Optimized Workflow for Enrichment and Identification of Biotinylated Peptides using Tamavidin 2-REV for BioID and Cell Surface Proteomics

Chemical or enzymatic biotinylation of proteins is widely used in various studies, and proximity-dependent biotinylation coupled to mass spectrometry is a powerful approach for analyzing protein–protein interactions in living cells. We recently developed a simple method to enrich biotinylated peptides using Tamavidin 2-REV, an engineered avidin-like protein with reversible biotin-binding capability. However, the low abundance of protein biotinylation in cells required large amounts of cellular proteins to detect enough biotinylated peptides. Here we optimized the workflow for efficient enrichment and identification of biotinylated peptides. The most efficient recovery was achieved by heat inactivation of trypsin, prewashing Tamavidin 2-REV beads, clean-up of biotin solution, mock elution, and the optimal temperature and salt concentration for elution. Using the optimized workflow, over 2-fold more biotinylated peptides were identified with higher purity from RAW264.7 macrophages expressing TurboID-fused STING. In addition, sequential digestion with Glu-C and trypsin led to the identification of biotinylation sites that were not identified by trypsin digestion alone. Furthermore, the combination of this workflow with TMT labeling enabled large-scale quantification of cell surface proteome changes upon EGF stimulation. This workflow would be useful not only for BioID and cell surface proteomics but also for various other applications based on protein biotinylation.

To efficiently enrich and identify biotinylated peptides, two groups have developed a revolutionary 51 method using anti-biotin antibody beads, from which biotinylated peptides can be efficiently but non-52 specifically eluted with strong acid and high concentration of organic solvent (15, 16). Very recently, 53 we developed a simple method to specifically enrich biotinylated peptides using Tamavidin 2-REV 54 beads (17). Tamavidin 2-REV is an engineered version of Tamavidin 2 that acquired reversible biotin-55 binding capability by mutating S36 that forms a hydrogen bond with biotin (17, 18). Because 56 biotinylated peptides can be mildly and specifically eluted from Tamavidin 2-REV beads by adding 57 excess free biotin, the enrichment efficiency was higher than that of anti-biotin antibody beads. 58 However, interference ions derived from our enrichment process prevented sensitive detection of 59 biotinylated peptides. Furthermore, the low abundance of protein biotinylation in BioID-expressing 60 cells required large amounts of cellular proteins to identify enough biotinylated peptides by mass 61 Tip SDB, evaporated in a SpeedVac concentrator, and redissolved in 0.1% TFA and 3% ACN. 136 137

LC-MS/MS of BioID Samples 138
Peptides were analyzed by an EASY-nLC 1200 UHPLC-coupled Orbitrap Fusion Tribrid mass 139 spectrometer (Thermo Fisher Scientific). Aliquots (10 µL) of each sample were loaded by 20 μL of 140 0.1% formic acid in water at a pressure of 200 bar onto a trap column (100 μm × 2 cm, PepMap 141 nanoViper C18 column, 5 μm, 100 Å; Thermo Fisher Scientific). The following mobile phases were 142 used: solvent A (0.1% formic acid in water) and solvent B (80% ACN and 0.1% formic acid in water). 143 The trapped peptides were eluted from an analytical column (75 μm × 15 cm, nano HPLC capillary 144 column, 3 μm; Nikkyo Technos, Tokyo, Japan) at a constant flow rate of 300 nL/min by changing the 145 gradient: 5% B at 0 min, 40% B at 60 min, 100% B at 70 min, and 100% B at 80 min. The Orbitrap 146 Fusion was operated in positive ion mode with nanospray voltage set at 2.0 kV, source temperature at 147 with default charge set to 2, maximum ion injection time of 50 ms, a resolution of 120,000, and 149 automatic gain control (AGC) value of 4e5. The MS1 scan cycle was followed by MS2 scans (ion trap) 150 of the most intense ions fulfilling predefined selection criteria (AGC: 1e5; maximum ion injection 151 time: 200 ms; isolation window: 1.6 m/z; spectrum data type: centroid; exclusion of unassigned, singly, 152 and >5 charged precursors; peptide match preferred; exclude isotopes on; dynamic exclusion time: 10 153 s) for 3 sec. The HCD collision energy was set to 30% normalized collision energy. 154 with up to two missed cleavages; (b) precursor mass tolerance of 10 ppm; (c) fragment mass tolerance 160 of 0.6 Da; (d) carbamidomethylation of cysteine as a fixed modification; (e) acetylation of protein N 161 terminus, oxidation of methionine, and biotinylation of lysine as variable modifications. Peptides were 162 filtered at a false discovery rate (FDR) of 1% using the Percolator node of Proteome Discoverer. 163

TMT-Based Quantitative Proteomics of Surface-Biotinylated Cells 165
EGF-stimulated and surface-biotinylated HeLa cells were lysed in guanidine buffer as described above. 166 After heating and sonication, proteins (100 µg each) were purified by methanol-chloroform 167 precipitation and resuspended in 20 µL of 0.1% RapiGest SF in 50 mM triethylammonium bicarbonate. 168 After sonication and heating at 95°C for 10 min, the proteins were digested with 1 µg trypsin/Lys-C 169 mix (Promega) at 37 °C overnight. After heating at 95°C for 10 min, the digested peptides from each 170 sample were labeled with 0.1 mg of TMT 11-plex reagents (Thermo Fisher Scientific) and pooled. The  Normalization was performed such that the total sum of the abundance values for each TMT channel 191 over all peptides was the same. Cell lysates prepared in guanidine buffer as described above were subjected to methanol-chloroform 202 precipitation and resuspended in SDS-sample buffer. The proteins (10 µg each) were separated by 203 electrophoresis on a 9% polyacrylamide gel and transferred onto a PVDF membrane. After blocking 204 with Blocking One (Nacalai Tesque) for 1 h at room temperature, the membrane was incubated with 205 primary antibodies in TBS-T (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.05% Tween-20) 206 containing 5% Blocking One overnight at room temperature. After washing twice with TBS-T for 10 207 min, the membrane was incubated with HRP-linked secondary antibodies for 1 h and then washed 208 twice with TBS-T for 10 min. To detect biotinylated proteins, the transferred membrane was blocked 209 with Blocking One and incubated with HRP-linked streptavidin (BioLegend, #405210, 1:10,000) in 210 TBS-T containing 5% Blocking One for 1 h. Protein bands on the membrane were detected using the 211 ChemiDoc Touch Imaging System (Bio-Rad) after incubation with Clarity Western ECL substrate 212 (Bio-Rad). Antibodies and dilutions used in this study were as follows: anti-STING (D2P2F) rabbit 213 mAb (Cell Signaling Technology, #13647, 1:10,000); anti-EGF receptor (D38B1) XP rabbit mAb 214 (Cell Signaling Technology, #4267, 1:1,000); anti-phospho-ERK1/2 (Thr202/Tyr204) (D13.14.4E) 215 XP rabbit mAb (Cell Signaling Technology, #4370, 1:2,500); anti-α-tubulin (DM1a) mouse mAb 216 (Sigma-Aldrich, T6199, 1:4,000); anti-rabbit IgG, HRP-linked antibody (Cell Signaling Technology, 217 these cells and its ability to biotinylate cellular proteins by adding 500 µM biotin to the culture medium 225 for 10 min ( Figure S1). This biotinylated sample was used throughout the experiments to optimize the 226 enrichment workflow for biotinylated peptides. 227 In the original workflow, cellular proteins were extracted with guanidine hydrochloride together 228 with one-step reduction and alkylation, and then purified by methanol-chloroform precipitation. After 229 digestion with trypsin in RapiGest surfactant, Pefabloc SC was added to inhibit trypsin activity. peptides. Therefore, we attempted to optimize each step of the enrichment process to reduce these 236 interference ions and efficiently elute biotinylated peptides from Tamavidin 2-REV. 237 First, to examine the effect of surfactants that solubilize precipitated proteins and promote enzymatic 238 digestion (31-33), we compared RapiGest (34) with a phase-transfer surfactant (PTS) composed of 239 sodium deoxycholate (SDC) and sodium lauroylsarcosinate (SLS) (24-26). As a result, there were 240 almost no differences in MS1 chromatograms between RapiGest and PTS ( Figure 1A,B, Condition 1 241 vs. Condition 2); however, we started using PTS because it is more cost-effective than RapiGest.

Optimization of Elution Conditions for Biotinylated Peptides from Tamavidin 2-REV 268
Next, we attempted to optimize the conditions for competitive elution of biotinylated peptides from 269 the Tamavidin 2-REV beads. First, we focused on the temperature at which biotinylated peptides were 270 eluted from the beads. In the original workflow, biotinylated peptides were eluted with the biotin 271 solution at 95 °C. We therefore compared four temperatures: 4, 37, 56, and 95 °C. Among them, the 272 highest number of biotinylated peptides was identified at 37 °C (Figure 2A). The interference ions 273 were increased with the increase in the elution temperature, suggesting that these interference ions 274 were non-specifically eluted from the beads by heating ( Figure S2). 275 be reduced by adding a step of 'mock' elution with biotin-free buffer immediately before elution with 277 the biotin solution. After the beads were incubated with biotin-free buffer for 15 min at 37 °C, the 278 buffer was removed and biotinylated peptides were eluted from the beads with the biotin solution at 279 37 °C. As a result, non-biotinylated peptides were efficiently removed from the eluate, which increased 280 the ratio of biotinylated to non-biotinylated peptides ( Figure 2B). 281 The previous crystal structure analysis had revealed that Tamavidin 2 forms multiple hydrogen 282 bonds with biotin (38), and therefore we considered the possibility that biotinylated peptides could be 283 efficiently eluted by increasing the salt concentration to weaken the hydrogen bonds. After the mock 284 elution, biotinylated peptides were eluted from the beads at 37 °C with the biotin solutions containing 285 various concentrations of NaCl. As a result, the highest number of biotinylated peptides was identified 286 around 500 mM NaCl ( Figure 2C). On the basis of these results of optimizing the elution conditions, 287 we eluted biotinylated peptides from Tamavidin 2-REV beads with the biotin solution containing 500 288 mM NaCl at 37 °C after the mock elution. 289

Comparison of Biotinylated Peptides Enriched by the Original and Optimized Workflows 290
To evaluate the optimized workflow, biotinylated peptides enriched by the original and optimized 291 workflows from 300 µg proteins of RAW264.7 cells expressing TurboID-STING were compared by 292 LC-MS/MS analysis. MS1 chromatograms showed that many peaks of interference ions in the original 293 workflow were greatly reduced in the optimized workflow ( Figure 3A). Furthermore, the number of 294 identified biotinylated peptides was increased by more than two-fold after the optimization, and the 295 contamination by non-biotinylated peptides was reduced to almost one-third, resulting in 296 approximately 95% of all identified peptides being biotinylated ( Figure 3B). Thus, we have established 297 an optimized workflow to efficiently and specifically enrich biotinylated peptides using Tamavidin 2-298 digestion with trypsin (39, 40). Previous studies of cross-linking MS using lysine-reactive cross-linkers 303 (e.g., DSSO) have shown that sequential digestion with Lys-C or other non-lysine-specific proteases 304 (e.g., Glu-C) followed by trypsin increases the coverage of identified cross-linked peptides (41, 42). 305

REV. 299
We therefore examined three combinations of different proteases to digest biotinylated proteins from 306 TurboID-STING-expressing cells: trypsin alone, Lys-C followed by trypsin, and Glu-C followed by 307 trypsin. First, we examined whether PTS buffer used to solubilize precipitated proteins was compatible 308 with digestion with these proteases, and confirmed that each protease cleaved at specific residues; that 309 is, > 97% of all cleavages by trypsin occurred at lysine or arginine residues, >96% of all cleavages by 310 Lys-C occurred at lysine residues, and 97% of all cleavages by Glu-C occurred at aspartate or 311 glutamate residues ( Figure S3  For example, many of identified biotinylation sites were shared by the two digestions ( Figure 4A,B). 316 On the other hand, sequential digestion with Glu-C and trypsin yielded many unique biotinylation sites 317 that were not identified by trypsin alone or by Lys-C and trypsin, although the total number of 318 identified biotinylation sites was less ( Figure 4A,B). 319 Consistent with the endoplasmic reticulum (ER) localization of STING at the steady state, many 320 biotinylation sites identified by the three digestions were located on ER proteins (Table S1). In the 321 case of an ER protein Reticulon-4/Nogo (44, 45), while four biotinylation sites were identified by 322 digestion with trypsin alone, six other biotinylation sites were identified by sequential digestion with 323 Glu-C and trypsin ( Figure 4C). All 10 biotinylation sites were located in the N-terminal cytoplasmic 324 Collectively, these results indicate that digestion with trypsin alone is most effective in identifying 326 peptides with biotinylated lysine residues. In addition, sequential digestion with Glu-C and trypsin can 327 identify further biotinylation sites and provide more detailed topological information of membrane 328 proteins.

Biotinylated Peptide Enrichment 332
Finally, quantitative cell surface proteome profiling was performed by applying this optimized 333 workflow. Selective biotinylation of surface-exposed primary amines on cell surface proteins is one of 334 the most frequent approaches, and the membrane-impermeable reagent Sulfo-NHS-biotin is typically 335 used (40). We examined changes in the surface proteome of HeLa cells stimulated with epidermal 336 growth factor (EGF), which is well known to induce the internalization and degradation of EGF 337 receptor (EGFR) (46). First, we confirmed that EGF stimulation of HeLa cells induced rapid and 338 transient phosphorylation of ERK1/2 and gradual degradation of EGFR, as previously reported ( Figure  339 S4A) (47, 48). Next, surface biotinylation was performed at four time points after EGF stimulation in 340 multiple biological replicates, followed by digestion of cellular proteins with trypsin. After TMT 341 labeling and mixing of digested peptides, biotinylated peptides were enriched using Tamavidin 2-REV 342 ( Figure 5A). In total, 7,996 biotinylated peptides were identified and quantified by LC-MS/MS 343 analysis (Table S2)