Monitoring Protein Import into the Endoplasmic Reticulum in Living Cells with Proximity Labeling

The proper trafficking of eukaryotic proteins is essential to cellular function. Genetic, environmental, and other stresses can induce protein mistargeting, and in turn threaten cellular protein homeostasis. Current methods for measuring protein mistargeting are difficult to translate to living cells, and thus the role of cellular signaling networks in stress-dependent protein mistargeting processes, such as ER pre-emptive quality control (ER pQC), are difficult to parse. Herein, we use genetically encoded peroxidases to characterize protein import into the endoplasmic reticulum (ER). We show that the ERHRP/cytAPEX pair provides good selectivity and sensitivity for identifying protein mistargeting, using the known ER pQC substrate transthyretin (TTR). Although ERHRP labeling induces formation of detergent-resistant TTR aggregates, this is minimized by using low ERHRP expression, without loss of labeling efficiency. cytAPEX labeling recovers TTR that is mistargeted as a consequence of Sec61 inhibition or ER stress-induced ER pQC. Furthermore, we demonstrate that stress-free activation of the ER stress-associated transcription factor ATF6 recapitulates the TTR import deficiency of ER pQC. Hence, proximity labeling is an effective strategy for characterizing factors that influence ER protein import in living cells.


INTRODUCTION
Eukaryotic cells contain diverse subcellular compartments. Proteins must traffic to the correct environment to allow proper function 1 . One third of the nascent proteome is translocated into the endoplasmic reticulum (ER), entering the secretory pathways. This includes proteins that are resident in the ER, the Golgi, and the plasma membrane, and proteins that are secreted into the extracellular space. While some short proteins can be translocated across the ER membrane post-translationally, most secretory proteins undergo co-translational translocation.
For both pathways, the Sec61 complex serves as the translocation channel 2 . Luminal secretory proteins are targeted via a hydrophobic N-terminal signal sequence (or signal peptide), while membrane proteins are recognized via their transmembrane (TM) domains. This recognition is governed by a series of quality control steps that ensure targeting of signal sequence-and TMcontaining proteins, while excluding proteins that lack a targeting sequence [3][4][5] . In addition to translocation of nascent peptides, other channels serve to retrotranslocate proteins from the ER to the cytosol, usually in response to stress. Misfolded proteins can be sent to the cytosol for proteasomal degradation through ER-Associated Degradation (ERAD) 6,7 , using Hrd1 or derlin protein channels. ER stress upregulates ERAD, lowering the ER misfolded protein burden. ER stress can also protect the ER by decreasing nascent protein import, thought the pre-emptive quality control (ER pQC) pathway [8][9][10][11][12][13][14] . Mistargeting of secretory proteins to the cytosol can be deleterious for that compartment, however, and has been implicated in neurodegeneration and diabetes [15][16][17][18] . A lack of high-throughput methods for measuring protein targeting efficiency in cells has challenged full characterization of secretory protein mistargeting 19 .
The gold standard technique for quantifying protein import into the ER is the microsomal import assay based on proteinase K protection 5,20 . In this method, nascent protein is metabolically 35 S-labeled, usually as part of an in vitro transcription/translation formulation in the presence of freshly prepared pancreatic microsomes. Protein that does not enter the ER is proteolytically digested, and the protein of interest is immunopurified from the microsomal fraction and quantified by autoradiography. While this assay has been foundational for characterizing protein import into the ER, and can be adapted to living cells 21 , it is low-throughput and can only consider a single substrate at a time. Alternatively, ER protein import can be inferred from ER-specific maturation events, such as signal sequence/peptide cleavage or Nlinked glycosylation. Since these chemical modifications require ER luminal peptidases and glycosylases, protein maturation faithfully reports on protein import. However, the lack of maturation does not imply exclusion from the ER [22][23][24] . Some proteins may get their signal sequence cleaved slowly 25 or cleaved without targeting the ER 26 . Site occupancy of N-glycan sequons varies 27 , and many ER-targeting proteins are not glycosylated. Thus, ER-specific maturation events alone cannot be universally used to determine ER import efficiency.
Proximity-dependent labeling has emerged as a widely successful method for characterizing subcellular proteomes and protein-protein interactions. [28][29][30][31][32] In particular, peroxidase labeling through the generation of membrane-impermeable phenoxy radicals has high spatiotemporal resolution. In this case, a heme peroxidase is genetically encoded to target a specific environment. The cell is pre-incubated with a biotin-phenol, and promiscuous labeling of proximal proteins is triggered by brief (1 min.) exposure to hydrogen peroxide 33,34 . Herein, we consider whether proximity-dependent labeling can be used to identify mistrafficking of a model protein, transthyretin (TTR), in the secretory pathway (Figure 1). and c). b) The cytosolic peroxidase should preferentially label proteins in the cytosol compared to proteins that have entered the ER. c) By contrast, an ER luminal peroxidase should preferentially label ER-directed proteins.

Comparison of HRP and APEX2 for selective ER protein labeling.
We considered two peroxidases for ER labeling: horseradish peroxidase (HRP) and the commonly used APEX2 variant of soybean ascorbate peroxidase 33 . Although APEX2 has been more widely used in recent studies, HRP has two possible advantages. First, the reported labeling efficiency of HRP is greater than that of APEX2. Second, HRP that fails to enter the ER should be unable to fold correctly due to the reducing environment and low [Ca 2+ ] in the cytosol, while APEX2 retains activity in all environments 35,36 . Our ER HRP construct 37 harbors (N to C) the signal sequence of immunoglobulin κ, a hemagglutinin tag (HA tag), an N175S mutation that provides for enhanced thermodynamic stability 38 , and a C-terminal KDEL sequence for ER. Our ER APEX construct consists of (N to C) the preprotrypsin signal sequence, FLAG tag, APEX2, and finally a C-terminal KDEL (see Materials and Methods). HEK239T cells transiently transfected with either ER HRP or ER APEX were subjected to biotin-phenol (BP) labeling. The biotin-labeled proteome was purified by avidin recovery, separated by SDS-PAGE, and select ER and cytosolic marker proteins assayed by immunoblotting. To determine whether mistargeted ER peroxidases label proteins in the cytosol, we treated cells with the Sec61 translocon inhibitor mycolactone A/B (ML A/B) [39][40][41][42] , which prevents ER import of newly synthesized proteins (Figure 2a).

Figure 2
Compartmental selectivity of ER luminal peroxidases. a) Mycolactone A/B inhibits the Sec61 translocon by direct binding in the channel, causing newly synthesized proteins to be delivered to the cytosol. b) Representative immunoblots comparing protein labeling by ER HRP and ER APEX following ML A/B treatment (16 h, 25 nM) as indicated. † indicates background from previous blotting (BiP in HSPA1A), ‡ for background from the current primary antibodies (KDEL). Full blots and blotting order are provided in Figure S2. Grp94 is obscured by poor transfer in lanes 2 and 3 of lysate.
Neither peroxidase labels any of the assayed proteins in the absence of BP or H2O2 (Figure 2b avidin purified, lanes 7 and 10). Consistent with the peroxidase labeling preference for tyrosine 43 , we see sharp drops in signal intensity for ER HRP by anti-HA (YPYDVPDYA) and for ER APEX by anti-FLAG (DYKDDDDK). In the avidin purifications, hardly any signal is observed.
Both ER luminal peroxidases label the abundant ER chaperones Grp94 and BiP (Figure 2b, avidin purified, lanes 8 and 11). The abundant cytosolic proteins a-tubulin, b-actin, GAPDH, and cytosolic chaperone HSPA1A are faintly labeled by both, consistent with reported selectivities of peroxidase labeling in other compartments 29 . However, in each case the relative labeling of cytosolic proteins is higher for ER APEX than for ER HRP, while the labeling of the ER luminal proteins is greater for ER HRP as compared to ER APEX. This indicates that ER HRP is more selective for ER proteins under these conditions than ER APEX. ML A/B treatment only slightly decreases the total labeled protein yield, as determined by Ponceau S stain of avidin-purified eluates ( Figure S1, avidin purified), suggesting that the accumulated peroxidase prior to treatment is adequate for nearly full labeling activity. Both total For both ER peroxidases, more a-tubulin is labeled after ML A/B treatment, reflecting higher atubulin abundance in the lysates. Overall, ER HRP shows more specificity than ER APEX for ER over cytosolic proteins, and Sec61 inhibition does not decrease this specificity, indicating that ER HRP is a better choice for labeling ER proteins under conditions that interfere with ER import.

Orthogonality of cytosolic and ER luminal peroxidases.
We similarly compared the cytosolic and ER specificities of cyt APEX (APEX2 without the signal sequence, but containing a nuclear export sequence 33 ) and ER HRP. We transfected HEK293T cells with the chosen peroxidases, respectively, and performed proximity labeling. As expected, cyt APEX preferentially labels the cytosolic housekeeping proteins over the ER luminal chaperones, while ER HRP preferentially labels ER luminal chaperones over cytosolic proteins (Figure 3a). In each case, there is slight labeling of cytosolic proteins by ER HRP and ER proteins by cyt APEX. Although ERdj3 import is reported to be inefficient when it is overexpressed in HeLa cells 46 , we do not observe the characteristic preERdj3 band, suggesting that the cyt APEX-labeled fraction has successfully entered the ER and corresponds to cross-labeling, not deficient import.
Cytosolic protein labeling by cyt APEX cannot be ascribed to labeling of non-imported proteins, as the molecular weight migrations are consistent with signal sequence cleavage. Similarly, upon overexpression of eGFP ( cyt GFP) or an ER-stable moxGFP 47 with an ER targeting signal sequence ( ER GFP), we see cyt GFP solely labeled by cyt APEX, and ER GFP strongly preferentially labeled by ER HRP (Figure S3). Minor ER GFP labeling by cyt APEX is consistent with the reported residual mistargeting 48 , and could be partly due to modest unfolded protein response (UPR) activation upon ER GFP expression as indicated by elevated levels of BiP 8,49 . We further investigated global endogenous protein labeling by ER HRP and cyt APEX by quantitative proteomics. To observe how Sec61 inhibition globally affects protein localization, we included conditions with 25 nM ML A/B, and 1 µM MG132 to inhibit proteasomal degradation ( Figure S4).
These experiments were performed using Tandem Mass Tag (TMT) multiplexing, and identified proteins were required to appear in at least 3 out of 6 replicate experiments. Figure 3b shows the ER-to-cytosol labeling ratios (ER/cyt) of 2415 proteins in the absence of ML A/B and under proteasomal inhibition. These proteins are ranked based on their ER/cyt ratio, and as expected ER luminal proteins and most of the ER membrane proteins are enriched on the left side (top ER/cyt ranks), based on Gene Ontology [50][51][52] . ER membrane proteins can present residues on either the cytosolic or ER side of the membrane, and hence we see a broader distribution of labeling preference for proteins in this class. Interestingly, mitochondrial proteins are labeled to a similar extent by both peroxidases. Cytosolic proteins, as expected, are relatively excluded from the high ER/cyt region. Hence, it is clear that the ER HRP/ cyt APEX pair does differentiate between cytosolic and ER luminal proteins. The long ML A/B and MG132 treatment times (16 h) unfortunately led to substantial protein-level changes that prevented clear interpretation of the results as solely due to changes in ER import. However, ER proteins with known rapid turnover (gene names e.g. ATF6B, Sil1, APP, etc.) 53 demonstrate sharply decreased labeling by ER HRP following ML A/B treatment, consistent with their depletion from the ER ( Figure S4).

Figure 3:
Orthogonality of this assay with respect to the cytosol and the ER lumen. a) representative immunoblots of some abundant cytosolic and ER luminal proteins. b) ER-labeling over cytosol-labeling ratio of proteins identified with TMT-MuDPIT analysis. 2415 proteins were identified in at least 3 out of 6 replicates. Raw intensities are normalized with those of pyruvate carboxylase. Protein IDs are ranked based on their ER-to-cytosol ratios. ER-to-cytosol ratio were plotted in the logarithm with base of 2 scale. ER HRP and cyt APEX are annotated with hexagons. Some common ER luminal proteins and cytosolic proteins are annotated with diamonds. The four endogenous biotin-binding carboxylases in mitochondrial matrix are annotated in squares. If an identified protein belongs to corresponding gene ontology terms (cellular components, downloaded from AmiGo, 2021-08-22), it will be represented as a colored line in the bar graphs at the bottom.

TTR labeling by subcellularly localized peroxidases
Transthyretin (TTR) is among the most abundant plasma proteins, partially mislocalizes to the cytosol both basally and under stress 10,11,54 , and has over a hundred variants (including the wild-type) that have been demonstrated to be responsible for systemic amyloidoses 55 . As such, its biophysics and processing have been well investigated 56,57 . Hence, we chose TTR as a model ER substrate to evaluate our ability to use peroxidases to quantify ER import efficiency.
TTR is only 14 kDa; for convenience we overexpressed a dual FLAG-tagged TTR construct, where the FLAG tag follows the native TTR signal sequence and adds 2 kDa to the apparent migration on SDS-PAGE; native TTR is not expressed in HEK293T cells. TTR contains a sequon for N-linked glycosylation, but this sequon is only exposed under severe misfolding and hence typically has no glycan occupancy 58 . In the literature, relative cleavage yield of the signal peptide has been used to infer ER import efficiency 10 . We evaluated proximity labeling of TTR by cyt APEX and ER HRP with immunoblotting.
Our initial experiments found that ER HRP and TTR co-expression nearly eliminated TTR protein levels; this result was similar for multiple plasmid preparations ( Figure S5). Despite the low apparent TTR protein levels, we still saw robust TTR labeling; the labeled TTR was found solely as aggregates, despite elution conditions that included boiling the sample in reducing Laemmli buffer (2% SDS) for 20 min. This could be due to oxidation of TTR by the peroxidase, which is associated with increased TTR misfolding and amyloidogenesis 59 . To minimize TTR oxidation and aggregation by labeling, and to avoid CMV promoter competition between ER HRP and TTR, we optimized the ER HRP plasmid DNA concentration. 60 We found that even a 10% equivalence (160 ng plasmid per 6-cm dish) of the ER HRP plasmid induces a majority of TTR to form SDS-resistant aggregates and damage products ( Figure S6). Interestingly, the total labeled protein across the full molecular weight range is primarily TTR after ER HRP labeling ( Figure S7). Although there is TTR SDS-PAGE migration similar to the "immature", uncleaved TTR, its dependence on the 1-min H2O2 treatment immediately prior to lysis is not consistent with this band representing uncleaved TTR; rather, this must represent a TTR oxidative modification. It is possible that the low amount of ER HRP necessary for robust labeling is a consequence of the oxidizing environment of the ER. Ultimately, we found that although we could not eliminate TTR aggregation due to labeling by ER HRP, we could maximize yield of monomeric TTR at 2 equiv. Flag TTR : 0.01 equiv. ER HRP plasmid ratio ( Figure S8). This condition was used to compare cyt APEX and ER HRP labeling of TTR in cells.

Figure 4
Labeling of FLAG TTR by cytosolic and ER peroxidases. Immunoblots of SDS-PAGE separated lysates or avidin-purifications from HEK293T cells overexpressing FLAG TTR alone (2 equiv.) or with heme peroxidases (no HPX, no heme peroxidase; cyt APEX, 1 equiv.; ER HRP, 0.01 equiv.) as indicated, and treated with BP (500 µM, 30 min prior to harvest) or H2O2 (1 mM for 1 min) as indicated. † indicates background from previous blotting (b-actin in FLAG tag; ERdj3 in TTR), ‡ for background from the current primary antibodies (b-actin; BiP). Blotting order: for both lysate and avidin purification: b-actin (mouse, 700 nm), FLAG tag (mouse, 700 nm), ERdj3 (rabbit, 800 nm), BiP (rabbit, 800 nm), TTR (rabbit, 800 nm). Immature (imm.) and mature (mat.) TTR bands represent TTR with the signal sequence uncleaved or cleaved respectively. For slices with increased brightness, the brightness is increased uniformly over the entire blot without changes in contrast. increases the band intensity at the "immature" molecular weight, illustrating how evaluating import based on apparent signal sequence cleavage by immunoblot can be misleading.

Proximity Labeling Detects Protein Partitioning Independently from Maturation.
To better understand the limits of signal sequence cleavage as a proxy for import, we considered a TTR sequence that is relatively resistant to cleavage in the ER due to a proline Cterminal to the cleavage site, TTR G1P (Figure 5a, b) 61,62 . As expected, TTR WT is present in lysates solely in the mature form, and this form is labeled by ER HRP but not by cyt APEX (lane 6, "avidin", Figure 5c). By contrast, both immature and mature TTR G1P are observed in lysates, due to incomplete cleavage. However, although the mature form presumably has entered the ER, the immature TTR G1P could either have failed to enter the ER, or alternatively have entered the ER and failed to undergo signal peptidase cleavage. ER HRP labeling demonstrates that both forms are primarily present in the ER (lane 8, "avidin", Figure 5c), and thus the primary source of immature TTR G1P is properly translocated protein that did not get processed. Interestingly, both forms are also labeled slightly by cyt APEX (lane 4, "avidin", Figure 5c). The presence of the immature form in the cytosol could represent ER pQC due to TTR G1P -induced ER stress.
Indeed, BiP levels increase in response to TTR G1P transfection, indicated UPR activation and likely ER stress ( Figure S9). Furthermore, consistent with UPR activation, TTR G1P is aggregation-prone. While TTR WT is entirely soluble, both mature and immature TTR G1P primarily aggregate following ultracentrifugation of lysates at 77,000 x g for 4 h at 4 °C ( Figure S10). The basis of cytosolic labeling of mature TTR G1P by cyt APEX is less clear; it could potentially be due to retrotranslocation through ERAD 63 . These cytosolic populations of TTR G1P persist despite a lack of proteasomal inhibition, maybe resulting from its inherent instability and aggregation.
Nevertheless, proximity labeling provides starkly different and more accurate information about how this TTR variant partitions between the cytosol and ER than could have been obtained by looking at the mature and immature protein levels alone.

Figure 5
Comparison between maturation and proximity labeling to identify ER-directed protein localization. a) Signal sequence cleavage site in TTRs. b) Schematic of signal sequence processing and translocation. c) Representative immunoblot of SDS-PAGE separated lysates and avidin purifications from HEK293T cells expressing TTR WT or TTR G1P as indicated. Immature (imm.) and mature (mat.) TTR bands represent TTR with the signal sequence uncleaved or cleaved respectively.

Proximity Labeling Reports TTR Accumulation in Cytosol in Response to Sec61 Inhibition
Since cyt APEX and ER HRP labeling effectively report on TTR partitioning between the cytosol and the ER, we considered whether they could identify a change in TTR import in response to pharmacological treatment with ML A/B 39,64,65 . HEK293T cells overexpressing Flag TTR and either cyt APEX or ER HRP were treated with ML A/B for 16 h, alongside MG132 to inhibit proteasomal degradation of mistargeted protein 10,66,67 . MG132 on its own is adequate to induce a small amount of TTR labeling by cyt APEX. There is minimal cytosolic TTR in the presence of ML A/B without MG132, but the combined treatment substantially increases the cytosolic TTR population (lane 4, "avidin", Figure 6). In the ER, on the other hand, the amount of labeled TTR decreases significantly, indicating that 16-h ML A/B treatment prevents TTR from populating the ER to replace the secreted population. Decreased labeling of BiP suggests that this long treatment time might be adequate to deplete ER HRP as well. Critically, mycolactone

Proximity labeling enables mechanistic characterization of ER pQC
FLAG TTR is a known ER pre-emptive quality control (ER pQC) substrate 10,11 . ER stress induces a decrease in TTR import, and the mistargeted protein is rerouted to proteasomal degradation. In the presence of a proteasomal inhibitor, mistargeted TTR should instead accumulate in the cytosol. We co-transfected HEK293T cells with cyt APEX and FLAG TTR, and treated with thapsigargin (Tg), a SERCA inhibitor that canonically induces ER stress through ER calcium depletion and consequent decreased activity of calcium-dependent chaperones. Cells were also treated with MG132 to prevent degradation of mistargeted TTR. The activity of Tg is  Figures 7a, b). Labeling of TTR by ER HRP activity is not affected, consistent with the low reporter yield of pQC. Despite TTR being among the best validated pQC substrates, the mechanism by which TTR import decreases is still unclear. Because the proximity labeling assay allows import to be characterized in intact cells, we decided to investigate whether signaling pathways downstream of ER stress could modulate TTR import. ER stress remodels cellular proteostasis through the Unfolded Protein Response (UPR), which consists of three signaling pathways: PERK, ATF6, and IRE1/XBP1s. PERK function is mediated by activation of the integrated stress response, common to many cellular stresses, while the transcription factors ATF6 and XBP1s act primarily by promoting transcription of ER folding and degradation factors. The HEK293 DAX cell line allows stress-free and orthogonal small-molecule induction of XBP1s (in response to doxycycline, or Dox, treatment) and ATF6 (in response to trimethoprim, or TMP, treatment) 49 . HEK293 DYG cells express GFP/YFP under control of Dox/TMP and serve as a control line. These two cell lines were transfected with FLAG TTR, treated with TMP or Dox for 16 h, and the cytosolic FLAG TTR fractions measured through our proximity labeling approach (Figure 8a, b). We found that ATF6, but not XBP1s, induces an increase in cytosolic TTR in the absence of stress, suggesting that ATF6 might mediate ER pQC in response to stress (Figure 8c, d). Previously, it was found that Derlin association with the Sec61 translocon is critical for ER pQC 10 . However, none of the three Derlins are induced by ATF6 49 . Hence, other molecular models for pQC in response to ER stress now need to be considered. Interestingly, XBP1s activation decreases cytosolic TTR accumulation. XBP1s broadly upregulates trafficking factors, which could affect trafficking efficiency 49 . ) in the presence of MG132 (4.5 h, 1 µM) as indicated. ATF6 activation is confirmed by intense upregulation of BiP, while Hyou1, as a target of both the ATF6 and XBP1s arms of the UPR, is upregulated by both TMP and Dox in the HEK293 DAX line. ‡ indicates background signal (BiP; b-actin). c) Activation of the UPR-associated transcription factor ATF6, but not XBP1s, also induces FLAG TTR mistargeting. ** p < 0.01 and * p < 0.05 by two-way Student's t-test comparison to vehicle (n = 7).

Summary
We find that proximity labeling is an effective and facile technique to identify perturbations in the localization of ER-directed proteins in response to stress. It only requires the expression of genetically encoded peroxidases, and thus is compatible with living cells. In the case of TTR, proximity labeling faithfully recapitulates the known effects of Tg-induced pQC, and furthermore shows that ATF6 activation recapitulates the TTR import deficiency. Hence, this approach could be useful generally for identifying cellular pathways involved in protein mistargeting. In contrast to protease protection assays, proximity labeling has the further advantage of being a promiscuous labeling method. It thus can be potentially coupled to mass spectrometry to determine proteome-wide changes in import. Further investigations will characterize factors that influence the mistargeted ER-directed proteome.  Table S1.

Supporting Information
All constructs were subjected to analytical digest and sequenced (Retrogen) to confirm identity. Primers were purchased from Integrated DNA Technologies (IDT), and all enzyme purchased from New England Biolabs. The TTR G1P plasmid was cloned from the TTR expression plasmid using site-directed mutagenesis. DPBS. 100 mM H2O2 stock (100x) was freshly made by diluting 30% H2O2 (9.8 M, Fisher) with 1x DPBS. 20 After the 30-min incubation with biotin phenol, 30 µL 100x H2O2 stock was added into each dish to a final concentration of 1 mM, and dishes were agitated immediately after addition. Exactly 1 min after the H2O2 delivery, media were aspirated, and cells were washed three times with 3 mL 1x quencher solution. Cells were then scraped in 1 mL 1x quencher solution and pelleted at 4 °C, 700 x g for 5 min.

Human
Cell pellets may be deeply frozen in −80 °C freezer before lysis. Frozen eluate samples were boiled for 5 min and homogenized again before separation by SDS-PAGE (2 µL eluate used for silver stain). Gels were washed briefly twice in H2O and fixed in 30% ethanol (EtOH)/10% AcOH in H2O for 30 min and overnight. Gels were then washed three times in 35% EtOH for at least 20 min each, sensitized in 0.02% sodium thiosulfate pentahydrate (Na2S2O3·5H2O, Fisher) for 2 min, washed three times in H2O for one min each, silver-stained in 0.2% AgNO3/0.076% formalin for 20 min to up to 16 hours. Stained gels were washed twice with H2O for 5 min and developed in 6% Na2CO3/0.05% formalin/0.0004% Na2S2O3·5H2O. Development was halted with 5% AcOH and gels were imaged on a transilluminator (UVP).

Mass spectrometry
Only MS grade organic solvents were used during sample preparation, except chloroform (CHCl3, certified ACS). Buffer A is 0.1% formic acid in 5% acetonitrile (ACN)/H2O. Buffer B is 0.1% formic acid in 80% ACN/H2O. Buffer C is 500 mM ammonium acetate in Buffer A. Sample preparation was 23 performed in low-bind Eppendorf tubes and reagents were mixed well by vortex mixing (before the addition of 1% Rapigest, Waters) or flicking. Eluate samples were brought to 100 µL with H2O, followed by addition of 300 µL MeOH. BP-labeled protein was precipitated by adding 100 µL CHCl3, and pelleted at 21,100x g for 15 min. Supernatants were removed carefully by aspiration. After clean-up procedure was repeated for another twice, protein pellets were air-dried, and resuspended in 3 µL 1% Rapigest in H2O. Resuspensions were brought to 50 µL with 100 mM HEPES, pH 8.0. Proteins were then reduced by 10 mM tris(2-carboxyethyl)phosphine (TCEP, Millipore Sigma) for 30 min at 37 °C, alkylated by 5 mM iodoacetamide (Millipore Sigma) for 30 min in dark at ambient temperature and digested by 5 µg/µL trypsin (Thermo Fisher Scientific) overnight at 37 °C with 600-rpm agitation. 50 µg of TMT (tandem mass tag) labels (Pierce) in 40 µL ACN were added accordingly to each sample and let incubate for 1 h at ambient temperature.
Labeling reaction was quenched with 0.4% ammonium bicarbonate (NH4HCO3) for 1 h at ambient temperature. Samples were then combined and concentrated to around 10 µL via vacuum centrifugation, followed by resuspension in 200 µL buffer A. The mixed sample was then acidified to pH <2.0 with formic acid (Acros) and stored at -80°C freezer. Before being loaded onto a triphasic loading column for multiple dimension protein identification technology (MuDPIT) analysis, the sample was heated at 37 °C for 1 hour and hard spun for 30 min to precipitate Rapigest and leftover agarose avidin bead. Clarified sample was transferred to a new low-bind tube. This process may be repeated for at least three times.
Mycolactone A/B 4-plex samples were analyzed using two dimensional LC-MS/MS on an LTQ Orbitrap Velos hybrid mass spectrometer (Thermo) interfaced with an Easy-nLC 1000 (Thermo) according to standard MuDPIT protocols 70 . Analyses were performed using a twelve-cycle chromatographic run, with progressively increasing ammonium acetate salt bumps injected before each cycle (0% C, 10% C, 20% C, 30% C, 40% C, 50% C, 60% C, 70% C, 80% C, 100% C, 90% B+10% C, 90% B+10% C; balance of each with buffer A, if needed; one replicate had two extra 90% C+10% B), followed by ACN gradient (For the 0% C bump run: 5 min from 1% B to 6% B, 60 min to 45% B, 15 min to 100% B, 5 min at 100% B, 5 min to 1% B, 10 min at 1%, 100 min in total, 300 nL/min flow rate; For the rest runs: 5 min from 1% B to 6% B, 90 min to 45% B, 20 min to 100% B, 5 min at 100% B, 5 min to 1% B, 20 min at 1%, 145 min in total; 400 nL/min flow rate). Eluted peptides were ionized by electrospray (3.0 kV) and scanned from 110 to 2000 m/z in the Orbitrap with resolution 30000 in data dependent acquisition mode. The top ten peaks from each full scan were fragmented by HCD using a normalized collision energy of 38%, a 100 ms activation time, and a resolution of 7500. Dynamic exclusion parameters were 1 repeat count, 30 ms repeat duration, 500 exclusion list size, 60 s exclusion duration, and 1.50 Da exclusion width. MS1 and MS2 spectra were searched with MSFragger (with FragPipe [71][72][73] ) against a combined database of Uniprot human proteome database (downloaded with FragPipe, 2021-07-09) and heme peroxidases and AVID_CHICK, and reverse sequences for each entry as the decoy set, with common contaminants (e.g. keratin, porcine trypsin, etc.). Closed searches were allowed for static modification of cysteine residues (57.02146 Da, carbamidomethylation), N-termini 25 and lysine residues (229.1629 Da, TMT-tagging), half tryptic peptidolysis specificity, and mass tolerance of 20 ppm for precursor mass and 20 ppm for product ion masses. Spectral matches were assembled and filtered with a FDR of 0.01. Average intensities of identified proteins from each TMT channel were normalized against the intensities of pyruvate carboxylase.