MS. i. Serum Sample Digestion Streptavidin-bound proteins were reduced and alkylated on bead via sequential 20-minute incubations with 5mM TCEP and 10mM iodoacetamide at room temperature in the dark while being mixed at 1200 rpm in an Eppendorf thermomixer. Proteins were then digested by the addition of 0.1μg Lys-C (FUJIFILM Wako Pure Chemical Corporation, 125-05061) and 0.8μg Trypsin (Thermo Scientific, 90057) while shaking at 37°C overnight. ii. TMT Labeling and CIF Fractionation 77 The supernatant was transferred to new tubes and 8 μl of carboxylate-modified magnetic beads (CMMB, and also widely known as SP3 218 ) was added to each sample. 100% acetonitrile was added to each sample to increase the final acetonitrile concentration to >95% and induce peptide binding to CMMB. CMMB were then washed 3 times with 100% acetonitrile and then resuspended with TMT labeling buffer. 25 ug of each sample was labeled using TMT10 plex Isobaric Labels (Thermo Fisher Scientific) and the resulting 8 labeled samples were pooled. The pooled sample was fractionated by CMMB-based Isopropanol Gradient Peptide Fractionation (CIF) method 219 into 3 fractions before MS analysis. iii. LC-MS Acquisition and Analysis. Fractionated samples were separated on a 75uM ID x 25cm C18 column packed with 1.9μm C18 particles (Dr. Maisch GmbH) using a 140-minute gradient of increasing acetonitrile and eluted directly into a Thermo Orbitrap Fusion Lumos mass spectrometer where MS spectra were acquired using SPS-MS3. Protein identification was performed using MaxQuant62 v 1.6.17.0. The complete Uniprot mouse proteome reference database (UP000000589) was searched for matching MS/MS spectra. Searches were performed using a 20 ppm precursor ion tolerance. TMT10plex was set as a static modification on lysine and peptide N terminal. Carbamidomethylation of cysteine was set as static modification, while oxidation of methionine residues and N-terminal protein acetylation were set as variable modifications. LysC and Trypsin were selected as enzyme specificity with maximum of two missed cleavages allowed. 1% false discovery rate was used as a filter at both protein and PSM levels. Statistical analysis was conducted with the MSstatsTMT Bioconductor package. The abundance of proteins missing from one condition but found in more than 2 78 biological replicates of the other condition for any given comparison were estimated by imputing intensity values from the lowest observed MS1-intensity across samples and p- values were randomly assigned to those between 0.05 and 0.01 for illustration purposes. Data analysis and statistics Data was analyzed using Microsoft Excel, R (version 4.0.0 (2020-04-24), Platform: x86_64-appledarwin17.0 (64-bit); RStudio Version 1.3.959) and Python. For secretion annotations, proteins were annotated based on the subcellular localization data from UniProt and the cellular component data from National Center for Biotechnology Information (NCBI) (https://www.ncbi.nlm.nih.gov). Proteins in fasta formats were uploaded to SignalP5.0 (https://services.healthtech.dtu.dk/service.php?SignalP-5.0) and TMHMM (v 2.0) (https://services.healthtech.dtu.dk/service.php?TMHMM-2.0) for the prediction of SignalP and transmembrane helix, separately. EnhancedVolcano was used to generate volcano plots based on log2FC and p value. We ran TissueEnrich (https://bioconductor.org/packages/TissueEnrich)(32) on a list of proteins to look for enrichment for tissue-specific genes using mouse ENCODE datasets. Gene ontology function annotation was performed on two platforms-DAVID (https://david.ncifcrf.gov) and EnrichGO in clusterProfiler (3.16.1). The top GO terms were visualized with dotplot in ggplot2. PCA was used to study the similarities between samples. The analysis was conducted without filtering any proteins. For ectodomain shedding analysis, we wrote a python program to map all the peptides identified in mass spec for each protein to their corresponding full-length protein. The reference sequences were annotated based on UniProt topology information. Domain information was based on SMART (http://smart.embl-heidelberg.de). 79 Data Availability The original mass spectra and the protein sequence databases used for searches have been deposited in the public proteomics repository MassIVE (http://massive.ucsd.edu) and are accessible at ftp://MSV000088848@massive.ucsd.edu. RNA sequencing data is available under BioProject PRJNA808087 from NCBI SRA at https://www.ncbi.nlm.nih.gov/sra. The following public databases were used: Uniprot (https://www.uniprot.org), mouse (https://www.uniprot.org/proteomes/UP000000589), SignalP 5.0 (https://services.healthtech.dtu.dk/service.php?SignalP-5.0), TMHMM 2.0 (https://services.healthtech.dtu.dk/service.php?TMHMM-2.0). All analysis code is available at https://github.com/asmeyer/A-genetic-model-for-in-vivo-proximity- labeling-of-the-mammalian-secretome.git. Corresponding authors will provide original data upon request. Source data are provided with this paper. RESULTS Activation of BirA*G3 in Sox2-BirA*G3 mice We previously reported a Cre-inducible BirA*G3 cassette, inserted in the Rosa26 (R26) “safe-harbor” locus in mouse embryo stem cells (mESCs) (Fig. 3.1A) 50 . Here, we derived mouse strains, from three independently targeted mESCs containing the Cre- inducible BirA*G3 cassette (A11, B1, and C2). As expected, no differences were observed comparing the three lines consequently; we refer to data as if from a single line and we chose the A11 line for more extensive characterization. The modified Rosa26 locus was designed to drive ubiquitous GFP expression downstream of a CAGGS (β-actin/CMV) regulatory sequence. GFP transcription blocks 80 downstream expression of a BirA*G3 cassette and mKate2 reporter (Fig. 3.1A)(13). BirA*G3 encodes a promiscuously active biotin ligase selected by protein evolution(12) with a signal peptide and ER retention signal, designed to target and retain BirA*G3 within the cell’s secretory pathway (Fig. 3.1B). mKate2 encodes a monomeric, photostable, pH resistant, low toxicity, bright far-red fluorescent protein designed as a cellular indicator of cells producing BirA*G3 and activating biotinylation of secretory pathway proteins on biotin administration (Fig. 3.1B) 180-182 . CRE recombination at loxP sites flanking the GFP cassette enables the tissue and time-dependent expression of BirA*G3 and mKate2 50 . To initially examine BirA*G3 activity body-wide to broadly calibrate the model, we crossed the R26 BirA*G3 mice to the Sox2-Cre strain which results in recombination and predicted activation of BirA*G3 and mKate2 in the epiblast, and thereafter, in all cell types of the conceptus 183 . As a consequence, cells with the non-recombined BirA*G3 allele will be green (GFP+), while those undergoing CRE-mediated recombination in Sox2-Cre; BirA*G3 (abbreviated as Sox2-BirA*G3) mice will be red (mKate2+) and BirA*G3 positive (Fig. 3.1C). Viability and fertility were normal in Sox2-BirA*G3 mice. As predicted, whole- mount images of selected organs from control mice (BirA*G3/+ unless otherwise stated) were GFP+/mKate2-, with variable GFP intensity across organs (Fig. 3.1D), while Sox2- BirA*G3 mice were GFP-/mKate2+, demonstrating excision of the GFP cassette and activation of mKate2 reporter (Fig. 3.1E). High-magnification confocal images showed that BirA*G3 extensively colocalized with the ER resident protein calnexin (Fig. 3.1F; Supplementary Fig. 3.1A), consistent with BirA*G3 localization to the ER. Focusing on the kidney, liver, and brain, we next sought to determine the distribution of BirA*G3 using representative cell markers, highlighting key cell populations. 81 In the Sox2-BirA*G3 kidney, mKate2 signal and BirA*G3 were present throughout the nephron and collecting system highlighted by co-localization with different cell markers (Fig. 3.1G; Supplementary Fig. 3.1B-C; Supplementary Fig. 3.2A-B). In Sox2- BirA*G3 liver, mKate2 signal and BirA*G3 were present in hepatocytes (Albumin+), though the distribution was patchy (Fig. 3.1H) and neighboring cholangiocytes (CK19+) showed much lower levels of reporter and BirA*G3 (Supplementary Fig. 3.1D). In Sox2- BirA*G3 brain, mKate2 signal and BirA*G3 expression was present in neurons (MAP2+ representative cortical neurons), astrocytes (GFAP+), and microglia (IBA1+) (Fig. 3.1I; Supplementary Fig. 3.1E-F; Supplementary Fig. 3.2C). Taken together, the data indicate a broad cell and tissue distribution for BirA*G3 and mKate2, though levels vary significantly depending on the cell type. To compare BirA*G3 and the mKate2 reporter with another CAGGS (β- actin/CMV) driven reporters targeted to a similar position with the same transcriptional orientation in the R26 locus, we crossed Sox2-Cre mice with the widely used R26 TdTomato reporter mouse strain 184 . We observed homogeneous production of TdTomato in Sox2-Cre; TdTomato (Sox2-TdTomato) tissues (Supplementary Fig. 3.5A-B) suggesting that uneven BirA*G3/mKate2 in Sox2-BirA*G3 mice is specific to the BirA*G3 targeted locus. To specifically examine the patchy hepatocyte distribution, mKate2low and mKate2high hepatocytes were sorted from Sox2-BirA*G3 liver and analyzed by qPCR (Supplementary Fig. 3.5C). No BirA*G3 mRNA expression was detected in control liver as expected (Supplementary Fig. 3.5D). In Sox2-BirA*G3 liver, BirA*G3 mRNA was present in both mKate2 low and mKate2 high cells, though BirA*G3 mRNA levels were also lower in mKate2 low cells (Supplementary Fig. 3.5D). In contrast, albumin mRNA expression was comparable in mKate2 low and mKate2 high populations (Supplementary 82 Fig. 3.5E). The lower BirA*G3 mRNA expression was consistent with lower BirA*G3 protein level in mKate2 low hepatocytes (Supplementary Fig. 3.5F-H). Protein biotinylation is still observed in mKate2 low cells, although at lower levels (Supplementary Fig. 3.5F-H), Thus, a mosaic reduction in transcriptional activity of the BirA*G3 locus likely underlies variable levels of mKate2 and BirA*G3 in hepatocytes. To determine whether down-regulation relates to the duration of allele activation, we examined BirA*G3 in Sox2-BirA*G3 and control pups 10 days after birth. Interestingly, at this early time point, BirA*G3 and mKate2 show a relatively homogenous distribution in hepatocytes (Supplementary Fig. 3.6A-B). While the molecular underpinnings for this observation are unclear, most cell/tissue types were not affected and inducible CRE strains can potentially overcome possible selection for allele silencing. Indeed, we observed homogenous BirA*G3 in GFP- hepatocytes of adult CAGGCre-ERTM; BirA*G3 (CAGG-BirA*G3) mice following tamoxifen injection to induce broad, mosaic excision of the GFP cassette in the adult animal (Supplementary Fig. 3.6C-D). Protein Proximity labeling in Sox2-BirA*G3 mice Next, we characterized biotinylation of proteins in Sox2-BirA*G3 and control mice. Mice were fed biotin in chow (2,000 ppm biotin) for 5 days ad libitum. Brain samples were collected and analyzed by western blotting to detect biotinylated proteins through streptavidin-IRDye 800CW binding. Prominent biotin/BirA*G3-dependent labeling was observed for a broad range of protein species (Fig. 3.2A). Additionally, we observed biotin/ BirA*G3 independent-labeling of two prominent proteins in control and experimental brain samples (Fig. 3.2A) which most likely represent cytoplasmic biotin conjugates with pyruvate carboxylase (~130kDa) and methylcrotonyl-CoA carboxylase/propionyl-CoA carboxylase (~75kDa)(14)(Fig. 3.2A; Supplementary Fig. 83 3.3A). Examining a broad range of total protein lysates from a range of organs showed strong evidence for BirA*G3-dependent biotinylation of target tissues (Fig. 3.2B; Supplementary Fig. 3.3B). Notably, each tissue displayed a unique biotinylation pattern (Fig. 3.2B), suggesting diverse secretomes across tissues. Further, analysis of serum detected a robust BirA*G3-dependent biotinylated protein signature (Fig. 3.2C; Supplementary Fig. 3.3C). Comparable serum labeling was obtained through multiday (7 days) labeling with biotin addition to either water (5mM) or chow (2,000ppm) (Fig. 3.2D; Supplementary Fig. 3.3D, E). Subcutaneous injection of biotin (180mM, 100 μL) resulted in significant labeling of serum proteins within 1 hour of injection with an increased recovery of proteins up to 12 hours (Fig. 3.2E; Supplementary Fig. 3.3F). These results highlight the rapid kinetics for secretion of biotinylated proteins though comparison with protein products observed with continuous 7-day labeling in chow demonstrates enhanced labeling of lower molecular weight protein species (Figure 3.2E). The accumulation of biotinylated proteins could also be visualized by streptavidin-dependent immunofluorescence directly in sections of selective tissues. Abundant biotinylated proteins were observed in Sox2-BirA*G3 muscle, heart, brain, compared to their control counterparts (Fig. 3.2F; Supplementary Fig. 3.4A-B), consistent with CRE-dependent biotinylation. For the liver and kidney, endogenous biotin stores result in equivalent strong streptavidin signals in both Sox2-BirA*G3 and control mice (Supplementary Fig. 3.4C-D), even though western analysis shows Sox2-BirA*G3 - dependent labeling of proteins (Fig. 3.2B). To look for detrimental effects of BirA*G3 in Sox2-BirA*G3 mice, we performed hematoxylin and eosin staining and bulk mRNA-sequencing. No pathological changes were detected by histology comparing Sox2-BirA*G3 and control mice fed biotin chow for 84 7 days (Supplementary Fig. 3.7A). Principal component analysis (PCA) of mRNA- sequencing comparing liver, brain, and kidney between control (n=2 per tissue) and Sox2- BirA*G3 (n=2 per tissue) mice showed a tight clustering by tissue (Supplementary Fig. 3.7B). Differential gene expression analysis of all Sox2-BirA*G3 samples compared to all control samples showed a total of only four differentially expressed genes (DEGs) (Supplementary Fig. 3.7C). Furthermore, analysis of ER stress, unfolded protein response (UPR), and cell death markers showed no difference (adj. p-value > 0.05) at the transcript level between Sox2-BirA*G3 and control samples (Supplementary Fig. 3.7D). Additionally, western blotting showed no indication of ER stress or UPR in Sox2-BirA*G3 and control livers compared to ER stress induced (tunicamycin treated) MEFs (Supplementary Fig. 3.7E-J). MS-based tissue proteomics of biotinylated proteins As a prelude to MS analyses of each of the three tissues (liver, brain, and kidney), affinity purified biotinylated proteins were visualized by both streptavidin western analysis and silver stain. The biotinylation banding patterns were comparable to those observed in total protein lysates of corresponding samples, suggesting efficient and unbiased enrichment of biotinylated proteins Supplementary Fig. 3.8A-D). Silver stain showed specific bands in Sox2-BirA*G3 samples, despite a strong background of non- specific binding of unlabeled proteins to streptavidin conjugated beads in all three tissues (Supplementary Fig. 3.8E-H). The biotinylated proteomes of the three tissues were defined by quantitative TMT- based LC-MS/MS following streptavidin bead enrichment from Sox2-BirA*G3 and control liver, brain, and kidney. Of the thousands of proteins detected and quantified in the Sox2-BirA*G3 tissues, several hundred proteins were found to be significantly 85 enriched in liver (n=189), brain (n=200), kidney (n=578) compared to their control counterparts (log2 fold change (FC)>1.0 and adj. p-value<0.05) (Fig. 3.3A-B; Supplementary Fig. 3.9A-C; Supplementary Data 3.1-3.3). Principal component analysis (PCA) demonstrated tight grouping of biological replicates within the Sox2-BirA*G3 group, indicating similar proteomic profiles among these samples (Supplementary Fig. 3.10A-C). Relative abundance of representative signature proteins for each tissue is shown in in Fig. 3.3C. An enrichment score (ES) was generated as a measure of the abundance of proteins in Sox2-BirA*G3 replicate samples compared to control samples, as previously described 29,50 . Briefly, for 6plex TMT ratios of each tissue, TMT ratios were calculated by comparing each of the three Sox2-BirA*G3 over each of the three control samples, giving rise to nine different datasets of TMT ratios. Then, we determined the false positive rate (FPR) based on proteins that are retained or traffic through the ER (positive control) and proteins that do not (negative control) (Supplementary Fig. 3.10D-E; see MS Hit Analysis in Methods). The number of TMT ratios for each protein that pass the TMT ratio cutoff based on an FPR of 0.1 was defined as enrichment score (ES) (Supplementary Fig. 3.10E). A protein where all its ratios pass the cutoffs has an enrichment score (ES) of 9 (highest confidence), where a protein where none of its ratios pass the cutoffs has a score of 0 (lowest confidence). Assigning an ES>=5 as the cutoff (Supplementary Fig. 3.10F) maximized the recovery of ER-targeted proteins with high specificity (Supplementary Fig. 3.10G-I). In the conventional secretory pathway, proteins with either a signal peptide (SignalP) or transmembrane helix (TMH) travel through ER-Golgi apparatus. We used SignalP (v 5.0) 129,185,186 and TMHMM (v 2.0) 128,187,188 to predict the presence of SignalP and TMH on each tissue sample. Our analysis revealed that higher ES correlated with higher 86 ratios of proteins with predicted SignalP/TMH (Supplementary Fig. 3.9D-F), suggesting that proteins with higher ES contain “hits” of higher confidence. Comparing across liver, brain, and kidney, 113 proteins were present in all three organ samples (Fig. 3.3D) and over 93% of these (106) showed a SignalP and/or TMH (Fig. 3.3E). Applying Gene Ontology (GO) functional category enrichment analysis to identify characteristic cellular component attributes of these tissue-shared proteins highlighted extracellular exosome and ER-Golgi terms in DAVID analysis (Fig. 3.3E; Supplementary Fig. 3.9G), consistent with ER localization of BirA*G3 and the expected labeling of proteins in the secretory pathway. These proteins are enriched in ER-Golgi related functions (Supplementary Fig. 3.9H). TissueEnrich 189 , a tool for tissue-specific gene enrichment, was applied to liver, brain, and kidney-specific proteins (Supplementary Fig. 3.11A-C). Notably, all organ samples showed a strong enrichment profile for the expected organ type. Significant enrichment of SignalP/TMH was also seen in tissue-specific hits (Fig. 3.3F; Supplementary Fig. 3.11D, E). For biological process analysis, significantly enriched terms in liver showed unique features of liver function, which include lipid catabolic process and blood coagulation (Fig. 3.3F). Similarly, significantly enriched terms in brain showed unique features of brain function, such as synapse organization and assembly (Supplementary Fig. 3.11F). However, kidney-specific functional annotation terms were not observed amongst top terms in the kidney data (Supplementary Fig. 3.11G). The large majority of kidney enriched transporter and channel proteins are confined to short segments within the renal epithelium, organ-wide enrichment is likely to select for more secretory pathway proteins that are broadly distributed, or particularly abundant proteins such as UMOD, that are segmentally restricted (Fig. 3.3C; Supplementary Fig. 3.11E, G). 87 Analysis of the serum secretomes of labeled tissues To examine biotinylated proteins secreted into the serum of female Sox2-BirA*G3 mice, serum from labeled mice was enriched using streptavidin-conjugated beads and bound fractions from secretomes of Sox2-BirA*G3 (n=3) mice and associated controls (n=3) were analyzed by MS (6plex) (Fig. 3.4A; Supplementary Fig. 3.13A; Supplementary Data 3.4-3.5). PCA demonstrated tight clustering of the Sox2-BirA*G3 group (Supplementary Fig. 3.12A-B). The affinity purified serum proteome identified by TMT- based MS has minimal background compared with tissue samples (Supplementary Fig. 3.13B). Accordingly, we used a log2 FC>1.0 and adj. p-value<0.05 to define BirA*G3 serum enriched proteins over control serum to score enriched serum proteins from two independent MS runs of the same samples processed by two different research labs at different universities and run by the same MS group (plex1 and plex2). The majority of serum proteins (70.6%) were identified in both MS runs (plexes) (Supplementary Fig. 3.12C), highlighting the reproducibility of this method. Most serum enriched proteins are annotated to contain a SignalP or TMH (64.8% of total and 76.6% of proteins shared between two plexes; Supplementary Fig. 3.13C) accounting for enriched cellular component terms for extracellular exosome/space/region (Supplementary Fig. 3.12D-E). As expected, biological process annotation for serum enriched proteins predicted with SignalP/TMH are enriched in immunity and complement-related terms (Supplementary Fig. 3.12F). To identify potential proteins secreted from tissues to serum, we compared the enriched proteins from serum with those from the three tissues and found 220 total overlapping proteins between serum and i) liver, ii) brain, or iii) kidney (Fig. 3.4B). A list of selective serum enriched proteins was shown in a heat map representing log2FC between Sox2-BirA*G3 and control sample, including classic 88 hepatocyte secreted proteins, such as Angiotensinogen (AGT) and Albumin (ALB), brain cell marker, neuron neuropeptide Oxytocin (OXT), as well as kidney markers, uromodulin (UMOD) and insulin-like growth factor-binding protein-3 (IGFBP3) (Fig. 3.4C). Out of the 220 tissue-serum shared proteins, 65.9% were annotated as secreted by UniProt/NCBI (Fig. 3.4D), and the majority (85.9%) were predicted to have a SignalP or TMH (Fig. 3.4E). These tissue-serum shared proteins were then analyzed for GO term enrichment analysis (Fig. 3.4F; Supplementary Fig. 3.13D-E). DAVID analysis demonstrated significant enrichment in ‘extracellular’ annotations in the cellular component annotation (Fig. 3.4F). The biological process GO term for these tissue-shared proteins with predicted SignalP/TMH are enriched in coagulation and hemostasis (Supplementary Fig. 3.13E). Ectodomain shedding is an important post-translational mechanism for regulating the function of cell surface proteins, which involves the proteolytic cleavage of transmembrane (TM) cell surface proteins, and release of circulating, soluble form 190,191 . Five well-documented TM proteins (LIFR, EGFR, VCAM1, Slc38a10 and PIGR) appeared in our serum proteomic datasets. For each, sequenced peptides exclusively mapped to the annotated extracellular domains, consistent with ectodomain shedding (Fig. 3.4G; Supplementary Fig. 3.13G-I). While ectodomain shedding has been reported for LIF receptor subunit α (LIFR) 192 , polymeric immunoglobulin receptor (PIGR) 193 , epidermal growth factor receptor (EGFR) 194 , and vascular cell adhesion molecule 1 (VCAM1) 195 , solute carrier family 38 member 10 (Slc38a10) have not been directly linked to shedding, though an extramembrane fragment of human Slc38a10 has been reported in plasma 196 (Fig. 3.4G-H; Supplementary Fig. 3.13F-H). Approximate protein cleavage sites can be assigned on the basis of the most N- or C-terminal peptides (Fig. 3.4G-H; Supplementary 89 Fig. 3.13F-H). For example, the in vivo cleavage site for LIFR is predicted to be between the fifth fibronectin type 3 (FN3) domain and the TM domain, similar to a previous report 49 . Intriguingly, we also detected thioredoxin (TXN) and Annexin1 (ANXA1) in our dataset (Fig. 3.4C), each of which are reported to be released from the cell through unconventional protein secretion (UPS) 197-199 , suggesting this method can give further insight into UPS, with potentially other candidates among the detected proteins lacking a predicted SignalP or TMH domain. To confirm key hits in these secreted serum protein data, we performed western blot analysis of 11 proteins observed in the MS analyses of biotinylated serum proteins (Supplementary Fig. 3.13I, J). These validation analyses were carried out in both male and female mice (Supplementary Fig. 13J). Consistent with the MS data, all 11 proteins were enriched in Sox2-BirA*G3 serum after affinity purification. Biotin labeling of the hepatocyte secretome We next applied and extended the approach to determine the feasibility of identifying a tissue cell type specific secretome. For this, we utilized Alb-Cre to enable hepatocyte-specific biotinylation of proteins in the liver 200 . Biotin chow was given to adult Alb-Cre; BirA*G3 (abbreviated as Alb-BirA*G3) and control mice for 5 days, and then well- perfused tissues were collected after exsanguination (Supplementary Fig. 3.14A). mKate2 was specifically detected in the Alb-BirA*G3 liver (Fig. 3.5A; Supplementary Fig. 3.15A; Supplementary Fig. 3.14B). BirA*G3 showed a homogenous distribution restricted to the hepatocytes of Alb-BirA*G3 liver (Fig. 3.5B). BirA*G3 activity in Alb-BirA*G3 liver enabled specific protein biotinylation in liver protein lysates to be compared with control mice (Fig. 3.5C). Alb-BirA*G3 serum samples showed strong, specific protein biotinylation relative to serum from biotin administered controls lacking active BirA*G3 90 (Fig. 3.5D). The biotinylation banding patterns of affinity purified biotinylated proteins were comparable to those observed in total protein lysates of liver and serum samples (Fig. 3.5C-D; Supplementary Fig. 3.15B-F). Silver staining of affinity purified biotinylated proteins showed specific bands in Alb-BirA*G3 serum, with similar size distributions between sexes (Fig. 3.5E). Identification of hepatocyte secretome Streptavidin purified serum proteins of Alb-BirA*G3 (n=4) and control (n=4) mice were analyzed by LC-MS/MS (8plex) to identify secreted hepatocyte proteins (Supplementary Data 3.6). PCA demonstrated male (n=2) and female (n=2) Alb-BirA*G3 samples cluster together, indicating similar hepatocyte secretome between the two sexes (Fig. 3.5F). Approximately 80% of proteins (181/189) were specifically enriched (log2FC>1.0 and adj. p-value<0.05) in the Alb-BirA*G3 group (Fig. 3.5G). A heatmap showed very low protein abundance in the control group, consistent with the absence of biotinylated proteins in the control serum samples (Fig. 3.6A; Fig. 3.5E). To determine cell labeling specificity, TissueEnrich on serum enriched proteins in the Alb-BirA*G3 group showed a highly liver-specific organ profile using the mouse ENCODE dataset (Fig. 3.6B). Over half (69.1%) of the proteins enriched in Alb-BirA*G3 serum were annotated as secreted and a majority (77.9%) of these proteins were predicted with SignalP/TMH (Fig. 3.6C, D). Functional annotation showed significant enrichment for ‘extracellular’ terms (Supplementary Fig. 3.16A-B). Biological process annotation for the subset of serum enriched proteins with predicted SignalP or TMH, specific to the Alb-BirA*G3 group demonstrated enriched terms for key liver protein functions: coagulation and hemostasis (Fig. 3.6E) 198 . Recently, other groups have used adeno-associated virus carrying ER-TurboID 91 under the control of the hepatocyte-selective Tbg (thyroxine binding protein) promoter (AAV-Tbg-ER-TurboID) 49 and adenovirus carrying Sec61b (Protein transport protein Sec61 subunit beta)-TurboID 48 to label the liver secretome of mice. While these two previous datasets only share 5 proteins, both displayed a significant overlap (23/64 and 25/27, respectively) with our Alb-BirA*G3 dataset, highlighting the power of the transgenic strain (Fig. 3.6F). Further, the 137 proteins uniquely detected in the Alb- BirA*G3 dataset displayed a strong correlation with liver-specific proteins and the majority (67.2%) was as predicted with SignalP/TMH (Supplementary Fig. 3.16C, D). Well-characterized hepatocyte secreted proteins, including APOE, AGT, IGFBP2, were only seen in the Alb-BirA*G3 unique dataset, highlighting an improved coverage of secreted liver proteins using this strategy. Finally, we asked if secreted proteins can be labeled in the organ of origin and subsequently detected in a destination organ after affinity purification. The renin- angiotensin-aldosterone system regulates blood pressure in conjunction with the liver, kidney, and lung 201 . In this pathway, angiotensinogen (AGT) is produced by the liver and sequentially converted by kidney and lung enzymes to vasoconstrictive angiotensin II (AGTII). Liver released AGT is reabsorbed from the renal filtrate by proximal tubule cells in the kidney 202 . Consistent with this view, AGT was detected in the liver, serum, and kidney total protein from both Alb-BirA*G3 and control mice (Fig. 3.6G-I), but following streptavidin affinity purification, AGT was only detected in the liver, serum, and kidney from Alb-BirA*G3 mice, but not in control mice (Fig. 3.6G-I). Together, these data support the trafficking of AGT produced in liver through the blood stream with uptake in the kidney. 92 DISCUSSION The genetic system presented here allows for rapid (within 1 hour) and broad (representative cell types) in vivo biotinylation of proteins within the secretory pathway of the living mouse. Long term studies of BirA*G3 production show no obvious pathology. Although we do observe cell-type variability in BirA*G3 levels, that may be countered using tissue or time-dependent CRE activator lines as we show here comparing Sox2-BirA*G3 with CAGG-BirA*G3. TMT-based LC-MS/MS captured a number of well-characterized secreted proteins in Sox2-BirA*G3 serum, including proteins with hormone like properties, known to circulate at µg/mL (ADIPOQ 203 ) and ng/mL (ANGPTL 204 , MST1 205 , MSTN 206 , RETN 203 , CXCL7 207 , IGF1 208 , FGF15 209 ) levels. Interestingly, these represent examples of proteins derived from multiple tissues: adipose tissue (ADIPOQ, Adipsin 210 , and RETN), muscle (MSTN), immune cells (CXCL7), liver (IGF1), and intestine (FGF15). Furthermore, the approach provided insight into ectodomain shedding and UPS. Three research groups have reported various applications of proximity ligation techniques to profile the mammalian secretome. Two groups 48,49 profiled the mouse hepatocyte secretome via viral delivery of TurboID. Viral vectors however present unavoidable problems, such as viral tropism, which limits the use in some tissues. For instance, kidney has been quite difficult to transduce with any viral vector currently available 211 . For brain, viruses need to be intracranially administered due to the blood- brain barrier, for higher transduction efficiency and brain-specific targeting and viral vectors have broad tropisms for neural and glial cell types 39,179 . Directly comparing the Alb-BirA*G3 mouse strain to the two viral studies profiling the hepatocyte secretome, our study showed an improved coverage for relevant hepatocyte proteins which may reflect 93 the stable and efficient expression of BirA*G3, as well as other differences in the downstream analytical pipeline. A third group 10 reported a transgenic mouse expressing BioID, a less efficient biotin ligase (as shown in various studies 6,10,212 ), in the endoplasmic reticulum. Characterization of the endothelial and muscle secretome showed few secreted proteins, likely due to the low activity of BioID2 in ER lumen 6 . Furthermore, the kinetics of labeling are reported to be much faster with BirA*G3, suggesting the line reported here will be better suited to dynamic labeling studies 6 . Incomplete retention of TurboID-KDEL in the endoplasmic reticulum has been documented 48 . In our MS data, we identified tryptic peptides for BirA*G3 in both Sox2- BirA*G3 and Alb-BirA*G3 serum. However, using western blotting, we did not detect BirA*G3 in total serum proteins or affinity purified proteins in Sox2-BirA*G3 serum (Fig. 3.2C; Supplementary Fig. 3.8D). Biotinylation, with biotin ligase, is an ATP-dependent reaction, and ATP is normally restricted within the cell 213 . Biotin concentration within the cell and the concentration of proteins in the secretory pathway, likely also impact the efficiency of the enzymatic reaction. Together, these kinetic arguments make it unlikely there is a significant impact from low levels of circulating BirA*G3 on biotinylation of serum proteins. This is supported by the Alb-BirA*G3 data which demonstrated a robust biotin profile for specific liver-derived proteins with no labeling of abundant proteins secreted by other organs, identified in the Sox2-BirA*G3 study. Recent studies have found that many cytosolic proteins lacking a SignalP/TMH (leaderless cargoes) are released through a type III UPS mechanism 197 . ANXA1 is one such protein detected in our serum dataset (Fig. 3.4C) 197 . UPS cargoes like ANXA1 could be translocated into the ER-Golgi intermediate compartment (ERGIC). Considering the BirA*G3-ER may transiently appear in the ERGIC, it explains the detection of UPS 94 cargoes such as ANXA1. Recent studies show that approximately 13% of the human protein-coding genes encode for roughly 2640 secreted proteins, but relatively few have been functionally annotated and characterized 172 . Since many protein functions are conserved within mammals, we anticipate our platform will serve as a valuable resource for deciphering the mammalian secretome, under healthy or diseased conditions. MAIN FIGURES AND TABLES 95 Fig. 3.1. Generation and characterization of Sox2-BirA*G3 mice. Schematic diagram shows CRE-mediated excision at loxP sites removes the GFP cassette resulting in 96 production of BirA*G3-ER and mKate2 fluorescent protein. Adapted from Droujinine et al. 2021 (A). Schematic diagram shows that proteins that reside or travel through ER would be biotinylated by BirA*G3 (B). Schematic diagram of mouse mating to generate Sox2-BirA*G3 mice. All cells in control mice are expected to express GFP. All cells in Sox2-BirA*G3 mice are expected to express mKate2. Adapted from Droujinine et al. 2021 (C). Native GFP and native mKate2 fluorescence in whole-mount organs (Scale Bar: 2mm) and tissue sections (Scale Bar: 50μm) of control mice. S.I: small intestine. L.I: large intestine (D). Native GFP and native mKate2 fluorescence in whole-mount organs (Scale Bar: 2mm) and tissue sections (Scale Bar: 50μm) of Sox2-BirA*G3 mice (E). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and ER marker Calnexin in the cell culture of isolated mouse embryonic fibroblast (MEFs) from Sox2-BirA*G3 and control mice. Scale Bar: 25μm (F). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and UMOD, which marks the thick ascending limb of Henle’s loop (TALH), in the kidney sections of Sox2-BirA*G3 and control mice. Scale Bar: 50μm (G). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and Albumin, a hepatocyte marker, in the liver sections of Sox2-BirA*G3 and control mice. Scale Bar: 50μm (H). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and MAP2, a cortical neuron-specific protein, in the brain sections of Sox2-BirA*G3 and control mice. Scale Bar: 50μm. Unlike BirA*G3, mKate2 expression is not restricted to the ER, which may result in varied signal intensity due to differences in cell morphology (I). 97 Fig. 3.2. Analysis of biotinylated proteins in Sox2-BirA*G3 and control mice. Western blotting of protein lysates from brain in Sox2-BirA*G3 mice (CRE+) compared to control mice (CRE-) with or without biotin chow administration for 5 days. Upper: Streptavidin labeling. Lower: BirA*G3 (~35kDa). Each lane is a biological replicate from individual mice (n=2/genotype) (A). Western blotting of protein lysates from selective tissues in Sox2-BirA*G3 mice compared to control mice. Note: due to varied streptavidin intensity by tissue, brain, heart, and muscle streptavidin westerns are shown with increased signal intensity. Upper: Streptavidin labeling. Lower: BirA*G3 (~35kDa). Each lane is a biological replicate from individual mice (n=1/genotype) (B). Western blotting of serum total protein in Sox2-BirA*G3 mice compared to control mice (CRE-) with or without biotin chow administration for 5 days. Upper: Streptavidin labeling. Lower: BirA*G3 (~35kDa). Each lane is a biological replicate from individual mice (n=2/genotype) (C). Streptavidin labeling of biotinylated proteins in total serum from Sox2-BirA*G3 and control mice administered with regular chow, biotin chow, biotin water, or biotin chow and water for 7 days. Each lane is a biological replicate from individual mice (n=2/biotin treatment) (D). Streptavidin labeling of affinity purified biotinylated proteins in serum from Sox2-BirA*G3 and control mice given biotin chow for 7 days (left) or given biotin by subcutaneous injection and water (5mM, pH 7.4) after the injection until collection (right) (E). Immunofluorescence images of native GFP, native mKate2, BirA*G3 staining, streptavidin staining in cryo-sectioned muscle tissues from Sox2-BirA*G3 and control mice. Scale Bar: 50μm (F). 98 Fig. 3.3. Identification of biotinylated proteins in Sox2-BirA*G3 liver, brain, and kidney tissues by mass 99 spectrometry. Representative schematic of TMT-based 6plex LC-MS/MS workflow for liver Sox2-BirA*G3 (n=3) and control (n=3) samples. The same 6plex LC-MS/MS design was used for brain and kidney for individual MS runs (A). Volcano plots of proteins detected in liver of Sox2-BirA*G3 mice compared to control mice after streptavidin pulldown. Log 2 FC were plotted on the x-axis and -10log 10 (p value) were plotted on the y-axis. Significantly enriched proteins (adj. p-value< 0.05 and log 2FC>1.0) in Sox2-BirA*G3 A mice compared to control mice are shown in green (B). Relative abundance (log 2FC) of representative proteins in liver, brain, and kidney. ES are labeled for the abundant proteins in their corresponding tissues (C). Venn diagram showed the overlap of enriched proteins (ES method) among liver, brain, and kidney (D). Shared enriched proteins among three tissues (113 proteins) were predicted with SignalP/TMH (left) and analyzed with DAVID analysis for cellular components annotation (right). Gene ratio indicates the percentage of genes annotated with the term over the total number of genes in the list (E). Liver-specific enriched proteins (115 proteins) predicted with SignalP/TMH were analyzed with clusterProfiler (3.16.1) EnrichGO analysis. Left: a pie chart displayed the distribution of liver-specific proteins with SignalP/TMH prediction. Right: dot plots displayed the functional categorization of liver-specific enriched based on EnrichGO annotation, and the number of each category is displayed based on biological process. Gene ratio indicates the percentage of genes annotated with the term over the total number of genes in the list (F). 100 Fig. 3.4. Analysis of biotinylated proteins secreted to peripheral blood. Schematic of tissue secreted proteins identified by LC-MS/MS (6plex; Sox2-BirA*G3 n=3, control n=3) from 101 affinity purified serum from Sox2-BirA*G3 mice (A). Upset plot showed the overlap of enriched proteins among serum, liver, brain, and kidney (B). Heatmap of representative proteins enriched in Sox2-BirA*G3 serum relative to controls. Log 2 expression values are shown by color and intensity of shading. Grey, low; red, high (C). Pie chart displayed the number of shared enriched proteins between serum and three tissues (220 proteins) that are annotated as secreted by UniProt/NCBI (D). Pie chart displayed the distribution of shared enriched proteins between serum and three tissues (220 proteins) with predicted SignalP/TMH (E). Shared enriched proteins between serum and three tissues were analyzed with DAVID analysis for cellular component annotation. Gene ratio indicates the percentage of genes annotated with the term over the total number of genes in the list (F). Schematic of detected peptides for LIFR mapped onto its respective reference sequences with SMART protein domain annotation. Reference sequence is annotated with extracellular, transmembrane (TM) and cytoplasmic based on UniProt topology information. Amino acid sequences of the most C-terminal peptide are labeled. FN3: fibronectin type 3 (G). Schematic of detected peptides for Slc38a10 mapped onto its respective reference sequences. Reference sequence is annotated with TM based on UniProt topology information. Amino acid sequences of the most N-terminal peptide are labeled (H). 102 Fig. 3.5. Generation and characterization of Alb-BirA*G3 mice. Whole mount images of native GFP and mKate2 fluorescence in livers of Alb-BirA*G3 and control mice (Scale Bar: 3mm) (A). Immunofluorescence staining shows expression of BirA*G3 and hepatocyte marker Albumin in the liver sections from Alb-BirA*G3 and control mice. Scale Bar: 50μm (B). Western blotting of total protein from liver (C) and serum (D) in Alb-BirA*G3 mice compared to control mice. Upper: Streptavidin labeling. Lower: BirA*G3 (~35kDa). Each lane is a biological replicate from individual mice (n=3/genotype) (C-D). Silver stain of affinity purified biotinylated proteins from serum in Alb-BirA*G3 mice compared to control mice. Each lane is a biological replicate from individual mice (n=4/genotype) (E). PCA of streptavidin-purified serum proteins from Alb-BirA*G3 and control mice. Each dot represents a sample, which is colored by the annotation of mouse genotype and sex. Alb-BirA*G3 samples are manually circled by grey shadow (F). Volcano plots of proteins detected in serum of Alb-BirA*G3 mice compared to control mice after streptavidin pulldown in mass spectrometry. Log 2 FC were plotted on the x-axis and -10log 10 (p value) were plotted on the y-axis. Significantly enriched proteins (adj. p-value< 0.05 and log 2FC>1) in Alb-BirA*G3 mice compared to control mice are shown in red (G). 103 Figure 3.6. Analysis of hepatocyte-secreted serum proteins in Alb-BirA*G3 mice. Cluster heatmap of proteins enriched in Alb-BirA*G3 serum relative to controls. Log 2 expression values are shown by color and intensity of shading. Blue, low; red, high. Serum enriched proteins in the Alb-BirA*G3 104 group are highlighted in black color in the hierarchical clustering (A). Enriched proteins in Alb-BirA*G3 serum were analyzed with TissueEnrich to calculate tissue-specific gene enrichment (B). Pie chart displayed the number of enriched proteins in Alb-BirA*G3 serum that are annotated as secreted by UniProt/NCBI (C). Pie chart displayed the distribution of enriched proteins in Alb-BirA*G3 serum with predicted SignalP/TMH (D). Enriched proteins in Alb-BirA*G3 serum predicted with SignalP/TMH were analyzed with clusterProfiler (3.16.1) EnrichGO analysis for biological process. Gene ratio indicates the percentage of genes annotated with the term over the total number of genes in the list (E). Area-proportional venn diagram showed the overlap among enriched proteins in Alb-BirA*G3 serum, AAV-Tgb-ER-TurboID plasma 15 , and Sec61-TurboID plasma 14 (F). AGT levels in total protein lysates (upper) and affinity purified biotinylated proteins (lower) from liver (F), serum (G), and kidney (H) in Alb-BirA*G3 mice compared to control mice. Input: 500μg proteins. Bead lanes are affinity purification negative control without protein input. Each lane is a biological replicate from individual mice (n=3/genotype) (G-I). 105 Supplementary Fig. 3.1. BirA*G3 expression and colocalization with ER marker Calnexin in Sox2- 106 BirA*G3 mice. Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and ER marker Calnexin in Sox2-BirA*G3 and control mouse kidney. Scale Bar: 50μm (A). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, Nitric Oxide Synthase 1 (NOS1), Wilms' tumor 1 (WT1) and Renin in the kidney sections of Sox2-BirA*G3 and control mice. Scale Bar: 50μm (B). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and Solute Carrier Family 12 Member 3 (Slc12a3), which marks distal convoluted tubule, in the kidney sections of Sox2-BirA*G3 and control mice. Scale Bar: 50μm (C). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and cytokeratin (CK19), a marker for bile duct, in the liver sections of Sox2-BirA*G3 and control mice. Scale Bar: 50μm (D). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and GFAP, a marker for glia, in the brain sections of Sox2-BirA*G3 and control mice. Cells indicated by the yellow arrow are magnified as in the rectangles. Scale Bar: 50μm (E). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and IBA-1, a marker for microglia, in the brain sections of Sox2-BirA*G3 and control mice. Cells indicated by the yellow arrow are magnified as in the rectangles. Scale Bar: 25μm (F). 107 Supplemental Fig. 3.2. Native GFP and mKate2 expression in kidney and brain sections. Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and α-SMA, a smooth muscle marker, in the kidney sections of Sox2-BirA*G3 and control mice. Scale Bar: 50μm (A). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and aquaporin 2 (AQP2), a marker for collecting duct principal cells, in the kidney sections of Sox2-BirA*G3 and control mice. Scale Bar: 20μm (B). Tile scan images showed the expression of native GFP, native mKate2 and MAP2 in the brain sections of Sox2-BirA*G3 and control mice. Scale Bar: 3mm (C). 108 Supplementary Fig. 3.3. Total protein stains used for western blot normalization and quantification. Total protein stain used for western blot normalization and quantification of total brain protein lysates from Sox2-BirA*G3 (Cre+) and control (Cre-) mice with or without biotin chow administration for 5 days in Fig 3.2A (A). Total protein stain of protein lysates from selective tissues in Sox2-BirA*G3 mice compared to control mice shows equal loading by tissue in Fig 3.2D (B). Total protein stain used for western blot normalization and quantification of total serum from Sox2-BirA*G3 (Cre+) and control (Cre-) mice administered with regular chow, biotin chow, biotin water, or biotin chow and water for 7 days of Fig 3.2B (C). Total protein stain of affinity purified biotinylated proteins in serum from Sox2-BirA*G3 (Cre+) and control (Cre-) mice given biotin chow for 7 days (left) or given biotin by subcutaneous injection and water (5μM, pH 7.4) after the injection until collection (right) (D). Quantification of biotinylation levels normalized to total protein in total serum from Sox2-BirA*G3 (Cre+) and control (Cre-) mice administered with regular chow, biotin chow, biotin water, or biotin chow and water for 7 days of Fig 3.2B (E). Total protein stain of affinity purified biotinylated proteins in serum from Sox2-BirA*G3 and control mice given biotin chow for 7 days (left) or given biotin by subcutaneous injection and water (5mM, pH 7.4) after the injection until collection (right) (F). 109 Supplementary Fig. 3.4. Detection of biotin and biotinylated proteins in Sox2-BirA*G3 and control tissues. Immunofluorescence images of native GFP, native mKate2, BirA*G3 staining, streptavidin staining in cryo- sectioned heart (A), brain (B), liver (C), and kidney (D) tissues from Sox2-BirA*G3 and control mice. Scale Bar: 50μm (A-D). 110 Supplementary Fig. 3.5. Partial BirA silencing observed in Sox2-BirA*G3 mice was affirmed by comparison with Sox2-Cre; TdTomato mice. Representative high magnification images of native fluorescence expression in Sox2-BirA*G3 kidney, livers, and pancreas. Scale Bar: 25μm (A). Representative high magnification images of native fluorescence expression in Sox2-TdTomato kidney, liver, and pancreas. Scale Bar: 25μm (B). mKate2 low and mKate2 high hepatocytes were sorted from Sox2-BirA*G3 liver by flow cytometry based on native GFP and native mKate2 expression (C). BirA*G3 RNA expression was analyzed by Q-PCR in mKate2 low and mKate2 high hepatocytes of Sox2-BirA*G3 liver compared to total control liver. One-way ANOVA with Tukey’s multiple comparisons test were used to test for significant differences between individual groups (D). Albumin RNA expression was analyzed by Q-PCR in mKate2 low and mKate2 high hepatocytes of Sox2-BirA*G3 liver compared to total control liver. One-way ANOVA with Tukey’s multiple comparisons test were used to test for significant differences between individual groups (E). Total protein stain of total protein lysates from mKate2 low and mKate2 high hepatocytes of Sox2-BirA*G3 liver compared to total control liver. Each lane is a biological replicate from individual mice (n=2/genotype) (F). Western blotting of total protein lysates from mKate2 low and mKate2 high hepatocytes of Sox2-BirA*G3 liver compared to total control liver. Upper: Streptavidin labeling. Lower: BirA*G3 (~35kDa). Each lane is a biological replicate from individual mice (n=2/genotype) (G). Western blot quantification of BirA*G3 levels in mKate2 low and mKate2 high hepatocytes of Sox2-BirA*G3 liver compared to total control liver. Each lane is a biological replicate from individual mice (n=2/genotype) (H). 111 Supplementary Fig. 3.6. Homogenous mKate2 and BirA*G3 expression was detected in Sox2-BirA*G3 and control mouse pups at 10 days post birth and in CRE-induced liver cells from CAGG-BirA*G3 mice. Immunofluorescence images showed native GFP, native mKate2, BirA*G3, and Albumin in the liver sections of Sox2-BirA*G3 and control mice at 10 days post birth. Scale bar: 20μm (A). Immunofluorescence images showed native GFP, native mKate2, BirA*G3, and CK19 in the liver sections of Sox2-BirA*G3 and control mice at 10 days post birth. Scale bar: 20μm (B). Immunofluorescence images showed native GFP, native mKate2, BirA*G3, and Albumin in the liver sections of CAGG-BirA*G3 and control mice. Scale bar: 20μm (C). Immunofluorescence images showed native GFP, native mKate2, BirA*G3, and CK19 in the liver sections of CAGG-BirA*G3 and control mice. Scale bar: 20μm (D). 112 Supplementary Fig. 3.7. No detrimental effects were observed in Sox2-BirA*G3 and control mice. Representative images of hematoxylin and eosin staining of paraffin-sectioned tissues from Sox2-BirA*G3 and control mice with 7-day biotin chow. Scale Bar: 100μm (A). PCA of Sox2-BirA*G3 and control RNA sequencing samples from all tissues (liver, brain, and kidney). Each 113 point represents one sample (biological replicate) where color is by tissue type and shape is by genotype (B). Volcano plot showing all differentially expressed genes (DEGs) in Sox2-BirA*G3 samples compared to control samples for all three tissues. Increased (red) are DEGs in the Sox2-BirA*G3 condition and decreased (blue) are DEGs in the control condition (C). Heatmap of ER stress and unfolded protein response genes log 2 fold change (scale) from differential expression analysis between Sox2-BirA*G3 and control samples showing hierarchical clustering by tissue (D). Western blot and quantification normalized to total protein stain (not shown) except EIF2𝛼 (normalized to total EIF2𝛼) of ER stress markers BIP (Grp78) (E, p-value 0.32), phosphorylated (p) EIF2𝛼 and total EIF2𝛼 (F, p-value 0.23), CHOP (G, p-value 0.26), ATF6 (H, p-value 0.37), XBP-1s (I, p-value 0.75), and apoptotic marker CC3 (J, p-value 0.35) of 25𝜇g liver lysate from control and Sox2-BirA*G3 samples compared to ER stress controls (mouse 3T3 cells treated with or without Tunicamycin (Tn)). Arrows in D indicate molecular weights for total ATF6 (predicted 75 kDa) and cleaved/active ATF6 (50 kDa). Each lane is a biological replicate from individual mice (n=3/genotype). Statistical significance (significance = p-value < 0.05; ns: non-significant) was calculated using a two tailed, Welch’s t-test between Sox2-BirA*G3 and control samples (E-J). Supplementary Fig. 3.8. Streptavidin purification of biotinylated proteins for LC-MS/MS from Sox2- BirA*G3 mice. Western blotting of streptavidin affinity purified biotinylated proteins from liver (A), brain (B), kidney (C), serum (D) in Sox2-BirA*G3 mice compared to control mice. Upper: Streptavidin labeling. Lower: BirA*G3 (~35kDa). Each lane is a biological replicate from individual mice (n=3/genotype) (A-D). Silver stain of streptavidin affinity purified biotinylated proteins from liver (E), brain (F), kidney (G), and serum (H) in Sox2-BirA*G3 mice compared to control mice. Bead lanes are affinity purification negative control without protein input to show streptavidin contribution to bound fractions from beads. Each lane is a biological replicate from individual mice (n=3/genotype) (E-H). 114 Supplementary Fig. 3.9. Quantitative LC-MS/MS data analysis for three tissues. Summary of information obtained from quantitative LC-MS/MS (A). Volcano plots of proteins detected in 115 brain (B) and kidney (C) of Sox2-BirA*G3 mice compared to control mice after streptavidin pulldown. Log 2 FC were plotted on the x-axis and -10log 10 (p value) were plotted on the y-axis. Significantly enriched proteins (adj. p-value< 0.05 and log 2FC>1) in Sox2-BirA*G3 mice compared to control mice are shown in green or red (B-C). Percentage of proteins with predicted SignalP/TMH in each ES category. As the ES increases, the fraction of proteins with predicted SignalP or TMH increases (D-F). Shared enriched proteins among three tissues (113 proteins) were analyzed with clusterProfiler (3.16.1) EnrichGO analysis for cellular components annotation. Gene ratio indicates the percentage of genes annotated with the term over the total number of genes in the list (G). Shared enriched proteins among three tissues predicted with SignalP/TMH were analyzed with clusterProfiler (3.16.1) EnrichGO analysis for biological process annotation. Gene ratio indicates the percentage of genes annotated with the term over the total number of genes in the list (H). 116 Supplementary Fig. 3.10. Identification of enriched proteins in Sox2-BirA*G3 tissues. PCA of streptavidin-purified liver (A), brain (B), and kidney (C) from Sox2-BirA*G3 and control mice. Each hollow dot represents a sample, which is colored by the annotation of its genotype. Sox2-BirA*G3 samples are manually circled by green shadow (A-C). Distribution of proteins on the positive and negative control lists over TMT ratios log 2FC (Sox2-BirA*G3/control). (Top) Genes on the positive control (PC) list. (Bottom) Genes on the negative control (NC) list (D). The threshold of the TMT ratio is determined and based on a false positive rate of 0.1 (dashed line), which means that a protein is 10 times more likely to be a true ER protein than a false positive (E). Bar plots show number of proteins with each enrichment score from kidney proteomic data. The ES was defined as the number of comparisons (from 9) in which TMT-ratio > threshold (score 9 is for most confident hits and 0 is background) (F). Venn diagrams showed the overlap of enriched proteins using ES method (in teal circle) or conventional method (in orange circle; log 2 FC>1.0 and adj. p- value<0.05) for liver (D), brain (E), and kidney (F) proteomic data (G-I). 117 Supplementary Fig. 3.11. Tissue-specific analysis for tissue-specific enriched proteins. Tissue-specific enriched proteins from liver (A), brain (B), and kidney (C) were analyzed with TissueEnrich, a tool that calculates tissue-specific gene enrichment in an input gene set (A-C). Pie plots showed the number of brain-specific enriched proteins (184 proteins) (D) and kidney-specific enriched proteins (517 proteins) (E) predicted with SignalP/TMH (D-E). Brain-specific enriched proteins predicted with SignalP/TMH were analyzed with clusterProfiler (3.16.1) EnrichGO analysis for biological process. Gene ratio indicates the percentage of genes annotated with the term over the total number of genes in the list (F). Kidney-specific enriched proteins predicted with SignalP/TMH were analyzed with clusterProfiler (3.16.1) EnrichGO analysis for biological process. Gene ratio indicates the percentage of genes annotated with the term over the total number of genes in the list (G). 118 Supplementary Fig. 3.12. Data analysis for serum mass spectrometry results. PCA of streptavidin-purified serum proteins (A: plex1; B: plex2) from Sox2-BirA*G3 and control mice. Each hollow dot represents a sample, which is colored by the annotation of its genotype. Sox2-BirA*G3 samples are manually circled by pink shadow (A, B). Venn diagram showed the overlap of enriched serum proteins between plex 1 and plex 2 mass spectrometry (C). Serum enriched proteins were analyzed with DAVID analysis for cellular component annotation (D). Serum enriched proteins were analyzed with clusterProfiler (3.16.1) EnrichGO analysis for cellular components annotation (E). Serum enriched proteins predicted with SignalP or/and TMH were analyzed with clusterProfiler (3.16.1) EnrichGO analysis for biological process annotation (F). 119 Supplementary Fig. 3.13. Quantitative MS data analysis and western blot validation for serum. 120 Summary of information obtained from quantitative MS (A). Volcano plot of proteins detected in serum (plex1) of Sox2-BirA*G3 mice compared to control mice after streptavidin pulldown. Log 2 FC were plotted on the x-axis and -10Log 10 (p value) were plotted on the y-axis. Significantly enriched proteins (adj. p- value< 0.05 and log 2FC>1.0) in Sox2-BirA*G3 mice compared to control mice are shown in red. Volcano plot of proteins detected in serum (plex2) of Sox2-BirA*G3 mice compared to control mice after streptavidin pulldown were not shown (B). Pie chart displayed the distribution of serum enriched proteins with predicted SignalP/TMH (C). Shared enriched proteins between serum and three tissues were analyzed with clusterProfiler (3.16.1) EnrichGO enrichment analysis for cellular component annotation (D). Shared enriched proteins between serum and three tissues were analyzed with clusterProfiler (3.16.1) EnrichGO enrichment analysis for biological process annotation (E). Schematic of detected peptides for PIGR mapped onto its respective reference sequences with SMART protein domain annotation. Reference sequence is annotated with extracellular, TM and cytoplasmic based on UniProt topology information. Amino acid sequences of the most C-terminal peptide are labeled. IG: immunoglobulin (F). Schematic of detected peptides for EGFR mapped onto its respective reference sequences. Reference sequence is annotated with extracellular, TM and cytoplasmic based on UniProt topology information. Amino acid sequences of the most C-terminal peptide are labeled (G). Schematic of detected peptides for VCAM1 mapped onto its respective reference sequences. Reference sequence is annotated with extracellular, TM and cytoplasmic based on UniProt topology information. Amino acid sequences of the most C-terminal peptide are labeled (H). Mass spectrometry hits of well characterized secreted proteins showing Log 2FC and significant adjusted p-value enrichment method compared to enrichment score (ES) method (I). Western blot validation of streptavidin affinity purified serum hits from (I) in additional control and Sox2-BirA*G3 mice (n=2/sex). Input: 100, 200μg, or 600μg protein. Each lane is a biological replicate from individual mice (n=4/genotype) (J). Supplementary Fig. 3.14. Characterization of Alb-BirA*G3 and control mice. Bright field images of well-perfused organs from Alb-BirA and control mice in comparison with organs without perfusion (A). Immunofluorescence staining shows expression of native GFP, native mKate2, and BirA*G3 expression in the kidney sections from Alb-BirA and control mice. Scale Bar: 50μm (B). 121 Supplementary Fig. 3.15. Characterization of protein biotinylation in Alb-BirA*G3 and control mice. Bright field images of well-perfused organs from Alb-BirA*G3 and control mice in comparison with organs without perfusion (A). Total protein stain of affinity purified biotinylated proteins from liver (B) and serum (C) in Alb-BirA*G3 mice compared to control mice. Specific bands are indicated by asterisks. Bead lanes are affinity purification negative control without protein input to show streptavidin contribution to bound fractions from beads. Each lane is a biological replicate from individual mice (n=3/genotype) (B-C). Streptavidin labeling of affinity purified biotinylated proteins from liver (D) and serum (E) in Alb-BirA*G3 mice compared to control mice. Lower: BirA*G3 (~35kDa). Bead lanes are affinity purification negative control without protein input. Each lane is a biological replicate from individual mice (n=3/genotype) (D- E). 122 Supplementary Figure 3.16. Analysis of Alb-BirA*G3 and control serum enriched proteins. Enriched serum proteins in Alb-BirA*G3 mice were analyzed with DAVID analysis for cellular component annotation. Gene ratio indicates the percentage of genes annotated with the term over the total number of genes in the list (A). Enriched serum proteins in Alb-BirA*G3 mice were analyzed with clusterProfiler (3.16.1) EnrichGO enrichment analysis for cellular component annotation. Gene ratio indicates the percentage of genes annotated with the term over the total number of genes in the list (B). Bar plot showed the percentage of unique (n=137) or shared (n=45) Alb-BirA*G3 secreted proteins (compared with previous datasets) with predicted SignalP (C). Unique Alb-BirA*G3 secreted proteins (n=137) were analyzed with TissueEnrich to calculate tissue-specific gene enrichment (D). 123 Supplementary Table 3.1. Primary antibodies information. Primary antibodies used. WB: western blot, IF: immunofluorescence, Rb: rabbit, Ms: mouse, Sh: sheep, Gt: goat, Ch: chicken, CST: cell signaling technologies Antibody Host Supplier Catalog Number WB IF A2M Sh Novus AF5798-SP 1:1000 - Adipoq Rb Abcam ab181699 1:1000 - Adipsin/Cfd Sh R&Dsystems AF5430 1:1000 - Agt Rb IBL 28101 1:1000 - Alb Ch Abcam ab106582 1:1000 - Alb Ch Sigma SAB3500217 1:1000 1:500 α-SMA-Cy3 Ms Sigma C6198 - 1:500 Apoa1 Rb Abcam ab126786 1:1000 - Aqp2 Ms Santa Cruz sc-515770 - 1:500 ATF4 Rb CST 11815S 1:1000 - ATF6 Ms Novus Bio NBP1-40256-0.1mg 1:1000 - BiP Rb CST 3177S 1:1000 - BirA Ms Abcam ab232732 1:1000 1:250 C7 Rb Abcam ab233166 1:1000 - Calx Rb Abcam ab22595 - 1:250 CC3 Rb CST 9661S 1:1000 - CHOP Rb CST 5554S 1:1000 - Ck19 Rb Abcam ab52625 - 1:500 p-EIF2α (Ser51) Rb CST 9721S 1:1000 - EIF2α Rb CST 9722S 1:1000 - GFAP Ch Aves Aves Chicken GFAP - 1:1000 Gpx-3 Rb Abcam ab256470 - 1:1000 IBA-1 Rb Wako/Fujifilm Distributor: 019- 19741 - 1:500 Map2 Ch Abcam ab5392 - 1:500 Myc-Tag Rb Abcam ab9106 1:1000 - NFH Ch Aves Aves Chicken NFH - 1:500 Nos1 Gt Novus Bio NB100-858 - 1:1000 PZP Rb Abcam ab11805 1:1000 - Renin Rt R&Dsystems MAB42771 - 1:500 Slc12a3 Rb Sigma HPA028748 - 1:500 Umod Rb Alfa Aesar J65429 - 1:500 Umod Rb Abcam ab207170 1:1000 - Wt-1 Rb Thermo MA1-46028 - 1:500 Xbp-1s Rb CST 12782S 1:1000 - 124 Supplementary Table 3.2. Secondary antibodies information. Secondary antibodies used. WB: western blot, IF: immunofluorescence, Dk: donkey, Sh: sheep, Gt: goat, Rt: rat. Antibody Host Supplier Catalog Number WB IF Alexa Fluor Plus 488 anti-Chicken IgY Gt Invitrogen A32931 - 1:500 Alexa Fluor 546 anti-Chicken IgY Gt Invitrogen A-11040 - 1:500 Alexa Fluor 546 anti-Rabbit IgG (H+L) Gt Invitrogen A11035 - 1:500 Alexa Fluor 546 anti-Rat IgG Gt Invitrogen A-11081 - 1:500 Alexa Fluor 647 anti-Chicken IgY Gt Invitrogen A-21449 - 1:500 Alexa Fluor® 647 anti-Mouse IgG1 Rt BioLegend 406618 - 1:500 IRDye 680RD anti-Chicken Dk Li-Cor 926-68075 1:10,000 - IRDye 680RD anti-Goat Dk Li-Cor 926-68074 1:10,000 - IRDye 680RD anti-Mouse Gt Li-Cor 926-68072 1:10,000 - IRDye 680RD anti-Rabbit Gt Li-Cor 926-68073 1:10,000 - IRDye 680RD anti-Rat Dk Li-Cor 926-68076 1:10,000 - IRDye 800CW anti-Chicken Dk Li-Cor 926-32218 1:10,000 - IRDye 800CW anti-Goat Dk Li-Cor 926-32214 1:10,000 - IRDye 800CW anti-Mouse Gt Li-Cor 926-32212 1:10,000 - IRDye 800CW anti-Rabbit Gt Li-Cor 926-32213 1:10,000 - IRDye 800CW anti-Rat Dk Li-Cor 926-32219 1:10,000 - Alexa Fluor® Sheep-790 IgG Dk Jackson Immuno Research Labs 713-655-147 1:50,000 - IRDye 680RD Streptavidin NA Li-Cor 926-68079 1:5,000 - IRDye 800CW Streptavidin Li-Cor 926-32230 1:5,000 - 125 Chapter 4 In vivo proximity labeling identifies adipocyte-specific signaling changes in diet induced obesity This work was led by myself with data collection and manuscript preparation and contribution done by myself. Jinjin Guo supported mouse experiments and colony maintenance. Namrata Udeshi, Charles Xu, and Steven A. Carr executed proteomic experiments and spectral analysis for the white adipose tissue data. Jihui Sha and James Wolhschlegel executed proteomic experiments and spectral analysis for the serum and brown adipose tissue data. Andrew P. McMahon supervised, advised on experimental design, data analysis, and manuscript preparation and review. Shingo Kajimura provided adipose tissue expertise and consulting. INTRODUCTION Obesity is a global disease affecting over 1 billion people in the world 58-61 . Obesity affects individuals at any or all life stages; from childhood to adulthood. It is a comorbidity and/or risk factor for most major common diseases and is marked by key endocrine changes 62-72 . Obesity drives dysregulation and dysfunction of adipose tissue leading to a multitude of cellular and endocrine level changes 58-72 . In obesity, adipocytes, the specialized energy storage cell, undergo hypertrophy and hyperplasia to accommodate the additional energy storage requirements. However, there is a point at which this compensation breaks down and adipose tissue becomes unable to bear the excess energy storage requirements. Adipocyte death increases in conjunction with 126 hypoxia, driving infiltration of pro-inflammatory macrophages thought to contribute to systemic low-grade inflammation 68,217-219 . Further, there is perturbation of adipocyte secretion leading to marked changes in circulating levels of adipokines which contributes to obesity-related pathophysiology in distal organs 215,217,219-223 . The dysregulation and dysfunction of adipocytes leads to disruptions in both locally and distally, however, despite the extensive research in these areas the full extent of mechanisms by which adipocytes normal signaling regulates local and distal cell processes remains unclear. Adipose tissue is found in specific-regions throughout the body, termed depots. These depots can be either white (energy storage) adipose tissue or brown (thermogenic) adipose tissue. White adipose tissue depots have been differentially linked to metabolic disease with visceral adipose (VAT) being more highly correlative with metabolic disease than subcutaneous adipose tissue (SAT) 213,220 . Brown adipose (BAT), rapidly increases (browning of fat) in response to cold exposure and B-adrenergic signaling and increases in caloric restriction 58,69,217,218,222 . Further, BAT levels are decreased in obesity while WAT levels are increased, known as whitening of fat 223 . Despite the evidence of depot-specific contribution to the pathogenicity of obesity, the mechanisms by which different depots are affected and contribute to obesity signatures is poorly understood. Limitations in methods to study cell-type specific secretion changes has contributed to the difficulty in identifying additional adipocyte signaling in homeostasis and obesity. In vitro studies have been used to identify obesity induced adipocyte secretion changes, however, these studies are unable to recapitulate distal endocrine signaling of blood circulating proteins. Further, locally secreted proteins such as extracellular matrix proteins and paracrine signals can confound distal signaling identification attempts. Approaches to profile circulating proteins, such as blood mass spectrometry, are often limited by difficulty in identifying low abundance proteins and 127 cannot inform on the tissue of origin. Further, studies are limited by the lack of readily available reagents, particularly antibodies, for poorly characterized proteins. To overcome these limitations, proximity labeling, the tagging of proximal proteins by a promiscuous enzyme, has been applied to study protein-protein interactions (PPI) and map subcellular compartments, such as the secretory pathway 6,22,25,51 . Proximity labeling, such as TurboID, allows for fast, non-toxic biotinylation of proteins which can be subsequently purified and identified using quantitative proteomics. Recently, our group and others, generated genetic proximity labeling mice with floxed alleles to allow for cell-type specific secretome and proteome profiling 10-12 . These models showed strong cell-type specific and subcellular compartment mapping capabilities in homeostasis but have yet to be applied to an in vivo disease system. Here, we apply endoplasmic reticulum (ER) localized BirA*G3 proximity labeling to identify adipocyte secretion changes in diet induced obesity 224 . Activation of BirA*G3 in mouse adipocytes via adiponectin-Cre 225 results in biotinylation of adipocyte secreted proteins. To study the effects of diet induced obesity on adipocyte secretion, we profiled the serum secretomes from standard chow (SD) and high fat diet (HFD) fed mice by TMT proteomics 224 . To investigate depot-specific secretion changes both locally and circulating, we profiled inguinal SAT and perirenal VAT from SD and HFD fed mice by TMT proteomics. In both serum and adipose tissue, we identify numerous secreted proteins including adipokines when compared to non-labeled controls. Further, we captured both known and novel diet induced secretion changes in serum (HP) and adipose tissue (LEP, FABP4). Of serum identified proteins, 84% were predicted to have a signal peptide. In adipose tissue, only ~12% were predicted to have a signal peptide and many proteins were annotated to be ER-resident proteins, transmembrane proteins, or extracellular matrix proteins supporting the capture of local secreted proteins in tissue 128 and distal circulating proteins in serum. Taken together, our data establishes the ability of proximity labeling to identify adipocyte secretion changes in DIO. Further, this method identified novel DIO induced secretion changes identifying potential targets for further mechanistic investigation. METHODS Animal studies Institutional Animal Care and Use Committees (IACUC) at the University of Southern California reviewed and approved all animal work as performed in this study. All work adhered to institutional guidelines. The Adipoq-Cre mice (B6;FVBTg(Adipoq- cre)1Evdr/J, strain: 010803, stock no.: 028020, The Jackson Laboratory) 183 were used as previously described and crossed to BirA*G3-ER mice (Gt(ROSA)26Sortm10.1(CAG- BirA*,-mKate2)Amc, strain: 037395, stock no.: 037395) 129 to activate BirA*G3 in mouse adipocytes. The Sox2-Cre mice (B6.Cg-Edil3Tg(Sox2-cre)1Amc/J, stock no.: 008454, The Jackson Laboratory) 183 and the Alb-Cre mice (B6.Cg-Speer6-ps1Tg(Alb-cre)21Mgn/J, stock no.: 003574) 200 were used as described previously. In vivo Assays For all mouse studies, Adipoq-BirA*G3 and control mice at 12-20 weeks were given biotin water (3mM, pH 7.4; Sigma B4639-5G) for 8 days. For all in vivo assays, tissues were collected as follows. Mice were euthanized at 12-20 weeks, blood was then collected from the inferior vena cava, followed by perfusion with 1X cold DPBS (Dulbecco’s Phosphate Buffered Saline). The blood was allowed to clot at room temperature for 30 minutes and then spun down at 2,000 x g for 15 minutes at 4°C. The serum was collected and spun again at 2,000 x g for 15 minutes at 4°C, then transferred 129 to a fresh tube and flash frozen (in liquid nitrogen) before being stored at -80°C until being used. After perfusion, tissues were collected and rinsed in 1X cold DPBS before being minced with a razor blade and aliquoted into tubes. Tissues were then flash frozen and stored at -80°C until being used. For diet induced obesity studies, Adipoq-BirA*G3 and control mice were put on a standard high fat diet (60% kCal from fat; Envigo Adjusted Calories Diet TD.06414) starting at 6 weeks of age for 12 or 20 weeks. Mice were massed weekly until collection. Whole Mount Staining Whole mount staining preparation was done previously as described 227 with the following modifications. Briefly, adipose tissues were harvested from DPBS-perfused mice, mice into approximately 5 x 5 mm pieces, and then fixed in 4% paraformaldehyde in a 96-well plate, overnight at 4°C. Tissues were then washed 3 times in 1X DPBS before being incubated in blocking buffer (2.0% Sea Block (Thermo 37527) + 0.125% Triton-X100 in 1X DPBS) for 1 hour at room temperature. Tissues were then incubated in 100μL primary antibody (Supplemental Table 1) diluted in blocking buffer, for 4 days at 4°C. After, primary antibodies were removed, and tissues were washed in blocking buffer four times for 20 minutes each. Tissues were then incubated in 100μL secondary antibody diluted in blocking buffer for 4 days at 4°C. Secondary antibody (Supplemental Table 2) was removed, and tissues were washed in blocking buffer four times for 20 minutes each. Tissues were then incubated in 1 mg/mL Hoechst 33342 (Thermo H3570) in 1X DPBS with calcium and magnesium for 1 hour at room temperature. Tissues were then washed twice in 1X DPBS with calcium and magnesium for 10 minutes each. Tissues were mounted in Immu-Mount (Thermo 9990402) by creating a vaseline barrier on a microscope slide. Tissues were placed within the barrier and mounting media was 130 flooded over the tissues. A coverslip was then applied and gently pressed onto the samples to ensure all tissues were in an even field. The slides were sealed with nail polish and let to set overnight at room temperature. The slides were imaged at 40X or 63X using the Leica SP8 confocal microscope. Frozen Tissue Preparation and Sectioning Briefly, tissues were harvested from DPBS-perfused mice and then fixed in 4% paraformaldehyde overnight at 4°C. Tissues were then washed 3 times in 1X DPBS with calcium and magnesium before being incubated in 30% sucrose overnight. The following day, tissues were washed in OCT 3 times to remove excess sucrose and then embedded in OCT (VWR, 25608-930) and frozen in a dry-ice ethanol bath before being stored at - 80°C. Tissue blocks were thawed to -20°C overnight and then cryosectioned at 10-16μm at -20°C and placed on glass slides. Slides were then stored at -80°C until immunostained. Immunofluorescent Staining and Confocal Microscopy Frozen sectioned tissues were thawed at room temp for 10 minutes. To remove lipids from adipose tissue sections, slides were incubated in ice-cold 50% acetone, 50% methanol for 10 minutes and then briefly dried 239 . Slides were then rinsed in 1X DPBS for 10 minutes. Slides were permeabilized in 0.25% Triton-X100 (Sigma X100-500ML) for 5 minutes, then incubated in blocking buffer (2.0% Sea Block (Thermo 37527) + 0.125% Triton-X100 in 1X DPBS) for 1 hour at room temperature. Slides were then incubated in primary antibody (Supplemental Table 1) diluted in blocking buffer, overnight at 4°C. The following day, primary antibodies were removed, and slides were washed in blocking buffer four times for 5 minutes each. Slides were then incubated in secondary antibody diluted in blocking buffer for 1 hour at room temperature. Secondary antibody 131 (Supplemental Table 2) was removed, and slides were washed in blocking buffer four times for 5 minutes each. Slides were then incubated in 1 mg/mL Hoechst 33342 (Thermo H3570) in 1X DPBS with calcium and magnesium for 10 minutes at room temperature. Slides were then washed twice in 1X DPBS with calcium and magnesium for 5 minutes each. Slides were mounted in Immu-Mount (Thermo 9990402) and imaged at 40X or 63X using the Leica SP8 confocal microscope. Whole tissue section scans were imaged using the Zeiss AxioScan Z1 Slide Scanner at 20X to generate high-resolution tiled images of tissues sections. ELISA Analysis Adiponectin sandwich ELISA (R&D systems MRP300) was performed on mouse serum (1:10,000 dilution) following manufacturer’s instructions. Detection antibody was HRP-based to avoid signal interference from biotinylated proteins. Protein lysate preparation Protein lysates were prepared as described previously 11 with the following modifications. homogenized in 500μL RIPA complete lysis buffer (RIPA buffer (ThermoFisher, 89901) with 1X cOmplete EDTA-free protease inhibitor cocktail (Sigma, 11873580001), 1mM benzamidine hydrochloride (VWR, TCB0013-100G), 4μM pepstatin A (Sigma EI10), 100μM PMSF (Sigma 11359061001)) and bead homogenized using stainless steel beads (NextAdvance SSB14B-RNA) for 5 minutes at setting 10, Bullet Blender Storm (NextAdvance BT24M). Samples were then centrifuged at 14,000 x g for 15 minutes at 4°C. Supernatants were transferred to protein loBind (Eppendorf 22431081) tubes. Protein lysate concentrations were determined using Pierce BCA (Thermo 23227) microplate assay per manufacturer’s instructions. Lysates were then stored at -80°C. 132 Streptavidin beads pulldowns Streptavidin pulldowns were performed as described previously13 with modifications. Streptavidin magnetic (Thermo 88817) or high-capacity strepatvidin agarose (Thermo 20359) beads were resuspended in lysis buffer (above) by magnetic separation (BioRad 1614916). Agarose beads were packed by centrifugation. We tested a series of volumes of beads and washing conditions and found that 20μL beads per 100μg protein for adipose tissue and 60μL of high-capacity streptavidin agarose beads (Thermo 20359) with 4mg of serum protein together with the following washing conditions are sufficient (results available upon request). Pulldown reactions were set up in 450μL lysis buffer with 5μL beads per 100μg protein. Pulldowns were then incubated overnight at 4°C in a wheel rotator. The following day, pulldown reactions were washed twice in lysis buffer, then once in 8M Urea (Sigma U5378-500G) in 100mM Tris (pH 8.5), and finally twice lysis buffer. After the final wash, lysis buffer was removed and beads were either boiled in 12μL 1X loading buffer (Li-Cor 928-40004) with 1.43M β-mercaptoethanol or resuspended in 100μL lysis buffer and flash frozen, for western blotting and mass spectrometry respectfully. Silver Stain Analysis Silver staining was done using Richard J. Simpson’s protocol from Cold Spring Harbor (CSH). Gels were fixed in a 50% methanol (VWR BDH1135-4LG), 5% glacial acetic acid (VWR 97064-482) solution, gently shaking at room temperature for 20 minutes, Gels were then incubated in 50% methanol for 10 minutes, gently shaking at room temperature, followed by a 10-minute incubation in dH 2 O. Gels were then soaked in 0.02% sodium thiosulfate (Sigma 72049) for 1 minute and then in dH 2 O for 1 minute, 133 twice. Gels were then incubated in chilled 0.1% silver nitrate (Sigma 209139) for 20 minutes, gently shaking at 4C in the dark. Gels were then rinsed twice in dH 2 O for 1 minute each. Gels were developed in a 2% sodium carbonate (Sigma 222321) and 0.04% formaldehyde (Thermo 28906) until desired intensity was reached. Developing was stopped with a 5% glacial acetic solution and gels were stored in 1% glacial acetic acid until being discarded. Fluorescent Western Blot Analysis Western blots were performed with standard protocols and the following modifications. Equal amounts of total protein lysate were loaded per sample per reaction with 1X Li-Cor loading buffer (Li-Cor, 928-40004) with 1.43 M β-mercaptoethanol. For streptavidin pulldowns, beads were resuspended in 12μL 1X Li-Cor loading buffer (Li- Cor, 928-40004) with 1.43 M β-mercaptoethanol. All samples were then boiled at 95°C for 5 minutes to elute, then briefly spun down and kept on ice prior to loading. Total protein samples and pulldown elutes were loaded on 10% SDS acrylamide gels and run in standard 1X SDS-Running buffer with Li-Cor 5μL one-color molecular marker (Li-Cor, 928-40000) at 60V for 30 minutes, followed by 120V for ~50 minutes or until loading dye ran off. Samples on the gel were transferred to methanol activated PVDF 0.45μm membranes using BioRad’s wet tank mini-protean system for 1-3 hours at 250-300 constant mA in a sample dependent context. After transfer, membranes were dried at 37°C for 5 minutes and then re-activated with methanol. Blots were stained with Li-Cor’s Revert-700 Total Protein Stain (Li-Cor, 926-11010) for normalization and imaged using a Li-Cor Odyssey Clx. Blots were then de-stained per kit instructions and put in block (Li- Cor Intercept block, 927-60001) for 1 hour, room temperature, shaking. Blots were then transferred to primary antibody (Supplementary Table 1) (block with 0.2% Tween20) 134 overnight at 4°C, shaking. The following day, blots were washed four times in TBS-T for 5 minutes each at room temperature, shaking, and then incubated in secondary antibody (Supplementary Table 2) in block with 0.2% Tween20 and 0.1% SDS, and/or streptavidin conjugate (1:5,000; 680 or 800, Li-Cor, 926-68079, 926-32230) if visualizing biotinylated proteins, for 1 hour at room temperature, shaking. Blots were then washed twice with TBS-T for 5 minutes each, room temperature, shaking, followed by two 5-minute TBS washes at room temperature, shaking. Blots were imaged on a Li-Cor Odyssey Clx using Li-Cor’s ImageStudio (Version 5.2.5). After imaging blots were dried at 37°C for 5 minutes, then stored. All western blot images were exported from Li-Cor, pseudo-colored and converted to RGB tiffs in ImageJ (v1.51S) for figures. For specific proteins, bands were selected based on molecular weight from antibody manufacturer information and literature. Fluorescent Western Blot Quantification Biotinylation levels and proteins of interest were quantified via western blot using Li-Cor’s fluorescent western blot ImageStudio (Version 5.2.5) and Emperia Studio (Version 1.3.0.83) analysis software and protocols. Total protein stain images of each blot were used to normalize biotinylation (streptavidin) or protein of interest signal intensity in RStudio (Version 1.3.959, R Version 4.1.3) by determining the lane normalization factor (Li-Cor protocol) for each blot per manufacturer’s instructions. ggplot2 (Version 3.3.5) and GraphPad Prism 9.4.1 were used to visualize normalized biotinylated protein and protein of interest signal. Adipoq-BirA*G3 Analysis by MS (corresponding to Figure 3 and Supplemental 135 Figures 3, 4) After streptavidin beads pulldowns, SAT and VAT Adipoq-BirA*G3 and BirA*G3 (control) samples were sent to The Broad Institute of Harvard and MIT for MS. LC-MS/MS Analysis i. On-bead digestion Samples were collected and enriched with streptavidin magnetic beads (Thermo 88817) (adipose tissue input: 1.5mg), washed twice with 200 μL of 50mM Tris-HCl buffer (pH 7.5), transferred into new 1.5 mL Eppendorf tubes, and washed 2 more times with 200 μL of 50mM Tris (pH 7.5) buffer. Samples were incubated in 0.4 μg trypsin in 80 μL of 2M urea/50mM Tris buffer with 1 mM DTT, for 1 h at room temperature while shaking at 1000 rpm. Following pre-digestion, 80 μL of each supernatant was transferred into new tubes. Beads were then incubated in 80 uL of the same digestion buffer for 30 min while shaking at 1000rpm. Supernatant was transferred to the tube containing the previous elution. Beads were washed twice with 60 μL of 2M urea/50mM Tris buffer, and these washes were combined with the supernatant. The eluates were spun down at 5000 × g for 1 min and the supernatant was transferred to a new tube. Samples were reduced with 4 mM DTT for 30 min at room temperature, with shaking. Following reduction, samples were alkylated with 10mM iodoacetamide for 45 min in the dark at room temperature. An additional 0.5 μg of trypsin was added and samples were digested overnight at room temperature while shaking at 700 × g. Following overnight digestion, samples were acidified (pH < 3) with neat formic acid (FA), to a final concentration of 1% FA. Samples were spun down and desalted on C18 StageTips as previously described56. Eluted peptides were dried to completion and stored at −80 °C. 136 ii. TMT labeling of peptides Desalted peptides were labeled with TMT 18-plex (adipose tissue) reagents (ThermoFisher Scientific). Peptides were resuspended in 80 μL of 50 mM HEPES and labeled with 20 uL 20mg/mL TMT18 reagents in ACN. Samples were incubated at RT for 1 h with shaking at 1000 × rpm. TMT reaction was quenched with 4 μL of 5% hydroxylamine at room temperature for 15min with shaking. TMT labeled samples were combined, dried to completion, reconstituted in 100 μL of 0.1% FA, and desalted on StageTips. iii. bRP stage tip fractionation of peptides 50% of the TMT labeled peptide sample was fractionated by basic reverse phase (bRP) fractionation. StageTips packed with 3 disks of SDB-RPS (Empore) material. StageTips were conditioned with 100 μL of 100% MeOH, followed by 100 μL 50% MeCN/0.1% FA and two washes with 100 μL 0.1% FA. Peptide samples were resuspended in 200 μL 1% FA (pH<3) and loaded onto StageTips. 6 step-wise elutions were carried out in 100 μL 20 mM ammonium formate buffer with increasing concentration of 5%, 10%, 15%, 20%, 25%, and 45% MeCN. Eluted fractions were dried to completion. iv. Liquid chromatography and mass spectrometry Single-shot LC-MS/MS analyses were performed on 50% of each sample. The remaining 50% of each sample was fractionated using bRP StageTip fractionation. For single shot and all fractionated samples, desalted peptides were resuspended in 9 μL of 3% MeCN/0.1% FA and 4 μL was injected. For serum samples, an Orbitrap Fusion Lumos 137 Tribrid Mass Spectrometer (ThermoFisher Scientific) was used. For all other plexes, an Orbitrap Exploris 480 (ThermoFisher Scientific) was used. Mass spectrometers were coupled online to a Proxeon Easy-nLC 1200 (ThermoFisher Scientific) as previously described56. Briefly, 4 μL of each sample was loaded at onto a microcapillary column (360 μm outer diameter × 75 μm inner diameter) containing an integrated electrospray emitter tip (10 μm), packed to approximately 24 cm with ReproSil-Pur C18-AQ 1.9 μm beads (Dr. Maisch GmbH) and heated to 50 °C. bRP fractionated samples were analyzed using a 110 min LC–MS. Mobile phase flow rate was 200 nL/min, comprises 3% acetonitrile/0.1% formic acid (Solvent A) and 90% acetonitrile /0.1% formic acid (Solvent B). The 110-min LC–MS/MS method used the following gradient profile: (min:%B) 0:2; 1:6; 85:30; 94:60; 95:90; 100:90; 101:50; 110:50 (the last two steps at 500 nL/min flow rate).. Data acquisition was done in the data-dependent mode acquiring HCD MS/MS scans (r = 15,000) after each MS1 scan (r = 60,000) on the top 12 most abundant ions using an MS1 AGC target of 4 x 105 and an MS2 AGC target of 5 × 104. The maximum ion time utilized for MS/MS scans was 120 ms; the HCD-normalized collision energy was set to 36 (Fusion Lumos) or 28 (Exploris 480); the dynamic exclusion time was set to 20 s, and the peptide match and isotope exclusion functions were enabled. Charge exclusion was enabled for charge states that were unassigned, 1 and >7. MS data analysis All protein trafficking MS data were analyzed using Spectrum Mill software package v 7.07 (proteomics.broadinstitute.org)). Similar MS/MS spectra acquired on the same precursor m/z within ±60 s were merged. MS/MS spectra were excluded from searching if they were not within the precursor MH+ range of 600–6000 Da or if they failed the quality filter by not having a sequence tag length >0. MS/MS spectra were 138 searched against a UniProt mouse database with a release date of December 28, 2017 containing 46,519 proteins and 264 common contaminants modified to include GFP, mKate2 and BirA*G3-ER. All spectra were allowed ±20 ppm mass tolerance for precursor and product ions, 40% minimum matched peak intensity, and “trypsin allow P” enzyme specificity with up to 2 missed cleavages. The fixed modifications were carbamidomethylation at cysteine, and TMT6 at N-termini. The variable modifications used were oxidized methionine and N-terminal protein acetylation. Individual spectra were automatically designated as confidently assigned using the Spectrum Mill autovalidation module. Specifically, a target-decoy-based false-discovery rate (FDR) scoring threshold criteria via a two-step auto threshold strategy at the spectral and protein levels was used. First, peptide mode was set to allow automatic variable range precursor mass filtering with score thresholds optimized to yield a spectral level FDR of <1.2%. A protein polishing autovalidation was applied to further filter the peptide spectrum matches using a target protein level FDR threshold of 0. Following autovalidation, a protein–protein comparison table was generated, which contained experimental over control TMT ratios. For all experiments, non-mouse contaminants and reverse hits were removed. Furthermore, the data were median normalized. For serum data, we performed a moderated T-test (limma R package v4.1) to identify proteins significantly enriched in the experimental conditions compared to controls. We corrected for multiple hypotheses (Benjamini–Hochberg procedure). Any protein with an adjusted p-value of less than 0.05 and a log2 fold change greater than 1 was considered statistically enriched. For tissue data, we used ES method described below (MS hit analysis) to identify enriched proteins. Adipoq-BirA*G3 Serum Analysis by MS (corresponding to Figures 4 and Supplemental 139 Figure 6, 7) After streptavidin beads pulldowns, Adipoq-BirA*G3 and BirA*G3 (control) samples were sent to the UCLA proteomics core, Department of Biological Chemistry, Geffen School of Medicine at UCLA for MS. i. Serum Sample Digestion Streptavidin-bound proteins were reduced and alkylated on bead via sequential 20-minute incubations with 5mM TCEP and 10mM iodoacetamide at room temperature in the dark while being mixed at 1200 rpm in an Eppendorf thermomixer. Proteins were then digested by the addition of 0.1μg Lys-C (FUJIFILM Wako Pure Chemical Corporation, 125-05061) and 0.8μg Trypsin (Thermo Scientific, 90057) while shaking at 37°C overnight. ii. TMT Labeling and CIF Fractionation The supernatant was transferred to new tubes and 8 μl of carboxylate-modified magnetic beads (CMMB, and also widely known as SP3(60)) was added to each sample. 100% acetonitrile was added to each sample to increase the final acetonitrile concentration to >95% and induce peptide binding to CMMB. CMMB were then washed 3 times with 100% acetonitrile and then resuspended with TMT labeling buffer. 25 ug of each sample was labeled using TMT10 plex Isobaric Labels (Thermo Fisher Scientific) and the resulting 8 labeled samples were pooled. The pooled sample was fractionated by CMMB-based Isopropanol Gradient Peptide Fractionation (CIF) method(61) into 3 fractions before MS analysis. iii. LC-MS Acquisition and Analysis. Fractionated samples were separated on a 75uM ID x 25cm C18 column packed with 1.9μm C18 particles (Dr. Maisch GmbH) using a 140-minute gradient of increasing acetonitrile and eluted directly into a Thermo Orbitrap Fusion Lumos mass spectrometer 140 where MS spectra were acquired using SPS-MS3. Protein identification was performed using MaxQuant62 v 1.6.17.0. The complete Uniprot mouse proteome reference database (UP000000589) was searched for matching MS/MS spectra. Searches were performed using a 20 ppm precursor ion tolerance. TMT10plex was set as a static modification on lysine and peptide N terminal. Carbamidomethylation of cysteine was set as static modification, while oxidation of methionine residues and N-terminal protein acetylation were set as variable modifications. LysC and Trypsin were selected as enzyme specificity with maximum of two missed cleavages allowed. 1% false discovery rate was used as a filter at both protein and PSM levels. Statistical analysis was conducted with the MSstatsTMT Bioconductor package63. The abundance of proteins missing from one condition but found in more than 2 biological replicates of the other condition for any given comparison were estimated by imputing intensity values from the lowest observed MS1-intensity across samples and p- values were randomly assigned to those between 0.05 and 0.01 for illustration purposes. MS Hit Analysis To identify enriched proteins from MS data, Adipoq-BirA*G3 were compared to BirA*G3 controls. Adipoq-BirA*G3 proteins with a log2 fold change (log2 FC) >= 1.0 and unadjusted p-value < 0.05 were considered enriched. For diet comparisons SD-Adipoq- BirA*G3 samples were compared to HFD-Adipoq-Bira*G3 samples. Proteins were considered SD enriched if they had a log2 FC >= 1.0 and unadjusted p-value < 0.05 and HFD enriched if they had a log2 FC <= -1.0 and unadjusted p-value < 0.05. Data Analysis and Statistics 141 Data was analyzed using Microsoft Excel, R (version 4.1.3 (2022-03-10) (64-bit); RStudio Version 1.3.959), and GraphPad Prism 9.4.1. For secretion annotations, proteins were annotated based on the subcellular localization data from UniProt and the cellular component data from National Center for Biotechnology Information (NCBI) (https://www.ncbi.nlm.nih.gov). Proteins in fasta formats were uploaded to SignalP6.0 (https://services.healthtech.dtu.dk/service.php?SignalP) and TMHMM (v 2.0) (https://services.healthtech.dtu.dk/service.php?TMHMM-2.0) for the prediction of SignalP and transmembrane helix, separately. Gene ontology function annotation was performed using EnrichGO in clusterProfiler (3.16.1). The top GO terms were visualized with barplots in ggplot2. PCA was used to visualize similarities between samples. Data Availability The original mass spectra and the protein sequence databases used for searches have been deposited in the public proteomics repository MassIVE (htt://massive.ucsd.edu) and are accessible at ftp://MSV000088848@massive.ucsd.edu. The following public databases were used: Uniprot (https://www.uniprot.org), mouse (https://www.uniprot.org/proteomes/UP000000589), SignalP 6.0 (https://services.healthtech.dtu.dk/service.php?SignalP), TMHMM 2.0 (https://services.healthtech.dtu.dk/service.php?TMHMM-2.0), and Rosen lab adipocyte single nuclear RNA-seq data (https://gitlab.com/rosen-lab/white-adipose- atlas/-/tree/master). Corresponding authors will provide original data upon request. Source data are provided with this paper. RESULTS 142 Adipocyte secretory compartment protein biotinylation by BirA*G3 in Adipoq- BirA*G3 mice. We previously reported a Cre-inducible endoplasmic reticulum (ER) localized BirA*G3 genetic biotin-based proximity labeling mouse strain 11,50 . Here, we use the BirA*G3 secretome mouse strain to activate proximity labeling enzyme, BirA*G3, in mouse adipocytes by crossing BirA*G3 mice to Adiponectin-Cre mice. This results in a GFP-stop cassette excision leading to BirA*G3 and mKate2 activation in all adipocytes (Fig. 1A). Western blot analysis of proteins from Adipoq-Cre; BirA*G3 (abbreviated as Adipoq-BirA*G3) subcutaneous (SAT) inguinal adipose tissue showed strong biotinylated protein signatures compared to BirA*G3 (control) samples and BirA*G3 expression in recombined mice (Fig. 1B). Biotinylated proteins observed in both subcutaneous (SAT), visceral (VAT), and brown (BAT) adipose tissue from Adipoq-BirA*G3 mice largely co-localize with BirA*G3+ cells as predicted (Fig. 1C-D, Supplementary Fig. 1A-B). Further, non-recombined cells in Adipoq-BirA*G3 are green (GFP+), while recombined adipocytes (FABP4+) are red (mKate2+) and BirA*G3+ (Fig. 1E, Supplemental Fig. 1C-D). Profiling adipocyte-derived circulating proteins through affinity purification and western blotting of serum from Adipoq-BirA*G3 mice showed a strong biotinylated protein signal and identified the known serum adipokines, adipsin and adiponectin (Fig. 1F). Additionally, adiponectin is a sexually dimorphic circulating protein (higher in biological females) 226 , which was captured by proximity labeling (Fig. 1F, G). Adipocyte secretion changes in diet induced obesity Due to the strong biotinylated protein labeling adipose tissue (Supplemental Fig. 2A-B), we next sought to apply this approach to a diet induced rodent obesity (DIO) 143 model to identify diet induced protein secretion changes 224 . For this, we gave a standard high fat diet chow (HFD; 60% kCal fat, Envigo TD.06414) or a standard diet chow (SD; LabDiet SWLP) to Adipoq-BirA*G3 and BirA*G3 mice from 6 weeks to 18 weeks of age given ad libitum (Fig. 2A). Biotin (3mM, pH 7.4) was supplemented through water during the final 8 days to enable BirA*G3 for biotinylation of proteins (Fig. 2A). Mice on HFD showed significant weight gain starting after 6 weeks compared to control mice (Fig. 2B). Western blot analysis of streptavidin affinity purified biotinylated proteins showed a general decrease in serum biotinylated proteins in HFD compared to SD and an increase in VAT HFD biotinylated proteins, with no observable differences in SAT and BAT diet conditions (Fig. 2C-F). SAT showed a marginal increase in the streptavidin bead-bound fraction by silver staining, but not by western blotting (Fig. 2D, Supplemental Fig. 2C), while VAT showed a difference in biotinylated proteins through western blotting, but not in the streptavidin bead-bound fraction by silver staining (Fig. 2E, Supplemental Fig. 2D). Surprisingly, silver staining of the streptavidin bead-bound fraction of serum from Adipoq-BirA*G3 mice showed no protein enrichment compared to BirA*G3 mice. When compared to previously characterized liver proximity labeling (Alb-BirA*G3) and a ubiquitous proximity labeling (Sox2-BirA*G3) mouse line, we saw strong recovery of biotinylated proteins by silver staining (Supplemental Fig. 2K). These findings indicate a diet induced biotinylated protein difference in serum at the western level, but optimization is required to achieve significant enough protein enrichment for proteomic analysis. To optimize serum biotinylated protein enrichment for mass spectrometry, we used silver staining and adiponectin as a benchmark. Adiponectin, an approximately 30 kDa (monomer weight) adipokine, is known to decrease in obesity. Due to the strong decrease in biotinylated protein at approximately 30 kDa, we looked to see if there was a 144 decrease in serum adiponectin. Quantitative western blotting of total serum (no affinity purification) showed no difference between SD and HFD treated mice but did still show a decrease in biotinylated proteins (Supplemental Fig. 2E-H). Quantification of serum adiponectin by ELISA further confirmed the quantitative western blot results showing no difference in adiponectin levels between SD and HFD mice (Supplemental Fig. 2I). Adiponectin levels between females and males were significantly different as expected (Supplemental Fig. 2J). Since, there was no diet induced change in adiponectin levels, we compared a variety of streptavidin bead types and serum input levels using adiponectin as a readout. The use of high-capacity streptavidin beads resulted in strong recovery of serum biotinylated proteins from Adipoq-BirA*G3 compared to BirA*G3 mice (Supplemental Fig. 2L). Further, western blotting showed strong recovery biotinylated proteins with a decrease in the HFD biotinylated protein levels compared to SD without a difference in adiponectin recovery (Supplemental Fig. M). We next sought to identify both adipocyte-derived secreted proteins within adipose tissue and distally secreted proteins in serum by quantitative mass spectrometry. Quantitative TMT proteomics of adipocyte-derived adipose tissue biotinylated proteins in diet induced obesity To investigate potential diet induced adipocyte secretion differences between subcutaneous (SAT) and visceral (VAT) adipose tissue in obesity, biotinylated proteins were affinity purified using streptavidin-conjugated beads from inguinal (SAT) and perirenal (VAT) adipose tissue. Bead bound-fractions of biotinylated proteins from SD (n=3/tissue) and HFD (n=3/tissue) treated mice from Adipoq-BirA*G3 and non-labeled BirA*G3 (control) backgrounds were analyzed by TMT (18plex: n=9 SAT, n=9 VAT) proteomics (Fig. 3A). PCA plots showed clustering of samples by labeled (Adipoq- 145 BirA*G3) or non-labeled (BirA*G3) condition, as well as separation of SAT SD and HFD on PC2 (Fig. 3B). VAT SD and HFD samples did not cluster by diet in the 18plex PCA but did when PCA was done using only VAT samples (Supplemental Fig. 3A-B). Proteins were filtered for enrichment of labeled condition (SD and HFD) over the non-labeled BirA*G3 controls by a log 2 fold change (log 2FC) >= 1.0 (SD/CTR: SD-Adipoq- BirA*G3/BirA*G3 or HFD/CTR: HFD-Adipoq-BirA*G3/BirA*G3) and a significant non- adjusted p-value < 0.05 resulting in 2449 SD/control and 2268 HFD/control enriched proteins in SAT and 2161 SD/control and 2262 HFD/control in VAT (Fig. 3C). The majority of proteins, including many adipokines, were enriched in the Adipoq-BirA*G3 samples compared to the BirA*G3 samples for SD and HFD SAT and VAT samples (Supplemental Fig. 3C-F). To see if we were capturing secreted proteins, we used a signal peptide prediction database (signalP6.0) to determine the percentage of signal peptide containing proteins in the enriched set. In total, both SAT (12.15%) and VAT (12.16%) had ~12.5 % proteins predicted to contain a signal peptide. The percentage of proteins predicted to contain a signal peptide increased with increasing log 2FC (Supplemental Fig. 3G-J). Gene ontology analysis of enriched Adipoq-BirA*G3 proteins revealed terms related to related to protein trafficking through the endoplasmic reticulum and golgi-apparatus (Supplemental Fig. 4A-D). Next, to determine protein enrichment between diet conditions, proteins were annotated as SD or HFD enriched by a log 2 fold change (log 2FC) >= 1.0 (SD) or =< -1.0 (HFD) and significant non-adjusted p-value < 0.05 of SD samples over HFD samples (SD- Adipoq-BirA*G3/HFD-Adipoq-BirA*G3) (Fig. 3B, C). Non-adjusted p-values were used based on previous work comparing multiple biotinylated proximity labeling conditions 12,43 . Volcano plots showed enriched proteins in SD and HFD for SAT and VAT tissues (Fig. 3D-E). SAT resulted in 13 SD and 16 HFD enriched proteins, while VAT 146 resulted in 19 SD and 13 HFD enriched proteins. Leptin (LEP) and fatty acid binding protein 4 (FABP4), two secreted adipocyte proteins that are increased in obesity were identified in the HFD enriched proteins from SAT and VAT supporting identification of diet induced changes. Amyloid precursor protein (APP) has been shown to be increased in HFD adipocytes and was enriched in the SAT HFD condition. Further, haptoglobin (HP) was identified in the VAT HFD enriched proteins. Haptoglobin has been shown to be adipocyte-secreted and increase in HFD as well 130,131 . Heatmaps of SD and HFD enriched proteins in SAT and VAT showed similar enrichment in biological replicates with the exception of one VAT HFD sample (Fig. 3F-G). This sample also showed strong separation from the other two replicates by PC2 (Supplemental Fig. 3B). Gene ontology analysis of SD enriched proteins revealed biological process terms related to transcription and metabolic processes (Fig. 3H-I). Gene ontology analysis of HFD enriched proteins revealed biological process terms related to cytokine signaling in SAT and lipid and cholesterol storage terms in VAT (Fig. 3J-K). Western blotting of FABP4 from SAT, showed similar total levels between SD and HFD, but had an enrichment in HFD samples after streptavidin affinity purification (Supplemental Fig. 5A-D). To determine if these proteins were expressed by adipocytes, we used a recently published single-nuclear RNA-sequencing (snRNA-seq) dataset comparing isolated adipocyte nuclei from SAT (ING: inguinal) and VAT (PG: perigonadal) depots SD (chow) and HFD treated mice. HFD enriched proteins had stronger correlation to the snRNA- seq data for both SAT and VAT samples, while the SD enriched proteins showed little correlation to snRNA-seq diet treatments (Supplemental Fig. 4E-H). Though there was little depot-specificity to our depot enriched SD and HFD proteins in the snRNA-seq data (Supplemental Fig. 4I-L). This was not surprising given the high overlap of identified proteins when Adipoq-BirA*G3 SAT and VAT samples were compared to the BirA*G3 147 controls. Quantitative TMT proteomics of adipocyte derived serum biotinylated proteins in diet induced obesity To identify adipocyte-derived diet induced circulating protein changes, biotinylated proteins were affinity purified using streptavidin-conjugated beads from serum of Adipoq-BirA*G3 SD (n=4), HFD (n=4) mice, and non-labeled BirA*G3 (control) mice (n=2) and bound fractions were analyzed by TMT (10plex) proteomics (Fig. 4A). PCA plot showed clustering of labeled (Adipoq-BirA*G3) and non-labeled BirA*G3 controls and mild separation of SD Adipoq-BirA*G3 and HFD Adipoq-BirA*G3 samples along PC2 (Fig. 4B). Proteins were filtered for enrichment of labeled condition (SD and HFD) over the non-labeled BirA*G3 controls by a log 2 fold change (log 2FC) >= 1.0 (SD/CTR: SD-Adipoq-BirA*G3/BirA*G3 or HFD/CTR: HFD-Adipoq-BirA*G3/BirA*G3) and a significant non-adjusted p-value < 0.05 resulting in 52 SD/control and 64 HFD/control enriched serum proteins (Fig. 4C). The majority of proteins were enriched in the Adipoq-BirA*G3 samples compared to the BirA*G3 samples (Supplemental Fig. 6A- B). Adiponectin and adipsin were the two most enriched proteins in the Adipoq-BirA*G3 samples (Supplemental Fig. 6A-B). We then determined the percent of proteins containing a signal peptide. Overall, ~84% of Adipoq-BirA*G3 serum enriched proteins were predicted to contain a signal peptide (SD: 84.12%, HFD: 84.13%). Unlike the adipose tissue enriched proteins, the percentage proteins predicted to contain a signal peptide did not show a correlation with increasing log 2FC (Supplemental Fig. 6C-D). This was expected given that the adipose tissue enriched proteins contain a broader range of proteins such as ER resident proteins, membrane proteins, and extracellular matrix proteins, while serum is primarily composed of secreted proteins. Gene ontology analysis 148 of Adipoq-BirA*G3 serum enriched proteins revealed an enrichment of biological process terms related to innate immunity and humoral immune response (Supplemental Fig. 6E- F). Next, to determine serum protein enrichment between diet conditions, proteins were annotated as SD or HFD enriched by a log 2 fold change (log 2FC) >= 1.0 (SD) or =< - 1.0 (HFD) and significant non-adjusted p-value < 0.05 of SD samples over HFD samples (SD-Adipoq-BirA*G3/HFD-Adipoq-BirA*G3) (Fig. 4C). Volcano plot showed enriched proteins in SD and HFD samples (Fig. 4D). Of the 23 enriched SD proteins, adipsin (Cfd) was the most enriched, while haptoglobin (HP) was the most HFD enriched. Further, butyrylcholinesterase (BCHE) has been shown to decrease in HFD and was enriched in the SD samples 132 . Heatmap of SD and HFD enriched proteins in serum showed similar enrichment by biological replicate (Fig. 4E). Gene ontology analysis of SD enriched proteins resulted in biological process terms related to innate immunity, likely due to the number of complement factors in the enriched SD set. We next used the snRNA-seq adipocyte dataset to examine expression of genes encoding proteins either diet enriched or Adipoq-BirA*G3 enriched over BirA*G3 (Supplemental 7A-I). Genes encoding the majority of mass spectrometry identified serum proteins were expressed by adipocytes. However, there was no significant correlation of SD and HFD gene expression with the SD and HFD enriched protein set. Further, when looking at depot-specific gene expression of serum enriched proteins we observed more gene expression on some proteins in SAT and others in VAT (Supplemental Fig. 7C-D). Protein network analysis by STRINGdb of SD enriched serum proteins showed numerous known interactions with tissue expression enriched terms for brown adipose tissue (BTO:0000156) and adipose tissue (BTO:0001487) further suggesting identification of adipocyte-derived circulating proteins (Supplemental Fig. 7E). Protein network 149 analysis of HFD enriched serum proteins did not have adipose tissue term enrichment but did show phenotype term enrichment (Monarch) for abnormal lipid levels (MP:0002118) and abnormal lipid homeostasis (MP:00001547) (Supplemental Fig. 7F). Finally, to look for potential new adipocyte-derived circulating proteins, we compared Adipoq-BirA*G3 enriched proteins from both adipose tissues and serum (Fig. 5A). We identified 16 proteins enriched in all tissues (Fig. 5B). Gene ontology analysis of these 16 shared proteins resulted in biological process terms related to innate immunity, such as humoral immune response, complement activation, and B-cell mediated immunity (Fig. 5C). SnRNA-seq showed strong expression of 10 out of 16 terms. These included previously described adipocyte-derived circulating proteins: adiponectin, hephaestin, and butyrylcholinesterase. Interestingly, C4b and C6, two complement factors were expressed in the snRNA-seq data and enriched in all MS profiled tissues (adipose and serum). DISCUSSION Here, we applied a secretory compartment localized proximity labeling genetic mouse model to identify adipocyte-derived secreted protein changes in a rodent diet induced obesity model. Quantitative mass spectrometry detected a number of well-known adipocyte-derived secreted proteins in subcutaneous and visceral adipose tissue as well as in serum. Further, we identified diet induced changes in subcutaneous and visceral adipose tissues and in serum including known HFD induced changes such as an increase in leptin and FABP4 in adipose tissue. We show that adipocyte-specific recombination of the BirA*G3 allele resulted in BirA*G3 expression and co-localization with adipocyte marker FABP4 and not with other 150 non-adipocyte cell types (Fig. 1C-D) and results in significant protein biotinylation (Fig 4.2). TMT LC-MS/MS adipose tissue profiling detected multiple adipocyte-specific proteins including ADIPOQ, RETN, LEP, NAMP, MIF, adipsin (Cfd), FABP4, FBN1 (Supplemental Fig. 4.3C-F) as well as secretory pathway proteins such as ER resident proteins, membrane proteins, and secreted proteins. Further analysis of adipose tissue biotinylated proteins from SD or HFD treated mice detected diet induced changes such as HFD enriched LEP, FABP4, HP, LPL (Fig. 4.3F-G). These data show that proximity labeling detected and profiled locally secreted proteins in adipose tissue. TMT LC-MS/MS serum profiling identified numerous adipocyte-derived enriched proteins (Supplemental Fig. 4.6A-B). Among these were the adipokines, adiponectin and adipsin, and a number of proteins related to innate immunity, including C4b and C6. Both C4b and C6 were enriched in subcutaneous and visceral adipose tissue MS data as well. Previously published mouse adipocyte single-nuclear RNA-seq data shows expression of C4b and C6 in adipocyte clusters 28 . Taken together, this suggests potential adipocyte contribution to circulating C4b and C6 levels. The extent of this potential contribution warrants further studies as C4b and C6 are primarily generated by the liver. There have been, however, cases of primarily liver generated proteins being secreted by adipocytes and implicated in obesity (e.g., haptoglobin, hephaestin) 130,131 . These data strongly support the capability of proximity labeling to detect disease induced changes and profile a specific subcellular compartment. Although diet induced changes were detected in adipose tissue and serum, the number of enriched proteins to each diet was low (Fig. 4.3C, 4.4C). As a relatively new technique in vivo, robust guidelines to determine enriched protein cutoffs are lacking. Here, we utilized a significant p-value and log 2 fold change +/- 1 to determine enriched proteins between SD and HFD samples. By using a higher, albeit potentially arbitrary 151 cutoff, this may result in increased false negatives in the data, but does provide higher confidence in the proteins identified. As a field, proximity labeling data analysis would greatly benefit from better established field standards. Determining quantitative significance between two or more proximity labeling conditions remains a challenge, but as increased groups utilize this approach it is likely that these standards will arise. Toxicity related to of large-scale biotinylation of proteins via proximity labeling has been a concern of the field. Biotin (vitamin B7) is a required co-factor for several carboxylases (pyruvate carboxylase, 3-methylcrotoyl-CoA carboxylase, propionyl-CoA carboxylase, and acetyl-CoA carboxylases 1 and 2), which facilitate metabolism related to gluconeogenesis, fatty acid synthesis, and amino acid metabolism 11,12 . Previous evidence has shown proximity labeling effects on cell growth rate (in vitro) and on body size in Drosophila 6 . However, there were no observable detrimental effects of proximity labeling in this study nor seen in the characterization of the mouse model used here 11 . Although we did observe proximity labeling toxicity, we cannot rule out the effect of exogenous biotin supplementation on metabolic pathways driven by the carboxylases that require biotin as a co-factor. Biotin supplementation was done during the final 8 days of this study, which could potentially induce metabolic changes. Shorter biotin supplementation periods may avoid any potential biotin-induced changes. Future studies may wish to consider the most appropriate timeframe for biotin supplementation and see if there is a discernable change in the carboxylase activity and metabolic pathways with the onset of biotin supplementation. Surprisingly, we did not identify resistin or leptin in our Adipoq-BirA*G3 serum. In previous studies using a ubiquitously expressing (Sox2-BirA*G3) BirA*G3 mouse, resistin was identified in the serum of these mice, but not in mice with hepatocyte specific 152 labeling (Alb-BirA*G3) serum 11,50 . In our previous studies, we have not identified leptin in any of MS data. Further, streptavidin affinity purification and western blot validation has failed to detect leptin in serum but has been successful adipose tissue samples. The effect of biotinylation on individual proteins remains unknown. Although we did not see any toxicity of this model in our previous work, we did not investigate biotinylation effects on an individual protein basis 11 . Thus, as we only induced labeling by biotin supplementation for the final week of this study, we cannot rule out potential biotinylation effects on leptin’s secretion. Further, BirA*G3 biotinylates proteins on lysine residues, which can contribute to protein-detection bias. Leptin only contains 5 lysine residues, so it may be difficult to detect in serum if levels are lower compared to within adipose tissue or due to the high abundance of adiponectin and adipsin occupying binding sites during affinity purifications. Serum adiponectin removal could increase enrichment of less abundant biotinylated proteins enhancing proteomic results. Altogether, the technical challenges and lack of knowledge of biotinylation effects on specific proteins remain critical areas to consider when utilizing this approach. Proximity labeling has become a powerful tool for cell-type specific proteome profiling and subcellular compartment protein mapping. However, it has yet to be widely applied to mammalian in vivo systems. This study demonstrates biotin-based proximity labeling and quantitative proteomics in identifying disease induced cell-type specific secretion changes. We have demonstrated that this approach can provide novel biological insight to investigate pathophysiology in a cell-type specific manner. We anticipate with the recent availability of multiple genetic proximity labeling mouse models and the success in using this strain in a disease context, that this approach can be rapidly implemented by many groups. 153 MAIN FIGURES AND TABLES Fig. 4.1. Adipocyte secretory compartment protein biotinylation in mice. 154 Schematic diagram of mouse mating to generate Adipoq-BirA*G3 mice. The BirA*G3 mouse was crossed to an adiponectin-Cre (Adipoq-Cre) mouse to generate mice (Adipoq-BirA*G3) resulting in GFP cassette excision in adipocytes resulting in expression of the BirA*G3-ER and mKate2 allele (A). Western blotting of streptavidin affinity purified inguinal adipose tissue from Adipoq-BirA*G3 or BirA*G3 mice treated with biotin chow for 2 weeks. Upper: streptavidin. Lower: BirA, Adipsin, adiponectin. Input: 200𝜇g (B). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and biotinylated proteins (SA: streptavidin) in subcutaneous inguinal adipose tissue. Scale Bar: 50μm (C). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and biotinylated proteins (SA: streptavidin) in visceral retroperitoneal adipose tissue. Scale Bar: 50μm (D). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and adipocyte marker FABP4 in subcutaneous inguinal adipose tissue. Scale Bar: 50μm (D). Western blotting of streptavidin affinity purified serum from Adipoq-BirA*G3 or BirA*G3 mice treated with biotin chow for 2 weeks. Upper: streptavidin. Lower: Adipsin, adiponectin. Input: 200𝜇g (F). Each lane is a biological replicate from individual mice (n=3/genotype) (B, F). Fig. 4.2. Analysis of adipocyte derived biotinylated proteins in SD or HFD treated mice. Schematic diagram of mouse diet induced obesity model. Mice were treated with either standard diet (SD) or high fat diet (HFD) starting at 6 weeks of age for 12 weeks with biotin supplementation in water during the final week (A). Mass of mice by week treated with SD or HFD for the 12-week diet treatment (B). Western blotting of streptavidin affinity purified interscapular brown adipose tissue from Adipoq-BirA*G3 or BirA*G3 mice treated with SD or HFD for 12 weeks. Upper: streptavidin. Lower: BirA. Input: 200𝜇g (C). 155 Western blotting of streptavidin affinity purified subcutaneous inguinal adipose tissue from Adipoq- BirA*G3 or BirA*G3 mice treated with SD or HFD for 12 weeks. Upper: streptavidin. Lower: adiponectin. Input: 200𝜇g (D). Western blotting of streptavidin affinity purified visceral perirenal adipose tissue from Adipoq-BirA*G3 or BirA*G3 mice treated with SD or HFD for 12 weeks. Upper: streptavidin. Lower: adiponectin. Input: 200𝜇g (E). Streptavidin western blotting of streptavidin affinity purified inguinal adipose tissue from Adipoq-BirA*G3 or BirA*G3 mice treated with SD or HFD for 12 weeks. Input: 200𝜇g (F). Each lane is a biological replicate from individual mice (n=3/genotype) (C-F). 156 Fig. 4.3. Identification of diet induced biotinylated proteins in subcutaneous and visceral white adipose tissue. Representative schematic of TMT-based 18plex LC-MS/MS workflow for subcutaneous (SAT) inguinal 157 adipose tissue and visceral (VAT) perirenal adipose tissue from Adipoq-BirA*G3 (n=6/depot) and BirA*G3 (n=3/depot) from SD or HFD treated mice (A). PCA plot shows clustering of SAT and VAT SD, HFD Adipoq-BirA and BirA*G3 (CTR) samples (B). Number of enriched proteins identified in SD treated Adipoq- BirA*G3/BirA*G3 (SD/CTR), HFD treated Adipoq-BirA*G3/BirA*G3 (HFD/CTR), or SD or HFD enriched proteins (SD Adipoq-BirA*G3/ HFD Adipoq-BirA*G3) in SAT and VAT (C). Volcano plot of SAT significantly enriched proteins (SD: log 2 fold change >= 1.0 and p-value < 0.05; HFD: log 2 fold change <= -1.0 and p- value < 0.05) in SD and HFD (D). Volcano plot of VAT significantly enriched proteins (SD: log 2 fold change >= 1.0 and p-value < 0.05; HFD: log 2 fold change <= -1.0 and p-value < 0.05) in SD and HFD (E). Heatmap of SAT SD and HFD significantly enriched proteins log 2 values by biological replicate (F). Heatmap of VAT SD and HFD significantly enriched proteins log 2 values by biological replicate (G). Gene ontology terms for biological process (BP), cellular compartment (CC), and molecular function (MF) for SD SAT (H) and VAT (I) enriched proteins (H-I). Gene ontology terms for biological process (BP), cellular compartment (CC), and molecular function (MF) for HFD SAT (J) and VAT (K) enriched proteins (J-K). 158 Fig. 4.4. Identification of diet induced circulating adipocyte-derived biotinylated proteins. Representative schematic of TMT-based 18plex LC-MS/MS workflow for serum from Adipoq-BirA*G3 (n=4/diet) and BirA*G3 (n=2) from SD or HFD treated mice (A). PCA plot shows clustering of serum SD, HFD Adipoq-BirA and BirA*G3 (CTR) samples (B). Number of enriched proteins identified in SD treated Adipoq-BirA*G3/BirA*G3 (SD/CTR), HFD treated Adipoq-BirA*G3/BirA*G3 (HFD/CTR), or SD or HFD enriched proteins (SD Adipoq-BirA*G3/ HFD Adipoq-BirA*G3) in serum (C). Volcano plot of serum significantly enriched proteins (SD: log 2 fold change >= 1.0 and p-value < 0.05; HFD: log 2 fold change <= - 1.0 and p-value < 0.05) in SD and HFD (D). Heatmap of serum SD and HFD significantly enriched proteins log 2 values by biological replicate (E). Gene ontology terms for biological process, cellular compartment, and molecular function for SD enriched serum proteins. Gene ontology returned no results for HFD enriched serum proteins (input N was too low) (G). 159 Fig. 4.5. Identification of adipocyte-derived circulating proteins. Schematic of adipocyte-derived circulating biotinylated proteins in serum. Biotin: pink, BirA*G3: blue, Proteins: purple (A). Upset plot showing intersections and unions of enriched proteins in Adipoq-BirA*G3 (Adipoq-BirA*G3/BirA*G3) samples (B). Gene ontology terms for biological process (BP), cellular compartment (CC), and molecular function (MF) for Adipoq-BirA*G3 enriched proteins shared (n=16) between SAT, VAT, and serum (C). Dot plot of single-nuclear RNA-seq data showing gene expression of 16 shared proteins (D). 160 Supplementary Fig. 4.1. Adipocyte specific BirA*G3 expression in Adipoq-BirA*G3 mice. Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and biotinylated proteins (SA: streptavidin) in Adipoq-BirA*G3 and BirA*G3 interscapular brown adipose tissue. Scale Bar: 161 50μm (A). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and biotinylated proteins (SA: streptavidin) in Adipoq-BirA*G3 and BirA*G3 axillary brown adipose tissue. Scale Bar: 50μm (B). Immunofluorescence staining shows expression of native GFP, native mKate2, BirA*G3, and adipocyte marker FABP4 in Adipoq-BirA*G3 and BirA*G3 inguinal adipose tissue. Scale Bar: 50μm (C). Immunofluorescence staining shows expression of native GFP, native mKate2, macrophage marker F4/80, and endothelial cells (Isolecin-B4) in Adipoq-BirA*G3 and BirA*G3 inguinal adipose tissue. Boxes represent zoom images of adipocyte, macrophage, and endothelial nuclei either GFP+/mKate2- or GFP-/mKate2+ in Adipoq-BirA*G3 and BirA*G3 inguinal adipose tissue. Scale Bar: 25μm (D). 162 Supplementary Fig. 4.2. Adipocyte-derived biotinylated serum protein enrichment. Western blotting of inguinal adipose tissue (25𝜇g) from Adipoq-BirA*G3 or BirA*G3 mice treated with biotin chow for one week. Each lane is a biological replicate from individual mice (n=6/genotype). Upper: 163 streptavidin. Lower: BirA (A). Total protein stain of western blot from panel A (B). Silver staining of streptavidin affinity purified inguinal SAT from Adipoq-BirA*G3 or BirA*G3 (control) mice treated with SD or HFD for 12 weeks. Input: 500𝜇g (C). Silver staining of streptavidin affinity purified perirenal VAT from Adipoq-BirA*G3 or BirA*G3 (control) mice treated with SD or HFD for 12 weeks. Input: 500𝜇g (D). Western blotting of serum (5𝜇g) from Adipoq-BirA*G3 or BirA*G3 mice treated with SD or HFD for 12 weeks. Each lane is a biological replicate from individual mice (n=3/genotype). Upper: streptavidin. Lower: adiponectin (E). Total protein stain used for western blot normalization and quantification of total serum from Adipoq- BirA*G3 and BirA*G3 (control) mice treated with SD or HFD for 12 weeks from panel E (F). Western blot quantification (from panel A) of biotinylated serum proteins from Adipoq-BirA*G3 and BirA*G3 (control) mice treated with SD or HFD for 12 weeks (G). Western blot quantification (from panel E) of serum adiponectin from Adipoq-BirA*G3 and BirA*G3 (control) mice treated with SD or HFD for 12 weeks (H). Serum adiponectin concentration by ELISA of mice treated with SD or HFD for 12 weeks (I). Serum adiponectin concentration by ELISA of mice treated with SD or HFD for 12 weeks by sex (J). Silver staining of streptavidin affinity purified serum from Adipoq-BirA*G3, Alb-BirA*G3, Sox2-BirA*G3, or BirA*G3 (control) mice. Bead lanes are affinity purification negative control without protein input to show streptavidin contribution to bound fractions from beads. Each lane is a biological replicate from individual mice (n=3/genotype). Input: 500𝜇g (K). Silver staining of streptavidin affinity purified serum from Adipoq- BirA*G3 or BirA*G3 (control) mice using high-capacity streptavidin agarose beads. Input: 500𝜇g (L). Western blotting of streptavidin affinity purified serum using high-capacity streptavidin beads from Adipoq-BirA*G3 (Cre+) or BirA*G3 (Cre-) mice treated with SD or HFD for 12 weeks. Each lane is a biological replicate from individual mice (n=2/genotype). Upper: streptavidin. Lower: Adipsin, adiponectin (M). 164 Supplementary Fig. 4.3. Quantitative MS data analysis of subcutaneous and visceral adipose tissue. PCA plot of subcutaneous (SAT) Adipoq-BirA*G3 and BirA*G3 samples (A). PCA plot of visceral (VAT) Adipoq-BirA*G3 and BirA*G3 samples (B). Volcano plot of significantly enriched SAT proteins (SD Adipoq- 165 BirA*G3/BirA*G3) by log 2 fold change >= 1.0 and p-value < 0.05 (C). Volcano plot of significantly enriched SAT proteins (HFD Adipoq-BirA*G3/BirA*G3) by log 2 fold change >= 1.0 and p-value < 0.05 (D). Volcano plot of significantly enriched VAT proteins (SD Adipoq-BirA*G3/BirA*G3) by log 2 fold change >= 1.0 and p-value < 0.05. Known adipocyte-derived proteins are labeled (E). Volcano plot of significantly enriched VAT proteins (HFD Adipoq-BirA*G3/BirA*G3) by log 2 fold change >= 1.0 and p-value < 0.05. Known adipocyte-derived proteins are labeled (F). Dot plot of percentage of significantly enriched SAT (SD: G, HFD: H) and VAT (SD: I, HFD: J) proteins (Adipoq-BirA*G3/BirA*G3) that are predicted to have a signal peptide (G-I). Signal peptide prediction was done using signalP6.0 166 Supplementary Fig. 4.4. MS analysis of adipose tissue diet enriched proteins and single-nuclear RNA- seq validation. Gene ontology terms for biological process, cellular compartment, and molecular function for significantly enriched proteins (Adipoq-BirA*G3/BirA*G3) by log 2 fold change >= 1.0 and p-value < 0.05 in SAT (SD: A, 167 HFD: C) and VAT (SD: B, HFD: D) (A-D). Expression of SAT and VAT diet enriched proteins using publicly available mouse adipocyte single-nuclear RNA-seq from SD (chow) or HFD treated mice. SAT diet enriched protein expression in SD adipocytes compared to HFD adipocytes (SD: E, HFD: F). VAT diet enriched protein expression in SD adipocytes compared to HFD adipocytes (SD: G, HFD: H). SAT diet enriched protein expression in inguinal (SAT) compared to perigonadal (VAT) depots (SD: I, HFD: J). VAT diet enriched protein expression in inguinal (SAT) compared to perigonadal (VAT) depots (SD: K, HFD: L). Supplementary Fig. 4.5. Adipose tissue MS enriched protein validation. Western blotting of subcutaneous (SAT) inguinal adipose tissue (25𝜇g) from Adipoq-BirA*G3 or BirA*G3 mice treated with biotin chow for one week. Each lane is a biological replicate from individual mice (n=6/genotype). Upper: streptavidin. Lower: FABP4 Input (25𝜇g), FABP4 affinity purified (AP; Input: 100𝜇g) (A). Total protein stain of western blot from panel A (B). Western blot quantification of biotinylated proteins from panel A (C). Western blot quantification of FABP4 (Input 25𝜇g) from panel A (D). 168 Supplementary Fig. 4.6 Quantitative MS data analysis of adipocyte-derived circulating biotinylated proteins. Volcano plot of significantly enriched circulating biotinylated serum proteins (SD Adipoq- BirA*G3/BirA*G3) by log 2 fold change >= 1.0 and p-value < 0.05 (A). Volcano plot of significantly enriched circulating biotinylated serum proteins (HFD Adipoq-BirA*G3/BirA*G3) by log 2 fold change >= 1.0 and p- value < 0.05 (B). Dot plot of percentage of significantly enriched serum (SD: C, HFD: D) proteins (Adipoq- BirA*G3/BirA*G3) that are predicted to have a signal peptide (C-D). Signal peptide prediction was done using signalP6.0. Gene ontology terms for biological process, cellular compartment, and molecular function for significantly enriched proteins (Adipoq-BirA*G3/BirA*G3) by log 2 fold change >= 1.0 and p-value < 0.05 in serum (SD: E, HFD: F) (E-F). 169 Supplementary Fig. 4.7. Adipocyte expression of adipocyte-derived circulating serum biotinylated proteins by single-nuclear RNA-seq validation. Expression of serum enriched proteins using publicly available mouse adipocyte single-nuclear RNA-seq 170 from SD (chow) or HFD treated mice. Serum enriched proteins (SD Adipoq-BirA*G3/BirA*G3) expression in SD adipocytes compared to HFD adipocytes (A). Serum enriched proteins (HFD Adipoq- BirA*G3/BirA*G3) expression in SD adipocytes compared to HFD adipocytes (B). Serum enriched proteins (SD Adipoq-BirA*G3/BirA*G3) expression in SD adipocytes compared to HFD adipocytes (C). Serum enriched proteins (HFD Adipoq-BirA*G3/BirA*G3) expression in inguinal (SAT) adipocytes compared to perigonadal (VAT) adipocytes (D). Protein network analysis using STRINGdb of SD enriched serum proteins (E). Blue indicates tissue expression brown adipose tissue (BTO:0000156) term enrichment and red indicates tissue expression adipose tissue (BTO:0001487) term enrichment. Protein network analysis using STRINGdb of HFD enriched serum proteins (F). Blue indicates Monarch abnormal lipid homeostasis (MP:00001547) term enrichment and red indicates) Monarch abnormal lipid level (MP:0002118) term enrichment. Edge color represents protein-protein interactions: known interactions (Cyan: curated from database, Pink: experimentally determined), predicted interactions (Green: gene neighborhood, Red: gene fusions, Blue: gene co-occurrence), and others (Yellow: text mining, Black: co-expression, Pale blue: protein homology). Supplementary Table 4.1. Primary antibodies information. Primary antibodies used. WB: western blot, IF: immunofluorescence, Rb: rabbit, Ms: mouse, Sh: sheep, Gt: goat, Ch: chicken, CST: cell signaling technologies Antibody Host Supplier Catalog Number WB IF ADIPOQ Rb Abcam ab181699 1:1000 - Adipsin/Cfd Sh R&Dsystems AF5430 1:1000 - BirA Ms Abcam ab232732 1:1000 1:250 F4/80 Rt BioLegend 123102 - 1:250 FABP4 Rb Abcam ab92501 1:1000 1:250 Myc-Tag Rb Abcam ab9106 1:1000 - 171 Supplementary Table 4.2. Secondary antibodies information. Secondary antibodies used. WB: western blot, IF: immunofluorescence, Dk: donkey, Sh: sheep, Gt: goat, Rt: rat. Antibody Host Supplier Catalog Number WB IF Alexa Fluor Plus 488 anti-Chicken IgY Gt Invitrogen A32931 - 1:500 Alexa Fluor 546 anti-Chicken IgY Gt Invitrogen A-11040 - 1:500 Alexa Fluor 546 anti-Rabbit IgG (H+L) Gt Invitrogen A11035 - 1:500 Alexa Fluor 546 anti-Rat IgG Gt Invitrogen A-11081 - 1:500 Alexa Fluor 647 anti-Chicken IgY Gt Invitrogen A-21449 - 1:500 Alexa Fluor® 647 anti-Mouse IgG1 Rt BioLegend 406618 - 1:500 IRDye 680RD anti-Chicken Dk Li-Cor 926-68075 1:10,000 - IRDye 680RD anti-Goat Dk Li-Cor 926-68074 1:10,000 - IRDye 680RD anti-Mouse Gt Li-Cor 926-68072 1:10,000 - IRDye 680RD anti-Rabbit Gt Li-Cor 926-68073 1:10,000 - IRDye 680RD anti-Rat Dk Li-Cor 926-68076 1:10,000 - IRDye 800CW anti-Chicken Dk Li-Cor 926-32218 1:10,000 - IRDye 800CW anti-Goat Dk Li-Cor 926-32214 1:10,000 - IRDye 800CW anti-Mouse Gt Li-Cor 926-32212 1:10,000 - IRDye 800CW anti-Rabbit Gt Li-Cor 926-32213 1:10,000 - IRDye 800CW anti-Rat Dk Li-Cor 926-32219 1:10,000 - Alexa Fluor® Sheep-790 IgG Dk Jackson Immuno Research Labs 713-655-147 1:50,000 - IRDye 680RD Streptavidin NA Li-Cor 926-68079 1:5,000 - IRDye 800CW Streptavidin Li-Cor 926-32230 1:5,000 - 172 Chapter 5 Conclusions & Discussion Subcellular protein mapping remains a critical, yet challenging problem to study. Cellular fractionations are complex, time consuming, and lead to potential changes in protein states during processing and compartment contamination 176,177 . To overcome this, proximity labeling has been localized to subcellular compartments to efficiently tag and identify all proteins within that region 1,6,7,9 . This approach has resulted in subcellular compartment proteomes in numerous model systems and organelles 9 . However, this method had yet to be applied in mammalian genetic systems, limiting its applicability and relevance to mammalian disease and physiological models. Here, we adapted biotin- based proximity labeling 6 to generate a genetic mouse model to study cell-type specific secretomes and proteomes 11,50 . We have further applied this system to profile diet induced obesity adipocyte secretion changes. We generated a genetic proximity labeling mouse by inserting an endoplasmic reticulum localized BirA*G3 floxed allele into the Rosa26 locus 11 . Upon Cre-mediated recombination, a GFP cassette is excised resulting in BirA*G3 and an mKate2 reporter expression (Fig 3.1A). Characterization of this mouse model in major organs and tissues revealed successful biotinylation of proteins transiting through the secretory compartment. Analysis of streptavidin-affinity purified serum from these mice identified well-known and characterized secreted proteins from major tissues including adipokines adiponectin, adipsin, and resistin 203,228 (Fig. 3.4A). Further analysis of hepatocyte specific proximity labeling captured numerous hepatocyte serum secreted proteins such as AGT 201 , APOE, IGFBP2, and ALB (Fig. 3.6H). Extensive characterization of this model 173 supported success of mouse secreted protein profiling without any observed detrimental effects. By generating this genetic tool and making it readily available, we anticipate broad implementation and use to study cell-type specific secretion by many. To better understand adipocyte signaling changes in diet induced obesity we utilized adipocyte-specific proximity labeling to label and subsequently identify secreted proteins. Adipocyte-specific allele recombination resulted in BirA*G3 expression and activity in all adipocytes (Fig. 4.1A-D, Supplemental Fig. 4.1A-B). Western blotting of affinity purified proteins from adipose tissues and serum showed a strong biotinylated protein signature and labeling of known adipocyte secreted proteins, adipokines, supporting success of proximity labeling in adipocytes (Fig. 4.1B, F). We then investigated secretome changes in mice with diet induced obesity compared to lean mice (Fig. 4.2A-F). We identified lean and obese enriched proteins in subcutaneous and visceral adipose tissue and in serum. In serum, we identified known adipocyte secreted proteins such as adiponectin and adipsin 222,223 . Further, serum adipsin was decreased in obese mice, which is has been shown in rodent obesity models 240,241 . Another known adipocyte secreted protein haptoglobin, which has been shown to increase in circulation in obesity was the most obese serum enriched protein 130,131 . These studies provide key insight into in vivo mammalian proximity labeling. Through them we have shown feasibility and “proof-of-principal” in cell-type specific secretome profiling in homeostasis and disease. Despite the ability to use proximity labeling in this capacity, there have been limitations. The technical enrichment of proteins is still plagued by protein abundance biases and biases towards larger proteins and those with higher percentages of lysine residues. Computational analyses are further limited in identifying unannotated proteins. Various groups have worked to overcome these limitations but are still early on in method development 51 . While these issues pose 174 limitations on data generated using these approaches their effect will likely decrease as the proximity labeling field becomes more established and as the technique is utilized by more groups. Our work demonstrates the use of proximity labeling to capture and identify cell-type specific secretomes and proteomes in homeostasis and disease. Although proximity labeling has been adapted in numerous model systems and subcellular compartment mapping, we are among the first to generate a genetic mouse model and apply this to a disease system. We anticipate increasing groups utilizing this method to interrogate a wide range of biological questions focusing on cell-type specific proteome profiles in homeostasis, development, and disease. This will help launch the next generation of cell- type specific subcellular compartment mapping generating datasets and systems that will provide key insight into cellular and molecular biology. 175 Chapter 6 Perspective on Proximity Labeling to Map Subcellular Compartment Proteomes Protein localization and regulation is a key aspect for normal cell function. Isolation of various subcellular compartments and organelles has been traditionally extremely difficult limiting our understanding of protein localization and mapping on a global scale. Recent advances in proximity labeling techniques have allowed for organelle or compartment localized proteome profiling. Specifically, a promiscuous ligase (e.g., biotin ligase, peroxidase) can be expressed and then localized to a compartment, such as the secretory pathway, where it will then tag nearby proteins. These modified tagged proteins can then be enriched and identified by proteomics. This approach bypasses the need to isolate an organelle, often a time consuming and low-throughput process. It further allows for proteome profiling in a more native state by greatly reducing sample processing time. As such, this approach has been of interested to many. However, the data generated using this technique has fallen short of what many have hoped. As the technique is becoming more widely used, more caveats and considerations have become key aspects in the use of this approach and its ability to generate striking new biological insight. Early proximity labeling approaches were limited by the lack of highly active, non- toxic enzymes 1,3-5,9 . The generation of TurboID, a fast, non-toxic, highly promiscuous biotin ligase with labeling times as short as 10 minutes greatly increased the capabilities of in vivo proximity labeling 1,6,7,9 . As a result, there have been multiple efforts to generate in vivo proximity labeling systems. To identify organelle-specific proteomes, numerous 176 groups have generated in vitro and in vivo proximity labeling tools in model organisms including bacteria, yeast, plants, flies, worms, zebrafish, and mice 1,6,7,911,12,35,43,49-51 .Cardiac regeneration studies in zebrafish using proximity labeling identified changes in localization and abundance of Epb41l5 during regeneration and Talin2 after cardiac injury 43 . Studies in C. elegans have profiled P granule specific proteomes by proximity labeling. This then provided a curated set of proteins for an RNA interference based screen where they identified two LOTUS-domain proteins, (EGGD-1, EGGD-2) that are essential for P granule assembly 35 . To adapt proximity labeling to mammalian systems, multiple recent studies have focused on localizing promiscuous biotin ligases to the endoplasmic reticulum (ER) to identify cell-type specific secretion in mice. These studies have provided strong “proof-of-concept” in proximity labeling secretome profiling through teratoma-derived circulating proteins and transduced and genetic hepatocyte secretion profiling 11,12,13,48-51 . Hepatocyte proteins such as ALB, AGT, IGFBP2, and others were identified from liver lysates and in circulation in the blood 11,12,13,48-51 . Proof-of-concept studies using a cytoplasmic proximity labeling mouse identified brain region specific neuron and astrocyte proteome differences 12 . These differences, however, were mainly descriptive and lacked any follow-up on potential cell-type specific proteins. While the data from these studies have identified well-known secreted proteins in a cell-type specific manner, they have yet to provide strong insight into new biology. The data generated from these studies has been less than striking. However, as mammalian in vivo proximity labeling is still relatively early on in its use to profile and map subcellular compartments. The field has yet to reach a consensus on approach and analysis for these types of experiments. For instance, enrichment of biotinylated proteins may be done at the protein or peptide level and may be done using streptavidin- conjugated beads or a biotin antibody. These approaches result in differences in the depth 177 and breadth of proteins identified 11,48,52 . Generally speaking, peptide enrichment can lead to more stringent data, but comes at the cost of losing data. There also lacks a strong set of guidelines for analyzing proteomic data from these experiments. Development and implementation of reproducible, yet flexible enough standards for analysis of these datasets is still needed. It’s likely that many of these “growing pains” may be worked out as more groups utilize this approach in various systems. Further, there has been only a limited number of studies published using proximity labeling to profile and map subcellular compartments in a disease, injury, or repair state 7,35,45 . Thus, the comparison methods of multiple proteomes from the same organelle in various states are still being established. The interpretation and analysis of these data should be carefully considered due to the numerous caveats of the system. Proximity labeling approaches offer a new, exciting way to investigate protein localization and map subcellular proteomes. Despite, the many caveats of this system, it can provide easy and rapid approach to extract organelle and subcellular compartment proteomes. While little data has been shown identifying truly novel biology, it is expected that with the increased use of proximity labeling systems, both wet lab and computational approaches will improve results and refine approaches. Further, with increasing availability of genetic proximity labeling systems in multiple model systems (e.g., human cells, mice, flies, worms) we anticipate more groups utilizing these methods, which will further contribute to establishing strong field standards and growth. 178 References 1. Qin, W., Cho, K.F., Cavanagh, P.E., and Ting, A.Y. (2021). Deciphering molecular interactions by proximity labeling. Nat. Methods 18, 133–143. 2. Chapman-Smith, A., and Cronan, J.E. (1999). Molecular biology of biotin attachment to proteins. J. Nutr. 129, 477–484. 3. Martell, J.D., Deerinck, T.J., Sancak, Y., Poulos, T.L., Mootha, V.K., Sosinsky, G.E., Ellisman, M.H., and Ting, A.Y. (2012). Engineered ascorbate peroxidase as a genetically encoded reporter for electron microscopy. Nat. Biotechnol. 30, 1143–1148. 4. Choi-Rhee, E., Schulman, H., and Cronan, J.E. (2008). Promiscuous protein biotinylation by Escherichia coli biotin protein ligase. Protein Sci. 13, 3043–3050. 5. Cronan, J.E. (2005). Targeted and proximity-dependent promiscuous protein biotinylation by a mutant Escherichia coli biotin protein ligase. J. Nutr. Biochem. 16, 416–418. 6. Branon, T.C., Bosch, J.A., Sanchez, A.D., Udeshi, N.D., Svinkina, T., Carr, S.A., Feldman, J.L., Perrimon, N., and Ting, A.Y. (2018). Efficient proximity labeling in living cells and organisms with TurboID. Nat. Biotechnol. 36, 880–898. 7. Cho, K.F., Branon, T.C., Udeshi, N.D., Myers, S.A., Carr, S.A., and Ting, A.Y. (2020). Proximity labeling in mammalian cells with TurboID and split-TurboID. Nat. Protoc. 15, 3971–3999. 8. May, D.G., Scott, K.L., Campos, A.R., and Roux, K.J. (2020). Comparative Application of BioID and TurboID for Protein-Proximity Biotinylation. Cells 9. 9. Bosch, J.A., Chen, C.-L., and Perrimon, N. (2021). Proximity-dependent labeling methods for proteomic profiling in living cells: An update. Wiley Interdiscip. Rev. Dev. Biol. 10, e392. 10. Liu, J., Jang, J.Y., Pirooznia, M., Liu, S., and Finkel, T. (2021). The secretome mouse provides a genetic platform to delineate tissue-specific in vivo secretion. Proc. Natl. Acad. Sci. U. S. A. 118. 11. Yang, R., Meyer, A.S., Droujinine, I.A., Udeshi, N.D., Hu, Y., Guo, J., McMahon, J.A., Carey, D.K., Xu, C., Fang, Q., et al. (2022). A genetic model for in vivo proximity labelling of the mammalian secretome. Open Biol. 12. 12. Rayaprolu, S., Bitarafan, S., Santiago, J. V., Betarbet, R., Sunna, S., Cheng, L., Xiao, H., Nelson, R.S., Kumar, P., Bagchi, P., et al. (2022). Cell type-specific biotin labeling in vivo resolves regional neuronal and astrocyte proteomic differences in mouse brain. Nat. Commun. 13, 1–18. 179 13. Kido, K., Yamanaka, S., Nakano, S., Motani, K., Shinohara, S., Nozawa, A., Kosako, H., Ito, S., and Sawasaki, T. (2020). AirID, a novel proximity biotinylation enzyme, for analysis of protein–protein interactions. Elife 9, 1–24. 14. Johnson, B.S., Chafin, L., Farkas, D., Adair, J., Elhance, A., Farkas, L., Bednash, J.S., and Londino, J.D. (2022). MicroID2: A Novel Biotin Ligase Enables Rapid Proximity-Dependent Proteomics. Mol. Cell. Proteomics 21, 0–12. 15. Kubitz, L., Bitsch, S., Zhao, X., Schmitt, K., Deweid, L., Roehrig, A., Barazzone, E.C., Valerius, O., Kolmar, H., and Béthune, J. (2022). Engineering of ultraID, a compact and hyperactive enzyme for proximity-dependent biotinylation in living cells. Commun. Biol. 5. 16. Kim, D.I., Jensen, S.C., Noble, K.A., Kc, B., Roux, K.H., Motamedchaboki, K., and Roux, K.J. (2016). An improved smaller biotin ligase for BioID proximity labeling. Mol. Biol. Cell 27, 1188–1196. 17. Roux, K.J., Kim, D.I., Raida, M., and Burke, B. (2012). A promiscuous biotin ligase fusion protein identifies proximal and interacting proteins in mammalian cells. J. Cell Biol. 196, 801–810. 18. Vandemoortele, G., De Sutter, D., Moliere, A., Pauwels, J., Gevaert, K., and Eyckerman, S. (2019). A Well-Controlled BioID Design for Endogenous Bait Proteins. J. Proteome Res. 18, 95–106. 19. Dingar, D., Kalkat, M., Chan, P.K., Srikumar, T., Bailey, S.D., Tu, W.B., Coyaud, E., Ponzielli, R., Kolyar, M., Jurisica, I., et al. (2015). BioID identifies novel c-MYC interacting partners in cultured cells and xenograft tumors. J. Proteomics 118, 95–111. 20. Zhang, Y., Song, G., Lal, N.K., Nagalakshmi, U., Li, Y., Zheng, W., Huang, P., Branon, T.C., Ting, A.Y., Walley, J.W., et al. (2019). TurboID-based proximity labeling reveals receptor- mediated immunity. Nat. Commun. 10, 3252. 21. Van Itallie, C.M., Tietgens, A.J., Aponte, A., Fredriksson, K., Fanning, A.S., Gucek, M., and Anderson, J.M. (2014). Biotin ligase tagging identifies proteins proximal to E-cadherin, including lipoma preferred partner, a regulator of epithelial cell-cell and cell-substrate adhesion. J. Cell Sci. 127, 885–895. 22. Xiong, Z., Lo, H.P., McMahon, K.A., Martel, N., Jones, A., Hill, M.M., Parton, R.G., and Hall, T.E. (2021). In vivo proteomic mapping through gfp-directed proximity-dependent biotin labelling in zebrafish. Elife 10, 1–21. 23. Schopp, I.M., Amaya Ramirez, C.C., Debeljak, J., Kreibich, E., Skribbe, M., Wild, K., and Béthune, J. (2017). Split-BioID a conditional proteomics approach to monitor the composition of spatiotemporally defined protein complexes. Nat. Commun. 8. 180 24. De Munter, S., Görnemann, J., Derua, R., Lesage, B., Qian, J., Heroes, E., Waelkens, E., Van Eynde, A., Beullens, M., and Bollen, M. (2017). Split-BioID: a proximity biotinylation assay for dimerization-dependent protein interactions. FEBS Lett. 591, 415–424. 25. Feng, W., Liu, C., Spinozzi, S., Wang, L., Evans, S.M., and Chen, J. (2020). Identifying the cardiac dyad proteome in vivo by a BioID2 knock-in strategy. Circulation 940–942. 26. Han, Y., Branon, T.C., Martell, J.D., Boassa, D., Shechner, D., Ellisman, M.H., and Ting, A. (2019). Directed Evolution of Split APEX2 Peroxidase. ACS Chem. Biol. 14, 619–635. 27. Cho, K.F., Branon, T.C., Rajeev, S., Svinkina, T., Udeshi, N.D., Thoudam, T., Kwak, C., Rhee, H.W., Lee, I.K., Carr, S.A., et al. (2020). Split-TurboID enables contact-dependent proximity labeling in cells. Proc. Natl. Acad. Sci. U. S. A. 117. 28. Rhee, H.W., Zou, P., Udeshi, N.D., Martell, J.D., Mootha, V.K., Carr, S.A., and Ting, A.Y. (2013). Proteomic mapping of mitochondria in living cells via spatially restricted enzymatic tagging. Science (80). 339, 1328–1331. 29. Hung, V., Udeshi, N.D., Lam, S.S., Loh, K.H., Cox, K.J., Pedram, K., Carr, S.A., and Ting, A.Y. (2016). Spatially resolved proteomic mapping in living cells with the engineered peroxidase APEX2. Nat. Protoc. 11, 456–475. 30. Kim, D.I., Birendra, K.C., Zhu, W., Motamedchaboki, K., Doye, V., and Roux, K.J. (2014). Probing nuclear pore complex architecture with proximity-dependent biotinylation. Proc. Natl. Acad. Sci. U. S. A. 111, 2453–2461. 31. Mick, D.U., Rodrigues, R.B., Leib, R.D., Adams, C.M., Chien, A.S., Gygi, S.P., and Nachury, M. V. (2015). Proteomics of Primary Cilia by Proximity Labeling. Dev. Cell 35, 497–512. 32. Gingras, A.C., Abe, K.T., and Raught, B. (2019). Getting to know the neighborhood: using proximity-dependent biotinylation to characterize protein complexes and map organelles. Curr. Opin. Chem. Biol. 48, 44–54. 33. Martell, J.D., Yamagata, M., Deerinck, T.J., Phan, S., Kwa, C.G., Ellisman, M.H., Sanes, J.R., and Ting, A.Y. (2016). A split horseradish peroxidase for the detection of intercellular protein-protein interactions and sensitive visualization of synapses. Nat. Biotechnol. 34, 774–780. 34. Youn, J.Y., Dunham, W.H., Hong, S.J., Knight, J.D.R., Bashkurov, M., Chen, G.I., Bagci, H., Rathod, B., MacLeod, G., Eng, S.W.M., et al. (2018). High-Density Proximity Mapping Reveals the Subcellular Organization of mRNA-Associated Granules and Bodies. Mol. Cell 69, 517-532.e11. 181 35. Price, I.F., Hertz, H.L., Pastore, B., Wagner, J., and Tang, W. (2021). Proximity labeling identifies lotus domain proteins that promote the formation of perinuclear germ granules in c. Elegans. Elife 10, 1–34. 36. Kwak, C., Shin, S., Park, J.S., Jung, M., My Nhung, T.T., Kang, M.G., Lee, C., Kwon, T.H., Park, S.K., Mun, J.Y., et al. (2020). Contact-ID, a tool for profiling organelle contact sites, reveals regulatory proteins of mitochondrial-associated membrane formation. Proc. Natl. Acad. Sci. U. S. A. 117. 37. Hung, V., Lam, S.S., Udeshi, N.D., Svinkina, T., Guzman, G., Mootha, V.K., Carr, S.A., and Ting, A.Y. (2019). Erratum: Correction: Proteomic mapping of cytosol-facing outer mitochondrial and ER membranes in living human cells by proximity biotinylation (eLife (2017) 6 PII: e50707). Elife 8, 1–39. 38. Han, S., Udeshi, N.D., Deerinck, T.J., Svinkina, T., Ellisman, M.H., Carr, S.A., and Ting, A.Y. (2017). Proximity Biotinylation as a Method for Mapping Proteins Associated with mtDNA in Living Cells. Cell Chem. Biol. 24, 404–414. 39. Spence, E.F., Dube, S., Uezu, A., Locke, M., Soderblom, E.J., and Soderling, S.H. (2019). In vivo proximity proteomics of nascent synapses reveals a novel regulator of cytoskeleton- mediated synaptic maturation. Nat. Commun. 10, 386. 40. Cho, I.T., Adelmant, G., Lim, Y., Marto, J.A., Cho, G., and Golden, J.A. (2017). Ascorbate peroxidase proximity labeling coupled with biochemical fractionation identifies promoters of endoplasmic reticulum–mitochondrial contacts. J. Biol. Chem. 292, 16382–16392. 41. Chan, C.J., Le, R., Burns, K., Ahmed, K., Coyaud, E., Laurent, E.M.N., Raught, B., and Melançon, P. (2019). BioID performed on golgi enriched fractions identify C10orf76 as a GBF1 binding protein essential for golgi maintenance and secretion. Mol. Cell. Proteomics 18, 2285–2297. 42. Li, J., Han, S., Li, H., Udeshi, N.D., Svinkina, T., Mani, D.R., Xu, C., Guajardo, R., Xie, Q., Li, T., et al. (2020). Cell-Surface Proteomic Profiling in the Fly Brain Uncovers Wiring Regulators. Cell 180, 373-386.e15. 43. Pronobis, M.I., Zheng, S., Singh, S.P., Goldman, J.A., and Poss, K.D. (2021). In vivo proximity labeling identifies cardiomyocyte protein networks during zebrafish heart regeneration. Elife 10, 1–22. 44. Lin, Q., Zhou, Z., Luo, W., Fang, M., Li, M., and Li, H. (2017). Screening of proximal and interacting proteins in rice protoplasts by proximity-dependent biotinylation. Front. Plant Sci. 8, 1–10. 45. Khan, M., Youn, J.Y., Gingras, A.C., Subramaniam, R., and Desveaux, D. (2018). In planta proximity dependent biotin identification (BioID). Sci. Rep. 8, 2–9. 182 46. Conlan, B., Stoll, T., Gorman, J.J., Saur, I., and Rathjen, J.P. (2018). Development of a Rapid in planta BioID System as a Probe for Plasma Membrane-Associated Immunity Proteins. Front. Plant Sci. 9, 1882. 47. Opitz, N., Schmitt, K., Hofer-Pretz, V., Neumann, B., Krebber, H., Braus, G.H., and Valerius, O. (2017). Capturing the Asc1p/Receptor for Activated C Kinase 1 (RACK1) Microenvironment at the head region of the 40s ribosome with quantitative BioID in Yeast. Mol. Cell. Proteomics 16, 2199–2218. 48. Kim, K. eun, Park, I., Kim, J., Kang, M.G., Choi, W.G., Shin, H., Kim, J.S., Rhee, H.W., and Suh, J.M. (2021). Dynamic tracking and identification of tissue-specific secretory proteins in the circulation of live mice. Nat. Commun. 12, 1–9. 49. Wei, W., Riley, N.M., Yang, A.C., Kim, J.T., Terrell, S.M., Li, V.L., Garcia-Contreras, M., Bertozzi, C.R., and Long, J.Z. (2021). Cell type-selective secretome profiling in vivo. Nat. Chem. Biol. 17, 326–334. 50. Droujinine, I.A., Meyer, A.S., Wang, D., Udeshi, N.D., Hu, Y., Rocco, D., McMahon, J.A., Yang, R., Guo, J.J., Mu, L., et al. (2021). Proteomics of protein trafficking by in vivo tissue- specific labeling. Nat. Commun. 12. 51. Na, Z., Dai, X., Zheng, S.J., Bryant, C.J., Loh, K.H., Su, H., Luo, Y., Buhagiar, A.F., Cao, X., Baserga, S.J., et al. (2022). Mapping subcellular localizations of unannotated microproteins and alternative proteins with MicroID. Mol. Cell 82, 2900-2911.e7. 52. Udeshi, N.D., Pedram, K., Svinkina, T., Fereshetian, S., Myers, S.A., Aygun, O., Krug, K., Clauser, K., Ryan, D., Ast, T., et al. (2017). Antibodies to biotin enable large-scale detection of biotinylation sites on proteins. Nat. Methods 14, 1167–1170. 53. Apweiler, R., Bairoch, A., Wu, C.H., Barker, W.C., Boeckmann, B., Ferro, S., Gasteiger, E., Huang, H., Lopez, R., Magrane, M., et al. (2004). UniProt: The universal protein knowledgebase. Nucleic Acids Res. 32. 54. N. Michael Green, E. John Toms; The properties of subunits of avidin coupled to Sepharose. Biochem J 1 August 1973; 133 (4): 687–698. 55. Henrikson, K.P., Allen, S.H.G., and Maloy, W.L. (1979). An avidin monomer affinity column for the purification of biotin-containing enzymes. Anal. Biochem. 94, 366–370. 56. Qureshi, M.H., and Wong, S.L. (2002). Design, production, and characterization of a monomeric streptavidin and its application for affinity purification of biotinylated proteins. Protein Expr. Purif. 25, 409–415. 57. Rafiee, M., Sigismondo, G., Kalxdorf, M., Förster, L., Brügger, B., Béthune, J., and Krijgsveld, J. (2020). Protease-resistant streptavidin for interaction proteomics. Mol. Syst. Biol. 16, 1–12. 183 58. Aguilar, M., Bhuket, T., Torres, S., Liu, B., and Wong, R.J. (2015). Prevalence of the metabolic syndrome in the United States, 2003-2012. Jama 313, 1973–1974. 59. T. Leff., J. G. Ganneman. (Eds). (2010). Adipose Tissue in Health and Disease. Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. 60. Organization, W.H. Obesity and overweight. (2020) 61. Chouchani, E.T., and Kajimura, S. (2019). Metabolic adaptation and maladaptation in adipose tissue. Nat. Metab. 1, 189–200. 62. Pirzgalska, R.M., Seixas, E., Seidman, J.S., Link, V.M., Sánchez, N.M., Mahú, I., Mendes, R., Gres, V., Kubasova, N., Morris, I., et al. (2017). Sympathetic neuron-associated macrophages contribute to obesity by importing and metabolizing norepinephrine. Nat. Med. 23, 1309–1318. 63. Zeng, W., Pirzgalska, R.M., Pereira, M.M.A., Kubasova, N., Barateiro, A., Seixas, E., Lu, Y.- H., Kozlova, A., Voss, H., Martins, G.G., et al. (2015). Sympathetic Neuro-adipose Connections Mediate Leptin-Driven Lipolysis. Cell 163, 84–94. 64. Ghaben, A.L., and Scherer, P.E. (2019). Adipogenesis and metabolic health. Nat. Rev. Mol. Cell Biol. 20, 242–258. 65. Hall, J.E., do Carmo, J.M., da Silva, A.A., Wang, Z., and Hall, M.E. (2019). Obesity, kidney dysfunction and hypertension: mechanistic links. Nat. Rev. Nephrol. 15, 367–385. 66. Kahn, B.B. (2019). Adipose tissue, inter-organ communication, and the path to type 2 diabetes: The 2016 banting medal for scientific achievement lecture. Diabetes 68, 3–14. 67. Frayn, K.N., Karpe, F., Fielding, B.A., Macdonald, I.A., and Coppack, S.W. (2003). Integrative physiology of human adipose tissue. Int. J. Obes. 27, 875–888. 68. Kajimura, S. (2017). Advances in the understanding of adipose tissue biology. Nat. Rev. Endocrinol. 13, 69–70. 69. Villarroya, F., Cereijo, R., Villarroya, J., and Giralt, M. (2017). Brown adipose tissue as a secretory organ. Nat. Rev. Endocrinol. 13, 26–35. 70. Fabbiano, S., Suárez-Zamorano, N., Rigo, D., Veyrat-Durebex, C., Stevanovic Dokic, A., Colin, D.J., and Trajkovski, M. (2016). Caloric Restriction Leads to Browning of White Adipose Tissue through Type 2 Immune Signaling. Cell Metab. 24, 434–446. 71. Coolbaugh, C.L., Damon, B.M., Bush, E.C., Welch, E.B., and Towse, T.F. (2019). Cold exposure induces dynamic, heterogeneous alterations in human brown adipose tissue lipid content. Sci. Rep. 9, 1-13 184 72. Sonne, S.B., Yadav, R., Yin, G., Dalgaard, M.D., Myrmel, L.S., Gupta, R., Wang, J., Madsen, L., Kajimura, S., and Kristiansen, K. (2017). Obesity is associated with depot-specific alterations in adipocyte DNA methylation and gene expression. Adipocyte 6, 124–133 73. Droujinine, I. A. & Perrimon, N. Interorgan Communication Pathways in Physiology: Focus on Drosophila. Annu Rev Genet 50, 539-570 (2016). 74. Zhang, M. & Schekman, R. Unconventional secretion, unconventional solutions. Science 340, 559-561 (2013). 75. Yi, P., Park, J.-S. & Melton, D. A. Betatrophin: a hormone that controls pancreatic β cell proliferation. Cell 153, 747-758 (2013). 76. Yi, P., Park, J.-S. & Melton, D. A. Retraction notice to: betatrophin: a hormone that controls pancreatic β cell proliferation. Cell 168, 326 (2017). 77. Egerman, M. A. et al. GDF11 increases with age and inhibits skeletal muscle regeneration. Cell Metab 22, 164-174 (2015). 78. Gillette, M. A. & Carr, S. A. Quantitative analysis of peptides and proteins in biomedicine by targeted mass spectrometry. Nat Methods 10, 28-34 (2013). 79. Lim, J.-M. et al. Defining the regulated secreted proteome of rodent adipocytes upon the induction of insulin resistance. J Proteome Res 7, 1251-1263 (2008). 80. Park, S. et al. A genetic strategy to measure circulating Drosophila insulin reveals genes regulating insulin production and secretion. PLoS Genet 10, e1004555 (2014). 81. Cao, C. & Brown, M. R. Localization of an insulin-like peptide in brains of two flies. Cell Tissue Res 304, 317-321 (2001). 82. Dreyer, A. P. et al. A circadian output center controlling feeding: fasting rhythms in Drosophila. PLoS Genet 15 (2019). 83. Cognigni, P., Bailey, A. P. & Miguel-Aliaga, I. Enteric neurons and systemic signals couple nutritional and reproductive status with intestinal homeostasis. Cell Metab 13, 92-104 (2011). 84. Chintapalli, V. R., Wang, J. & Dow, J. A. Using FlyAtlas to identify better Drosophila melanogaster models of human disease. Nat Genet 39, 715-720 (2007). 85. Fogerty, F. J. et al. Tiggrin, a novel Drosophila extracellular matrix protein that functions as a ligand for Drosophila alpha PS2 beta PS integrins. Development 120, 1747-1758 (1994). 86. Bunch, T. A. et al. The PS2 integrin ligand tiggrin is required for proper muscle function in Drosophila. Development 125, 1679-1689 (1998). 87. Green, N. et al. A Common Suite of Coagulation Proteins Function in Drosophila Muscle Attachment. Genetics 204, 1075-1087 (2016). 88. Palm, W. et al. Lipoproteins in Drosophila melanogaster—assembly, function, and influence on tissue lipid composition. PLoS Genet 8, e1002828 (2012). 185 89. Rodríguez-Vázquez, M., Vaquero, D., Parra-Peralbo, E., Mejía-Morales, J. E. & Culi, J. Drosophila lipophorin receptors recruit the lipoprotein LTP to the plasma membrane to mediate lipid uptake. PLoS Genet 11, e1005356 (2015). 90. Drechsler, M., Schmidt, A. C., Meyer, H. & Paululat, A. The conserved ADAMTS-like protein lonely heart mediates matrix formation and cardiac tissue integrity. PLoS Genet 9, e1003616 (2013). 91. Hu, Y. et al. An integrative approach to ortholog prediction for disease-focused and other functional studies. BMC Bioinformatics 12, 357 (2011). 92. Katewa, S. D. et al. Intramyocellular fatty-acid metabolism plays a critical role in mediating responses to dietary restriction in Drosophila melanogaster. Cell Metab 16, 97-103 (2012). 93. Banerjee, K. K., Ayyub, C., Sengupta, S. & Kolthur-Seetharam, U. Fat body dSir2 regulates muscle mitochondrial physiology and energy homeostasis nonautonomously and mimics the autonomous functions of dSir2 in muscles. Mol Cell Biol 33, 252-264 (2013). 94. Arrese, E. L. & Soulages, J. L. Insect fat body: energy, metabolism, and regulation. Annu Rev Entomol 55, 207-225 (2010). 95. Demontis, F. & Perrimon, N. FOXO/4E-BP signaling in Drosophila muscles regulates organism-wide proteostasis during aging. Cell 143, 813-825 (2010). 96. Grundmann, U., Römisch, J., Siebold, B., Bohn, H. & Amann, E. Cloning and expression of a cDNA encoding human placental protein 11, a putative serine protease with diagnostic significance as a tumor marker. DNA Cell Biol 9, 243-250 (1990). 97. Laneve, P. et al. The tumor marker human placental protein 11 is an endoribonuclease. J Biol Chem 283, 34712-34719 (2008). 98. Poe, J. C. et al. EndoU is a novel regulator of AICD during peripheral B cell selection. J Exp Med 211, 57-69 (2014). 99. Solter, D. in Manipulating the mouse embryo (eds Beddington Hogan & Lacy Constantini) (Cold Spring Harbor Press, 1994). 100. Solter, D. From teratocarcinomas to embryonic stem cells and beyond: a history of embryonic stem cell research. Nat Rev Genet 7, 319-327 (2006). 101. Howarth, M. et al. Monovalent, reduced-size quantum dots for imaging receptors on living cells. Nat Methods 5, 397-399 (2008). 102. Iwaki, T. & Castellino, F. J. A single plasmid transfection that offers a significant advantage associated with puromycin selection in Drosophila Schneider S2 cells expressing heterologous proteins. Cytotechnology 57, 45-49 (2008). 103. Katzen, F. Gateway® recombinational cloning: a biological operating system. Expert Opin Drug Discov 2, 571-589 (2007). 104. Perkins, L. A. et al. The transgenic RNAi project at Harvard Medical School: resources and validation. Genetics 201, 843-852 (2015). 186 105. Kwon, Y. et al. Systemic organ wasting induced by localized expression of the secreted insulin/IGF antagonist ImpL2. Dev Cell 33, 36-46 (2015). 106. Rajan, A. & Perrimon, N. Drosophila cytokine unpaired 2 regulates physiological homeostasis by remotely controlling insulin secretion. Cell 151, 123-137 (2012). 107. Emery, P. Protein extraction from Drosophila heads. Methods Mol Biol, 375-377 (2007). 108. Ericsson, C. 2-D protein extracts from Drosophila melanogaster. Methods Mol Biol, 35-41 (1999). 109. Jensen, K. et al. Purification of transcripts and metabolites from Drosophila heads. J Vis Exp (2013). 110. Soler, C., Daczewska, M., Da Ponte, J. P., Dastugue, B. & Jagla, K. Coordinated development of muscles and tendons of the Drosophila leg. Development 131, 6041-6051 (2004). 111. Hartenstein, V. Atlas of Drosophila development. Vol. 328 (Cold Spring Harbor Laboratory Press, 1993). 112. Demontis, F., Patel, V. K., Swindell, W. R. & Perrimon, N. Intertissue control of the nucleolus via a myokine-dependent longevity pathway. Cell Rep 7, 1481-1494 (2014). 113. Krupp, J. J. & Levine, J. D. Dissection of oenocytes from adult Drosophila melanogaster. J Vis Exp, e2242 (2010). 114. Lung, O. & Wolfner, M. F. Drosophila seminal fluid proteins enter the circulatory system of the mated female fly by crossing the posterior vaginal wall. Insect Biochem Mol Biol 29, 1043-1052 (1999). 115. Calleja, M. et al. How to pattern an epithelium: lessons from achaete-scute regulation on the notum of Drosophila. Gene 292, 1-12 (2002). 116. Gilbert, S. F. Developmental biology. (2006). 117. Chakravarti, B., Fathy, P., Sindicich, M., Mallik, B. & Chakravarti, D. N. Comparison of SYPRO Ruby and Flamingo fluorescent stains for application in proteomic research. Anal Biochem 398, 1-6 (2010). 118. Chen, C.-L. et al. Proteomic mapping in live Drosophila tissues using an engineered ascorbate peroxidase. Proc Natl Acad Sci 112, 12093-12098 (2015). 119. Goetzenich, A. et al. The role of hypoxia-inducible factor-1α and vascular endothelial growth factor in late-phase preconditioning with xenon, isoflurane and levosimendan in rat cardiomyocytes. Interact Cardiovasc Thorac Surg 18, 321-328 (2014). 120. Liabakk, N.-B., Nustad, K. & Espevik, T. A rapid and sensitive immunoassay for tumor necrosis factor using magnetic monodisperse polymer particles. J Immunol Methods 134, 253-259 (1990). 187 121. Hopp, T. P. et al. A short polypeptide marker sequence useful for recombinant protein identification and purification. Biotechnology 6, 1204-1210 (1988). 122. Wang, M., et al. PaxDb, a database of protein abundance averages across all three domains of life. Mol Cell Proteomics 11, 492-50 (2012). 123. Ben-Shlomo, I., Hsu, S. Y., Rauch, R., Kowalski, H. W. & Hsueh, A. J. Signaling receptome: a genomic and evolutionary perspective of plasma membrane receptors involved in signal transduction. Sci STKE 2003, re9-re9 (2003). 124. Farrah, T. et al. A high-confidence human plasma proteome reference set with estimated concentrations in PeptideAtlas. Mol Cell Proteomics 10, M110. 006353 (2011). 125. Liu, T. et al. High dynamic range characterization of the trauma patient plasma proteome. Mol Cell Proteomics 5, 1899-1913 (2006). 126. Petersen, T. N., Brunak, S., von Heijne, G. & Nielsen, H. SignalP 4.0: discriminating signal peptides from transmembrane regions. Nat Methods 8, 785-786 (2011). 127. Krogh, A., Larsson, B., Von Heijne, G. & Sonnhammer, E. L. Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. J Mol Biol 305, 567-580 (2001). 128. Armenteros, J. J. A. et al. SignalP 5.0 improves signal peptide predictions using deep neural networks. Nat Biotechnol 37, 420-423 (2019). 129. Bendtsen, J. D., Jensen, L. J., Blom, N., Von Heijne, G. & Brunak, S. Feature-based prediction of non-classical and leaderless protein secretion. Protein Eng Des Sel 17, 349-356 (2004). 130. Gamucci, O., Lisi, S., Scabia, G., Marchi, M., Piaggi, P., Duranti, E., Virdis, A., Pinchera, A., Santini, F., and Maffei, M. (2012). Haptoglobin deficiency determines changes in adipocyte size and adipogenesis. Adipocyte 1, 142–183. 131. Chiellini, C., Bertacca, A., Novelli, S.E., Görgün, C.Z., Ciccarone, A., Giordano, A., Xu, H., Soukas, A., Costa, M., Gandini, D., et al. (2002). Obesity modulates the expression of haptoglobin in the white adipose tissue via TNF-alpha. J. Cell. Physiol. 190, 251–258. 132. Chen, V.P., Gao, Y., Geng, L., Stout, M.B., Jensen, M.D., and Brimijoin, S. (2016). Butyrylcholinesterase Deficiency Promotes Adipose Tissue Growth and Hepatic Lipid Accumulation in Male Mice on High-Fat Diet. Endocrinology 157, 3086–3095. 133. Brown, G. R. et al. Gene: a gene-centered information resource at NCBI. Nucleic Acids Res 43, D36-D42 (2014). 134. Alvarez-Llamas, G. et al. Characterization of the human visceral adipose tissue secretome. Mol Cell Proteomics 6, 589-600 (2007). 135. Chen, X., Cushman, S. W., Pannell, L. K. & Hess, S. Quantitative Proteomic Analysis of the Secretory Proteins from Rat Adipose Cells Using a 2D Liquid Chromatography− MS/MS Approach. J Proteome Res 4, 570-577 (2005). 188 136. Lehr, S. et al. Identification and validation of novel adipokines released from primary human adipocytes. Mol Cell Proteomics 11, M111. 010504 (2012). 137. Molina, H. et al. Temporal profiling of the adipocyte proteome during differentiation using a five-plex SILAC based strategy. J Proteome Res 8, 48-58 (2008). 138. Rosenow, A. et al. Identification of novel human adipocyte secreted proteins by using SGBS cells. J Proteome Res 9, 5389-5401 (2010). 139. Wu, L. E. et al. Identification of fatty acid binding protein 4 as an adipokine that regulates insulin secretion during obesity. Mol Metab 3, 465-473 (2014). 140. Zhong, J. et al. Temporal profiling of the secretome during adipogenesis in humans. J Proteome Res 9, 5228-5238 (2010). 141. Zhou, H. et al. Quantitative analysis of secretome from adipocytes regulated by insulin. Acta Biochim Biophys Sin 41, 910-921 (2009). 142. Zvonic, S. et al. Secretome of Primary Cultures of Human Adipose-derived Stem Cells Modulation of Serpins by Adipogenesis. Mol Cell Proteomics 6, 18-28 (2007). 143. Bortoluzzi, S., Scannapieco, P., Cestaro, A., Danieli, G. A. & Schiaffino, S. Computational reconstruction of the human skeletal muscle secretome. Proteins 62, 776-792 (2006). 144. Chan, C. X. a. et al. Identification of differentially regulated secretome components during skeletal myogenesis. Mol Cell Proteomics 10, M110. 004804 (2011). 145. Deshmukh, A. S., Cox, J., Jensen, L. J., Meissner, F. & Mann, M. Secretome analysis of lipid-induced insulin resistance in skeletal muscle cells by a combined experimental and bioinformatics workflow. J Proteome Res 14, 4885-4895 (2015). 146. Hartwig, S. et al. Secretome profiling of primary human skeletal muscle cells. Biochim Biophys Acta 1844, 1011-1017 (2014). 147. Hittel, D. S., Berggren, J. R., Shearer, J., Boyle, K. & Houmard, J. A. Increased secretion and expression of myostatin in skeletal muscle from extremely obese women. Diabetes 58, 30-38 (2009). 148. Norheim, F. et al. Proteomic identification of secreted proteins from human skeletal muscle cells and expression in response to strength training. Am J Physiol 301, E1013- E1021 (2011). 149. Raschke, S., Eckardt, K., Holven, K. B., Jensen, J. & Eckel, J. Identification and validation of novel contraction-regulated myokines released from primary human skeletal muscle cells. PLoS ONE 8, e62008 (2013). 150. Yoon, J. H. et al. Proteomic analysis of the palmitate-induced myotube secretome reveals involvement of the annexin a1-formyl peptide receptor 2 (fpr2) pathway in insulin resistance. Mol Cell Proteomics 14, 882-892 (2015). 189 151. Brunner, E. et al. A high-quality catalog of the Drosophila melanogaster proteome. Nat Biotechnol 25, 576 (2007). 152. Yuan, M., Breitkopf, S. B., Yang, X. & Asara, J. M. A positive/negative ion–switching, targeted mass spectrometry–based metabolomics platform for bodily fluids, cells, and fresh and fixed tissue. Nat Protoc 7, 872 (2012). 153. Martínez-Maqueda, D., Hernández-Ledesma, B., Amigo, L., Miralles, B. & Gómez-Ruiz, J. Á. in Proteomics in Foods 21-50 (Springer, 2013). 154. Feist, P. & Hummon, A. B. Proteomic challenges: sample preparation techniques for microgram-quantity protein analysis from biological samples. Int J Mol Sci 16, 3537-3563 (2015). 155. Tirumalai, R. S. et al. Characterization of the low molecular weight human serum proteome. Mol Cell Proteomics 2, 1096-1103 (2003). 156. Villén, J. & Gygi, S. P. The SCX/IMAC enrichment approach for global phosphorylation analysis by mass spectrometry. Nat Protoc 3, 1630 (2008). 157. Zhou, Y., Aebersold, R. & Zhang, H. Isolation of N-linked glycopeptides from plasma. Anal Chem 79, 5826-5837 (2007). 158. Zhang, H. Glycoproteomics using chemical immobilization. Curr Protoc Protein Sci, 24.23. 21-24.23. 14 (2007). 159. Baycin-Hizal, D. et al. GlycoFly: a database of Drosophila N-linked glycoproteins identified using SPEG–MS techniques. J Proteome Res 10, 2777-2784 (2011). 160. Tian, Y., Zhou, Y., Elliot, S., Aebersold, R. & Zhang, H. Solid-phase extraction of N-linked glycopeptides. Nat Protoc 2, 334 (2007). 161. Parkhitko, A. A. et al. Tissue-specific down-regulation of S-adenosyl-homocysteine via suppression of dAhcyL1/dAhcyL2 extends health span and life span in Drosophila. Genes Dev 30, 1409-1422 (2016). 162. Riedel, F., Gillingham, A. K., Rosa-Ferreira, C., Galindo, A. & Munro, S. An antibody toolkit for the study of membrane traffic in Drosophila melanogaster. Bio Open, bio. 018937 (2016). 163. Nezis, I. P. et al. Ref (2) P, the Drosophila melanogaster homologue of mammalian p62, is required for the formation of protein aggregates in adult brain. J Cell Biol 180, 1065-1071 (2008). 164. Song, W. et al. Activin signaling mediates muscle-to-adipose communication in a mitochondria dysfunction-associated obesity model. Proc Natl Acad Sci 114, 8596-8601 (2017). 165. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat Methods 9, 676-682 (2012). 166. Kapur, J. N., Sahoo, P. K. & Wong, A. K. A new method for gray-level picture thresholding using the entropy of the histogram. Comput Vis Graph Image Proc 29, 273-285 (1985). 190 167. Zack, G., Rogers, W. & Latt, S. Automatic measurement of sister chromatid exchange frequency. J Histochem Cytochem 25, 741-753 (1977). 168. Mitchell KJ, Pinson KI, Kelly OG, Brennan J, Zupicich J, Scherz P, et al. Functional analysis of secreted and transmembrane proteins critical to mouse development. Nat Genet. 2001;28(3):241-9. 169. Kotas ME, Medzhitov R. Homeostasis, inflammation, and disease susceptibility. Cell. 2015;160(5):816-27. 170. Anbazhagan R, Kavarthapu R, Mathur PP, Kostic TS, Prakash H. Editorial: Systemic Regulation of Organ Homeostasis and Implications of Hormones and Immunity. Front Endocrinol (Lausanne). 2021;12:740835. 171. Uhlén M, Karlsson MJ, Hober A, Svensson AS, Scheffel J, Kotol D, et al. The human secretome. Sci Signal. 2019;12(609). 172. Basu RK, Wheeler DS. Kidney-lung cross-talk and acute kidney injury. Pediatr Nephrol. 2013;28(12):2239-48. 173. Husain-Syed F, Slutsky AS, Ronco C. Lung-Kidney Cross-Talk in the Critically Ill Patient. American journal of respiratory and critical care medicine. 2016;194(4):402-14. 174. Gourd NM, Nikitas N. Multiple Organ Dysfunction Syndrome. J Intensive Care Med. 2020;35(12):1564-75. 175. Pietrowska M, Wlosowicz A, Gawin M, Widlak P. MS-Based Proteomic Analysis of Serum and Plasma: Problem of High Abundant Components and Lights and Shadows of Albumin Removal. Adv Exp Med Biol. 2019;1073:57-76. 176. Keshishian H, Burgess MW, Specht H, Wallace L, Clauser KR, Gillette MA, et al. Quantitative, multiplexed workflow for deep analysis of human blood plasma and biomarker discovery by mass spectrometry. Nat Protoc. 2017;12(8):1683-701. 177. Strack R. Revealing the secretome. Nat Methods. 2021;18(11):1273. 178. Uezu A, Kanak DJ, Bradshaw TW, Soderblom EJ, Catavero CM, Burette AC, et al. Identification of an elaborate complex mediating postsynaptic inhibition. Science. 2016;353(6304):1123-9. 179. Zhou K, Ding Y, Vuletic I, Tian Y, Li J, Liu J, et al. In vivo long-term investigation of tumor bearing mKate2 by an in-house fluorescence molecular imaging system. Biomed Eng Online. 2018;17(1):187. 180. Cranfill PJ, Sell BR, Baird MA, Allen JR, Lavagnino Z, de Gruiter HM, et al. Quantitative assessment of fluorescent proteins. Nat Methods. 2016;13(7):557-62. 181. Shcherbo D, Murphy CS, Ermakova GV, Solovieva EA, Chepurnykh TV, Shcheglov AS, et al. Far-red fluorescent tags for protein imaging in living tissues. Biochem J. 2009;418(3):567-74. 182. Hayashi S, Lewis P, Pevny L, McMahon AP. Efficient gene modulation in mouse epiblast using a Sox2Cre transgenic mouse strain. Mech Dev. 2002;119 Suppl 1:S97-S101. 191 183. Madisen L, Zwingman TA, Sunkin SM, Oh SW, Zariwala HA, Gu H, et al. A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat Neurosci. 2010;13(1):133-40. 184. Nielsen H. Predicting Secretory Proteins with SignalP. Methods Mol Biol. 2017;1611:59-73. 185. Nielsen H, Engelbrecht J, Brunak S, von Heijne G. Identification of prokaryotic and eukaryotic signal peptides and prediction of their cleavage sites. Protein Eng. 1997;10(1):1- 6 186. Möller S, Croning MD, Apweiler R. Evaluation of methods for the prediction of membrane spanning regions. Bioinformatics. 2001;17(7):646-53. 187. Sonnhammer EL, von Heijne G, Krogh A. A hidden Markov model for predicting transmembrane helices in protein sequences. Proc Int Conf Intell Syst Mol Biol. 1998;6:175- 82. 188. Jain A, Tuteja G. TissueEnrich: Tissue-specific gene enrichment analysis. Bioinformatics. 2019;35(11):1966-7. 189. Hooper NM, Karran EH, Turner AJ. Membrane protein secretases. Biochem J. 1997;321 ( Pt 2):265-79. 190. Lichtenthaler SF, Lemberg MK, Fluhrer R. Proteolytic ectodomain shedding of membrane proteins in mammals-hardware, concepts, and recent developments. EMBO J. 2018;37(15). 191. Althoff K, Müllberg J, Aasland D, Voltz N, Kallen K, Grötzinger J, et al. Recognition sequences and structural elements contribute to shedding susceptibility of membrane proteins. Biochem J. 2001;353(Pt 3):663-72. 192. Hamburger AE, West AP, Bjorkman PJ. Crystal structure of a polymeric immunoglobulin binding fragment of the human polymeric immunoglobulin receptor. Structure. 2004;12(11):1925-35. 193. Sanderson MP, Keller S, Alonso A, Riedle S, Dempsey PJ, Altevogt P. Generation of novel, secreted epidermal growth factor receptor (EGFR/ErbB1) isoforms via metalloprotease- dependent ectodomain shedding and exosome secretion. J Cell Biochem. 2008;103(6):1783- 97. 194. Garton KJ, Gough PJ, Philalay J, Wille PT, Blobel CP, Whitehead RH, et al. Stimulated shedding of vascular cell adhesion molecule 1 (VCAM-1) is mediated by tumor necrosis factor-alpha-converting enzyme (ADAM 17). J Biol Chem. 2003;278(39):37459-64. 195. Hashimoto M, Yamazaki A, Ohno A, Kimura T, Winblad B, Tjernberg LO. A Fragment of S38AA is a Novel Plasma Biomarker of Alzheimer's Disease. J Alzheimers Dis. 2019;71(4):1163-74. 196. Zhang M, Liu L, Lin X, Wang Y, Li Y, Guo Q, et al. A Translocation Pathway for Vesicle- Mediated Unconventional Protein Secretion. Cell. 2020;181(3):637-52.e15. 192 197. Rubartelli A, Bajetto A, Allavena G, Wollman E, Sitia R. Secretion of thioredoxin by normal and neoplastic cells through a leaderless secretory pathway. J Biol Chem. 1992;267(34):24161-4. 198. Nüchel J, Ghatak S, Zuk AV, Illerhaus A, Mörgelin M, Schönborn K, et al. TGFB1 is secreted through an unconventional pathway dependent on the autophagic machinery and cytoskeletal regulators. Autophagy. 2018;14(3):465-86. 199. Postic C, Shiota M, Niswender KD, Jetton TL, Chen Y, Moates JM, et al. Dual roles for glucokinase in glucose homeostasis as determined by liver and pancreatic beta cell-specific gene knock-outs using Cre recombinase. J Biol Chem. 1999;274(1):305-15. 200. Lavoie JL, Sigmund CD. Minireview: overview of the renin-angiotensin system--an endocrine and paracrine system. Endocrinology. 2003;144(6):2179-83. 201. Matsusaka T, Niimura F, Shimizu A, Pastan I, Saito A, Kobori H, et al. Liver angiotensinogen is the primary source of renal angiotensin II. J Am Soc Nephrol. 2012;23(7):1181-9. 202. Catta-Preta M, Martins MA, Cunha Brunini TM, Mendes-Ribeiro AC, Mandarim-de- Lacerda CA, Aguila MB. Modulation of cytokines, resistin, and distribution of adipose tissue in C57BL/6 mice by different high-fat diets. Nutrition. 2012;28(2):212-9. 203. Shimamura M, Matsuda M, Yasumo H, Okazaki M, Fujimoto K, Kono K, et al. Angiopoietin-like protein3 regulates plasma HDL cholesterol through suppression of endothelial lipase. Arterioscler Thromb Vasc Biol. 2007;27(2):366-72. 204. Leonard EJ, Skeel AH. Isolation of macrophage stimulating protein (MSP) from human serum. Exp Cell Res. 1978;114(1):117-26. 205. Burch PM, Pogoryelova O, Palandra J, Goldstein R, Bennett D, Fitz L, et al. Reduced serum myostatin concentrations associated with genetic muscle disease progression. J Neurol. 2017;264(3):541-53. 206. Spaks A. Role of CXC group chemokines in lung cancer development and progression. J Thorac Dis. 2017;9(Suppl 3):S164-S71. 207. Elis S, Courtland HW, Wu Y, Rosen CJ, Sun H, Jepsen KJ, et al. Elevated serum levels of IGF-1 are sufficient to establish normal body size and skeletal properties even in the absence of tissue IGF-1. J Bone Miner Res. 2010;25(6):1257-66. 208. Katafuchi T, Esterházy D, Lemoff A, Ding X, Sondhi V, Kliewer SA, et al. Detection of FGF15 in plasma by stable isotope standards and capture by anti-peptide antibodies and targeted mass spectrometry. Cell Metab. 2015;21(6):898-904. 209. Cook KS, Min HY, Johnson D, Chaplinsky RJ, Flier JS, Hunt CR, et al. Adipsin: a circulating serine protease homolog secreted by adipose tissue and sciatic nerve. Science. 1987;237(4813):402-5. 193 210. Favre D, Ferry N, Moullier P. Critical aspects of viral vectors for gene transfer into the kidney. J Am Soc Nephrol. 2000;11 Suppl 16:S149-53. 211. Kim DI, Jensen SC, Noble KA, Kc B, Roux KH, Motamedchaboki K, et al. An improved smaller biotin ligase for BioID proximity labeling. Mol Biol Cell. 2016;27(8):1188-96. 212. Stachon P, Geis S, Peikert A, Heidenreich A, Michel NA, Ünal F, et al. Extracellular ATP Induces Vascular Inflammation and Atherosclerosis via Purinergic Receptor Y2 in Mice. Arterioscler Thromb Vasc Biol. 2016;36(8):1577-86. 213. Hughes ED, Qu YY, Genik SJ, Lyons RH, Pacheco CD, Lieberman AP, et al. Genetic variation in C57BL/6 ES cell lines and genetic instability in the Bruce4 C57BL/6 ES cell line. Mamm Genome. 2007;18(8):549-58. 214. Raymond CS, Soriano P. High-efficiency FLP and PhiC31 site-specific recombination in mammalian cells. PLoS One. 2007;2(1):e162. 215. Hayashi S, McMahon AP. Efficient recombination in diverse tissues by a tamoxifen- inducible form of Cre: a tool for temporally regulated gene activation/inactivation in the mouse. Dev Biol. 2002;244(2):305-18. 216. Postic C, Shiota M, Niswender KD, Jetton TL, Chen Y, Moates JM, et al. Dual roles for glucokinase in glucose homeostasis as determined by liver and pancreatic beta cell-specific gene knock-outs using Cre recombinase. J Biol Chem. 1999;274(1):305-15. 217. Hughes CS, Foehr S, Garfield DA, Furlong EE, Steinmetz LM, Krijgsveld J. Ultrasensitive proteome analysis using paramagnetic bead technology. Mol Syst Biol. 2014;10:757. 218. Deng W, Sha J, Plath K, Wohlschlegel JA. Carboxylate-Modified Magnetic Bead (CMMB)- Based Isopropanol Gradient Peptide Fractionation (CIF) Enables Rapid and Robust Off- Line Peptide Mixture Fractionation in Bottom-Up Proteomics. Mol Cell Proteomics. 2021;20:100039. 219. Kahn, B.B. (2019). Adipose tissue, inter-organ communication, and the path to type 2 diabetes: The 2016 banting medal for scientific achievement lecture. Diabetes 68, 3–14. 220. Frayn, K.N., Karpe, F., Fielding, B.A., Macdonald, I.A., and Coppack, S.W. (2003). Integrative physiology of human adipose tissue. Int. J. Obes. 27, 875–888. 221. Benites-Zapata, V.A., Toro-Huamanchumo, C.J., Urrunaga-Pastor, D., Guarnizo-Poma, M., Lazaro-Alcantara, H., Paico-Palacios, S., Pantoja-Torres, B., and Ranilla-Seguin, V. del C. (2019). High waist-to-hip ratio levels are associated with insulin resistance markers in normal-weight women. Diabetes Metab. Syndr. Clin. Res. Rev. 13, 636–642. 194 222. Coolbaugh, C.L., Damon, B.M., Bush, E.C., Welch, E.B., and Towse, T.F. (2019). Cold exposure induces dynamic, heterogeneous alterations in human brown adipose tissue lipid content. Sci. Rep. 9, 1-13 223. Kleinert, M., Clemmensen, C., Hofmann, S.M., Moore, M.C., Renner, S., Woods, S.C., Huypens, P., Beckers, J., de Angelis, M.H., Schürmann, A., et al. (2018). Animal models of obesity and diabetes mellitus. Nat. Rev. Endocrinol. 14, 140–162. 224. Eguchi, J., Wang, X., Yu, S., Kershaw, E.E., Chiu, P.C., Dushay, J., Estall, J.L., Klein, U., Maratos-Flier, E., and Rosen, E.D. (2011). Transcriptional control of adipose lipid handling by IRF4. Cell Metab. 13, 249–259. 225. Gui, Y., Silha, J. V., and Murphy, L.J. (2004). Sexual dimorphism and regulation of resistin, adiponectin, and leptin expression in the mouse. Obes. Res. 12, 1481–1491. 226. Berry, R., Church, C.D., (2014). Imaging of Adipose Tissue. O.A., Mcdougald (Ed.) Methods in Enzymology: Methods of Adipose Tissue Biology, Part A (Vol. 537, 47-73) (Academic Press) 227. Wong, G.W., Wang, J., Hug, C., Tsao, T.S., and Lodish, H.F. (2004). A family of Acrp30/adiponectin structural and functional paralogs. Proc. Natl. Acad. Sci. U. S. A. 101, 10302–10307.Lo, J.C., Ljubicic, S., Leibiger, B., Kern, M., Leibiger, I.B., Moede, T., Kelly, M.E., Chatterjee Bhowmick, D., Murano, I., Cohen, P., et al. (2014). Adipsin is an adipokine that improves β cell function in diabetes. Cell 158, 41–53. 228. Gómez-Banoy, N., Guseh, J.S., Li, G., Rubio-Navarro, A., Chen, T., Poirier, B.A., Putzel, G., Rosselot, C., Pabón, M.A., Camporez, J.P., et al. (2019). Adipsin preserves beta cells in diabetic mice and associates with protection from type 2 diabetes in humans. Nat. Med. 25, 1739–1747.

Asset Metadata

Creator Meyer, Amanda Stevens (author)

Core Title Mammalian secretome profiling to identify adaptive and maladaptive signaling in homeostasis and diet induced obesity

Contributor Electronically uploaded by the author (provenance)

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Molecular Biology

Degree Conferral Date 2022-12

Publication Date 12/06/2023

Defense Date 11/14/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag adipocyte,cell secretion,mouse models,OAI-PMH Harvest,obesity,proximity labeling

Format theses (aat)

Language English

Advisor McMahon, Andrew (committee chair), Chen, Xiaojiang (committee member), Dean, Matthew (committee member), Pinaud, Fabien (committee member), Youn, Jang-Hyun (committee member)

Creator Email asmeyer46@gmail.com,meyeras@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC112617320

Unique identifier UC112617320

Identifier etd-MeyerAmand-11339.pdf (filename)

Legacy Identifier etd-MeyerAmand-11339

Document Type Dissertation

Format theses (aat)

Rights Meyer, Amanda Stevens

Internet Media Type application/pdf

Type texts

Source 20221207-usctheses-batch-994 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email uscdl@usc.edu

Abstract (if available)

Abstract Identifying cell-type specific circulating proteins remains a key challenge. Here, we adapted biotin-based proximity labeling to in vivo mammalian systems. We demonstrated in vivo proximity labeling “proof-of-principle” by identifying teratoma-derived circulating proteins in mice. We then generated a genetic proximity labeling mouse where a promiscuous biotin ligase, BirA*G3, is localized to the endoplasmic reticulum allowing for biotinylation and subsequent identification of secreted proteins. To profile secretion in a cell-type specific manner, we used a floxed BirA*G3 allele when crossed to a Cre-driver excises a GFP-stop cassette resulting in BirA*G3 expression. Extensive characterization of this mouse shows the ability to label and identify secreted proteins from numerous tissues including liver (ALB, AGT), kidney (UMOD), immune cells (CXCL7), muscle (MSTN), fat (ADIPOQ, Cfd), gut (FGF15), and brain (OXT). We then sought to apply this technique to identify diet induced obesity adipocyte secretion changes in mice. High fat diet treatment for 12 weeks resulted in significant weight gain and altered biotinylated protein profiles. Proteomic analysis of these samples identified diet specific protein changes. These studies demonstrate proximity labeling detection of cell-type specific secretomes in homeostasis and in disease. This mouse model provides a new tool for the research community to apply to numerous biological systems. By using this mouse model to investigate adipocyte secretion we have identified new insight in adipocyte-derived circulating factors and provided strong evidence of proximity labeling approaches to identify disease signatures. Together, this body of work has contributed to the advancement and development of in vivo proximity labeling techniques.