bioRxiv preprint doi: https://doi.org/10.1101/264788; this version posted February 13, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Liu et al. An integrated approach to comprehensively map the molecular context of proteins Xiaonan Liu1,2, Kari Salokas1,2, Fitsum Tamene1,2,3 and Markku Varjosalo1,2,3* 1Institute of Biotechnology, University of Helsinki, Helsinki 00014, Finland 2Helsinki Institute of Life Science, University of Helsinki, Helsinki 00014, Finland 3Proteomics Unit, University of Helsinki, Helsinki 00014, Finland *Corresponding author Abstract: Protein-protein interactions underlie almost all cellular functions. The comprehensive mapping of these complex cellular networks of stable and transient associations has been made available by affi nity purifi cation mass spectrometry (AP-MS) and more recently by proximity based labelling methods such as BioID. Due the advancements in both methods and MS instrumentation, an in-depth analysis of the whole human proteome is at grasps. In order to facilitate this, we designed and optimized an integrated approach utilizing MAC-tag combining both AP-MS and BioID in a single construct. We systematically applied this approach to 18 subcellular localization markers and generated a molecular context database, which can be used to defi ne molecular locations for any protein of interest. In addition, we show that by combining the AP-MS and BioID results we can also obtain interaction distances within a complex. Taken together, our combined strategy off ers comprehensive approach for mapping physical and functional protein interactions. Introduction: Majority of proteins do not function in isolation geting the endogenous bait protein, allowing and their interactions with other proteins defi ne purifi cation of the bait protein together with the their cellular functions. Therefore, detailed under- associating proteins (preys). This approach has standing of protein-protein interactions (PPIs) is been proven well suited for even large-scale high- the key for deciphering regulation of cellular net- throughput studies, and to yield highly repro- works and pathways. During the last decade, the ducible data in both intra- and inter-laboratory versatile combination of affi nity purifi cation and usage2. The most commonly used epitope tags in mass spectrometry (AP-MS) revolutionized the medium to large-scale studies include FLAG3, His4, detailed characterization of protein complexes MYC5, HA6, GFP7and Strep8, of which the Strep-tag and protein-interaction networks1. The AP-MS has become the gold-standard in affi nity purifi - approach relies on expression of a bait protein cation proteomics due to unparalleled protein coupled with an epitope tag or antibodies tar- purity in physiological purifi cation conditions as 1 bioRxiv preprint doi: https://doi.org/10.1101/264788; this version posted February 13, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. An integrated approach to comprehensively map the molecular context of proteins well as the possibility for native competitive elu- we included as well a nine amino acid hemagglu- tion using biotin. tinin (HA)-epitope. The HA-epitope also facilitates AP-MS can also be combined with quantitative additional follow-up approaches such as ChIP- proteomics approaches to better understand the Seq17 and purifi cation of the crosslinked proteins protein complex stoichiometry9 and the dynamics for cross-linking coupled with mass spectrometry of protein–complex (dis)assembly1,10. The combi- (XL-MS)18, making the MAC-tag almost as versatile nation of AP-MS with other techniques, such as as the Swiss Army knife. biochemical fractionation, intact mass measure- To benchmark the usability and performance of ment and chemical crosslinking11,12, has been used the MAC-tag we applied it to 18 bona fi de subcel- to characterize supramolecular organization of lular localization marker proteins. This allowed us protein complexes. to validate the correct localization of the MAC- Although AP-MS remains the most used tagged marker proteins as well as to monitor the method for mapping protein-protein inter- localization of the in vivo biotinylated interactors. actions, the recently developed proximity Additionally the interactions provide new infor- labeling approaches, such as BioID13 and APEX14, mation about these 18 marker proteins and their have become complementary and somewhat cellular functions. Furthermore the 18 localization competing methods. BioID involves expression markers and their 1911 interactions, form a basis of the protein of interest fused with a prokary- of the reference molecular context repository, otic biotin ligase (BirA) and the subsequent which we show can be used for “mass spectrom- biotinylation of the amine groups of the neigh- etry (MS) microscopy” analysis of any protein of boring proteins when excess of biotin is added interest. The combinatory analysis with AP-MS and to the cells. Whereas the wild-type BirA from E. BioID also provided information, which effi ciently Coli is capable of transferring the biotin only to a could be used to derive relative spatial distances substrate bearing a specifi c recognition sequence, for proteins in a complex. Taken together, our the generation of a promiscuous BirA* (Arg118Gly devised combinatory MAC-tag and analysis mutant) allows the biotinylation of any protein approaches around it provide a plethora of infor- found within a 10 nm labeling radius13,15. While mation of the cellular functions and the molecular BioID has the abilities to capture weak and/or context of any studied protein. transient protein-protein interactions, the identi- fi ed interactions are not limited to direct binders RESULTS but can include proximate proteins as well. MAC- tag AP-MS and BioID pipeline for In order to avoid artefactual interactions detection of physical and functional caused by overexpression of the bait proteins, interactions majority of the large-scale interaction proteomic studies employ the Flp-In™ T-REx 293 cell line To generate a versatile approach for identifi- allowing moderate and inducible bait protein cation of both stable physical and transient expression from isogenic cell clones16. Although functional protein-protein interactions we inte- the system allows rapid generation of transgene grated and optimized the BioID approach with 10,23 stably expressing cell lines, comprehensive our single-strep Strep AP-MS pipeline . Both analyses utilizing complementarily both AP-MS of these approaches have become the method and BioID is resource-intense in the respect of cell of choice for interactomics analyses. We have line generation. To address this caveat and allow recently shown the eff ectiveness of using these 10,23 high-throughput comprehensive interactome approached complimentarily . However, the analyses, we generated a Gateway®-compatible complementary use of the two techniques has MAC (Multiple Approaches Combined) -tag been labor-intense, involving tagging of the bait enabling both the single-step Strep AP-MS as well proteins with BirA* and Strep-tag individually, as the BioID analysis with a single construct, which as well as generation of two set of cell lines per decreases the number of required individual cell bait. To overcome the major limitations, we have lines by 50%. In addition to allow visualization of developed an integrated experimental workfl ow tagged bait protein by immunohistochemistry, utilizing a MAC-tag containing both StrepIII-tag 2 bioRxiv preprint doi: https://doi.org/10.1101/264788; this version posted February 13, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Liu et al. and BirA* (Supplementary Fig. 1a). In addition to included: mitochondria (Apoptosis-inducing fac- optimizing the experimental steps, we focused tor 1, AIFM1); endoplasmic reticulum (Calnexin, on the compatibility of the two methods and to CALX); peroxisome (Catalase, CATA); early endo- the simplicity of the analysis pipelines to gener- some (Early endosome antigen 1, EEA1); cytoplas- ate a process with improved performance and mic peripheral plasma membrane marker (Ezrin, reproducibility on detecting protein-protein EZRI); nucleolus marker (rRNA 2’-O-methyltrans- interactions. The two pipelines diff er only in the ferase fi brillarin, FBRL); cis-Golgi marker (Golgin activation of the BirA* by addition of biotin to the subfamily A member 2, GOGA2); chromatin (His- cell culture media and harsher lysis condition in tone H3.1, H31); exosome (Heat shock cognate 71 the BioID pipeline (Fig. 1, Supplementary Fig. 1). kDa protein, HSP7C); lysosome (Lysosome-associ- Without biotin addition the BirA* in the MAC-tag ated membrane glycoprotein 1, LAMP1); nuclear remains inactive (Supplementary Fig. 1b, c), result- envelope marker (Prelamin-A/C, LMNA); protea- ing in identical (cor=0.88-0.99) single-step affi nity some (Proteasome subunit alpha type-1, PSA1); purifi cation results as vector with only StrepIII-tag recycling endosome (Ras-related protein Rab-11A, (Supplementary Fig. 1d, e). Similarly, when biotin RAB11A); late endosome (Ras-related protein Rab- was added the results compare (cor=0.95-0.97) to 9A, RAB9A); microtubule (Tubulin alpha-1A chain, that of a