Materials and Methods

Molecular cloning and generation of stable cell lines

All plasmids used in this study were constructed by PCR and Gibson isothermal assembly. The expression ORFs, hereafter named bait-BirA, were constructed by fusing BioID2 to the C-terminal of each model protein (with a glycine-serine linker added between) and a 3XFLAG tag at C-terminal to simplify the immuno detection. ORFs were inserted into pcDNA5/FRT (Invitrogen), which allows targeted integration of the transgenes into the host genome. Gibson assembly primers were designed by SnapGene software and used to amplify the corresponding fragments and vectors with long overlapping overhangs, which were then assembled using Gibson Assembly Master Mix (NEB). To obtain secretable BioID2 (without any bait protein), Gibson assembly was employed to fuse the signal peptide of SERPINC1 gene to the N-terminal of BirA (hereafter referred to as Signal-BirA). Assembly products were transformed to the chemically competent e. coli , and recombinant plasmids were verified by restriction digestion and sequencing. For all experiments, Flp‐In 293 cells (Invitrogen) were cultured in DMEM media supplemented with fetal bovine serum (10 %) and antibiotics (penicillin, 100 U mL−1 and streptomycin, 100 μg mL−1) and maintained at 37 °C under 5 % CO2. For generating stable cell lines, Flp-In 293 cells were seeded in 6 well plates at a density of 0.5×106 cells per well the day before transfection. Cells were then co-transfected with each pcDNA5/FRT vector containing expression cassette and pOG44 plasmid using Lipofectamine® 2000 according to the manufacturer’s directions. After recovery from transfection, cells were grown in DMEM containing 10% FBS, 1% PenStrep, and 150 μg/mL Hygromycin B to select hygromycin-resistant cells. Individual resistant colonies were isolated, pooled, and seeded in 24-well plates for further scaling up and screened for expression of the fusion proteins by Western Blotting.

Immunofluorescence

Recombinant HEK293 cells expressing BioID2 fusions were grown in complete medium supplemented with 50 uM biotin on coverslips until 70% confluent. Cells were then fixed in PBS containing 4% PFA for 10 min at room temperature. Blocking was performed by incubating fixed cells with 1% BSA and 5% normal goat serum in PBST. Anti-flag mouse monoclonal antibody-Dylight 650 conjugate (Thermofisher), targeting the bait-BirA, and streptavidin-DyLight 594 conjugate (Thermofisher), targeting the biotinylated proteins, were diluted at 1:300 and 1:1000 in blocking buffer, respectively and incubated with fixed cells for 30 minutes at room temperature. Cells were then washed, counterstained with DAPI, mounted on the slide using antifade vectashield mountant, and imaged using Leica SP8 Confocal with Lightning Deconvolution. Colocalization quantification was performed for the deconvolved images using Fiji’s (ImageJ 1.52p) Coloc_2 analysis tool between the 650 (anti-flag) and 594 (Streptavidin) channels (Schindelin, Rueden, Hiner, & Eliceiri, 2015). This tool generates a comprehensive report for evaluating pixel intensity colocalization of two channels by various methods such as Pearson’s Coefficient (range: -1.0 to 1.0), Manders’ Colocalization Coefficients (MCC, range: 0 to 1.0), and Li’s Intensity Correlation Quotient (IQC, range: -0.5 to 0.5) (Li, 2004; Manders, Verbeek, & Aten, 1993). Background pixel intensity was subtracted using Fiji’s rolling ball algorithm and a region of interest (ROI). Thresholds were determined using Coloc_2’s bisection method, which is further used to adjust for background. Above threshold metrics were reported.

Western blotting

To validate the secretion of bait-BirA proteins, supernatants of cultures expressing fusion proteins were collected, centrifuged to remove cell debris, and protein content was concentrated using Amicon Ultra 4 mL 10 KD filter unit (MilliporeSigma) and quantified using Bradford assay (Expedeon). 20 ug of total protein was loaded on SDS-PAGE gel for electrophoresis and resolved proteins were transblotted to nitrocellulose membrane using Trans-Blot Turbo Transfer System from Bio-RAD. The membrane was blocked with 5% skim milk in TBST and probed with HRP-conjugated anti-flag mouse monoclonal antibody (Thermofisher) diluted at 1:10000 in the blocking buffer. The membrane was washed, and Clarity Western ECL Substrate was added. Proteins’ bonds were visualized using G:Box Gel Image Analysis Systems (SYNGENE). For staining of intracellular biotinylated proteins, cells were grown in complete medium supplemented with 50 μM biotin, lysed by RIPA buffer, and protein content was quantified using Bradford assay. 20 ug of total protein was loaded and resolved and transblotted as described earlier. The membrane was blocked by 3% BSA in TBST and probed with HRP-conjugated streptavidin diluted in blocking buffer at 1:2000 ratio. For visualizing the proteins’ bands, the same Clarity Western ECL Substrate was used.

ELISA

Aliquots of clarified supernatants from esiRNA transfected cultures were taken out of −80°C and thawn on the ice and immediately subjected to ELISA. Flag levels were quantified from secreted bait-BirA fusion proteins by competitive anti-flag ELISA using the Flag-Tag Detection ELISA Kit (Cayman Chemicals). All measurements were performed according to the manufacture instructions in triplicate. The effect of PDIA’s knockdown on secretion of the model proteins was measured in comparison with the negative control of each cell line transfected with EGFP esiRNA as negative control (see below).

RNAi knockdown experiment

esiRNA targeting ERP44 was ordered from Sigma. HEK293 cell expressing SERPINC1-BirA were seeded at 0.6X105 cells/well in 24-well plates with complete medium and reverse transfected with 72 ng of ERP44 specific esiRNA or EGFP esiRNA as a negative control or KIF11 esiRNA as positive control using Lipofectamine RNAiMAX (Invitrogen). All transfections were performed according to the manufacturer’s guidelines. Targeted gene knockdown by esiRNA was allowed to occur for 48 and 72 h, culture supernatants were harvested, clarified by low-speed centrifugation, then aliquoted and stored at −80°C for further experiments.

Mass Spectrometry

Cells were grown in 245 mm plates (one plate per biological replicate in triplicate) to approximately 70% confluence in complete media and then incubated for 24 h with 50 μM biotin. Cells were harvested and washed twice in cold PBS, lysed with vigorous shaking (20 Hz) in 8M urea, 50mM ammonium bicarbonate lysis buffer, extracted proteins were centrifuged at 14,000 x g to remove cellular debris and quantified by BCA assay (Thermo Scientific) as per manufacturer recommendations. Affinity purification of biotinylated proteins was carried out in a Bravo AssayMap platform (Agilent) using AssayMap streptavidin cartridges (Agilent), and the bound proteins were subjected to on-cartridge digestion with mass spec grade Trypsin/Lys-C Rapid digestion enzyme (Promega, Madison, WI) at 70ºC for 2h. Digested peptides were then desalted in the Bravo platform using AssayMap C18 cartridges and the organic solvent was removed in a SpeedVac concentrator prior to LC-MS/MS analysis. Dried peptides were reconstituted with 2% acetonitrile, 0.1% formic acid, and analyzed by LC-MS/MS using a Proxeon EASY nanoLC system (Thermo Fisher Scientific) coupled to a Q-Exactive Plus mass spectrometer (Thermo Fisher Scientific). Peptides were separated using an analytical C18 Acclaim PepMap column 0.075 x 500 mm, 2µm particles (Thermo Scientific) in a 93-min linear gradient of 2-28% solvent B at a flow rate of 300nL/min. The mass spectrometer was operated in positive data-dependent acquisition mode. MS1 spectra were measured with a resolution of 70,000, an AGC target of 1e6 and a mass range from 350 to 1700 m/z. Up to 12 MS2 spectra per duty cycle were triggered, fragmented by HCD, and acquired with a resolution of 17,500 and an AGC target of 5e4, an isolation window of 1.6 m/z and a normalized collision energy of 25. Dynamic exclusion was enabled with a duration of 20 sec.

MS data Analysis

All mass spectra were analyzed with MaxQuant software (Tyanova et al., 2016) version 1.5.5.1. MS/MS spectra were searched against the Homo sapiens Uniprot protein sequence database (version January 2018) and GPM cRAP sequences (commonly known protein contaminants). Precursor mass tolerance was set to 20ppm and 4.5ppm for the first search where initial mass recalibration was completed and for the main search, respectively. Product ions were searched with a mass tolerance 0.5 Da. The maximum precursor ion charge state used for searching was 7. Carbamidomethylation of cysteines was searched as a fixed modification, while oxidation of methionines and acetylation of protein N-terminal were searched as variable modifications. Enzyme was set to trypsin in a specific mode and a maximum of two missed cleavages was allowed for searching. The target-decoy-based false discovery rate (FDR) filter for spectrum and protein identification was set to 1%. Enrichment of proteins in streptavidin affinity purifications were calculated as the ratio of intensity. To remove the systematic biases introduced during various steps of sample processing and data generation, dataset were normalized using the LOESS method (Smyth, 2005) integrated into Normalyzer (Chawade, Alexandersson, & Levander, 2014). Perseus software (Tyanova & Cox, 2018) was employed for data preparation, filtering, and computation of differential protein abundance. The DEP package (Zhang et al., 2018) was used to explore whether missing values in the dataset are biased to lower intense proteins. Left-censored imputation was performed using random draws from shifted distribution. A Student’s t‐test with a multi‐sample permutation‐based correction for an FDR of 0.05 was employed to identify differentially expressed proteins using log2 transformed data.

Detection of significant interactions

The threshold for significant interactions was determined using the known secretory pathway components as a gold standard. We set the cutoffs for FDR at 0.1 and removed all interactors with negative fold changes, as this optimizes the enrichment of known secretory pathway components among the significant interactors. The enrichment for two independent secretory pathway-related gene sets also peaked around the cutoffs set through the gene set of known secretory pathway components, suggesting the optimal cutoffs are robust to the gold standards chosen.

Aggregation of interactions and estimation of interaction synergy

To further generalize the interactions between individual SecPs and their interactors, we aggregated proteins based on several features. Given the interactions between the SecPs and their interactors, we can predict the important structural features implicated in the interactions between the SecPs and a given SecM. The interactions between the BirA-fused samples and the secretory pathway interactors can be pooled according to shared properties of the SecPs to reveal interdependencies between components of the secretory pathway and their products. These common properties include shared structural motifs (Blatch & Lässle, 1999), known sites of PTM from Uniprot and phosphosite (Blatch & Lässle, 1999; Hornbeck et al., 2015). The secretory pathway interactors can also be aggregated into curated biological pathways (mSigDB (Liberzon et al., 2011), KEGG (Kanehisa, Furumichi, Tanabe, Sato, & Morishima, 2017), Reactome (Croft et al., 2014)) and subcellular localizations (Thul et al., 2017).

Bayesian modeling framework

To quantify such associations of each interacting protein with protein features, we calculate the effective total frequency (𝛿f,g) of interactions between each feature-gene pair (f,g) by going through every SecP in our data and counting the number of times this feature occurs in a SecP p (fp) .