News

Thank you for visiting Nature.com. The browser version you are using has limited CSS support. For the best experience, we recommend that you use an updated browser (or disable Compatibility Mode in Internet Explorer). In the meantime, to ensure continued support, we will render the site without styles and JavaScript.
Enzymatic proximity labeling methods based on activated esters or phenoxy radicals are widely used to map subcellular proteomes and protein interactors in living cells. However, activated esters are less reactive, resulting in a wide labeling radius, and phenoxy radicals generated by peroxide treatment can interfere with redox pathways. Here we report a proximity labeling dependent photoactivation (PDPL) method developed by genetically linking the miniSOG photosensitizer protein to a protein of interest. Triggered by blue light and controlled by exposure time, singlet oxygen is generated and then spatiotemporally resolved labeling of histidine residues by the aniline probe is achieved. We demonstrate its high fidelity through organelle-specific proteome mapping. A side-by-side comparison of PDPL with TurboID shows a more specific and comprehensive proteomic coverage of PDPL. Next, we applied PDPL to the disease-associated transcriptional coactivator BRD4 and E3 Parkin ligase and found previously unknown interactors. By overexpression screening, two unknown substrates, Ssu72 and SNW1, were identified for Parkin, whose degradation is mediated by the ubiquitination-proteasome pathway.
Accurate characterization of protein networks underlies many fundamental cellular processes. Therefore, highly accurate spatiotemporal mapping of protein interactions will provide a molecular basis for deciphering biological pathways, disease pathology, and disrupting these interactions for therapeutic purposes. To this end, methods capable of detecting temporal interactions in living cells or tissues are highly desirable. Affinity Purification Mass Spectrometry (AP-MS) has historically been used to identify binding partners of proteins of interest (POIs). With the development of quantitative proteomics methods, Bioplex3.0 was created, the largest database of protein networks based on AP-MS. Although AP-MS is very powerful, the cell lysis and dilution steps in the workflow are biased towards weak and transient binding interactions and introduce post-lysis artifacts such as spurious interaction pairs that lack compartmentalization prior to lysis.
To address these issues, unnatural amino acids (UAA) with crosslinking groups and enzymatic nearby labeling (PL) platforms (eg APEX and BioID)5 have been developed. Although the UAA method has been successfully applied in many scenarios and provides information on direct protein adhesives, optimization of the UAA insertion site is still required. More importantly, it is a stoichiometric labeling method that lacks a catalytic reversal of labeling events. In contrast, enzymatic PL methods, such as the BioID method, fuse the engineered biotin ligase to POI7, which subsequently activates biotin to form a reactive biotinyl-AMP ester intermediate. The enzyme thus catalyzes and releases an activated biotin “cloud” that labels proximal lysine residues. However, BioID requires more than 12 hours to obtain a sufficient labeled signal, which precludes its use with temporal resolution. Using directed evolution based on yeast display, TurboID was designed based on BioID to be more efficient, allowing efficient labeling with biotin within 10 minutes, allowing more dynamic processes to be studied. Because TurboID is highly active and endogenous biotin levels are sufficient for low-level labeling, background labeling becomes a potential problem when highly enhanced and timed labeling is required by the addition of exogenous biotin. In addition, activated esters are poorly reactive (t1/2 ~5 min), which can lead to a large labeling radius, especially after saturation of neighboring proteins with biotin 5. In another approach, genetic fusion of engineered ascorbate peroxidase (i.e. biotin-phenol radicals and allows protein labeling within one minute9,10. APEX is widely used to identify subcellular proteomes, membrane protein complexes, and cytosolic signaling protein complexes11,12. However, the need for high concentrations of peroxides can affect redox proteins or pathways, disrupting cellular processes.
Thus, a new method capable of generating more reactive labeled-radius suppression species with high spatial and temporal accuracy without significantly disrupting cellular pathways will be an important addition to existing methods. Among the reactive species, singlet oxygen aroused our attention due to its short lifetime and limited diffusion radius (t1/2 < 0.6 µs in cells)13. Among the reactive species, singlet oxygen aroused our attention due to its short lifetime and limited diffusion radius (t1/2 < 0.6 µs in cells)13. Среди активных форм наше внимание привлек синглетный кислород из-за его короткого времени жизни и ограниченного радиуса диффузии (t1/2 < 0,6 мкс в клетках)13. Among the active forms, singlet oxygen attracted our attention due to its short lifetime and limited diffusion radius (t1/2 < 0.6 µs in cells)13.在活性物质中，单线态氧因其寿命短和扩散半径有限（细胞中t1/2 < 0.6 µs）而引起了我们的注意13。 1/2 < 0.6 µs）而引起了我们的注意13 Среди активных форм наше внимание привлекает синглетный кислород из-за его короткого времени жизни и ограниченного радиуса диффузии (t1/2 < 0,6 мкс в клетках). Among active forms, singlet oxygen attracts our attention because of its short lifetime and limited diffusion radius (t1/2 < 0.6 μs in cells). Singlet oxygen has been reported to randomly oxidize methionine, tyrosine, histidine and tryptophan, making it polar 14,15 for attachment to amine or thiol based probes16,17. Although singlet oxygen has been used to label subcellular compartment RNA, strategies for repurposing endogenous POI proximity markers remain unexplored. Here, we present a platform called photoactivation-dependent proximity labeling (PDPL), where we use blue light to illuminate POIs fused with a miniSOG photosensitizer and trigger singlet oxygen generation to oxidize proximal residues, followed by amine-containing modifications to oxidize chemical probes into intermediate living cells. . We tested a group of chemical probes to maximize tag specificity and identified modification sites using an open proteomics workflow. A side-by-side comparison of PDPL with TurboID shows a more specific and comprehensive proteomic coverage of PDPL. We applied this approach to organelle-specific markers of the subcellular proteome and general proteome identification of binding partners for the cancer-associated epigenetic regulatory protein BRD4 and the Parkinson’s disease-associated E3 ligase Parkin, which confirmed both a known and an unknown network of protein interactions. . The ability of PDPL to recognize E3 substrates in large protein complexes represents a situation where recognition of indirect binders is required. Two unknown parkin substrates mediated by ubiquitination-proteasome have been confirmed in situ.
Photodynamic therapy (PDT)19 and chromophore-assisted laser inactivation (CALI)20, in which light irradiation with photosensitizers generates singlet oxygen, can inactivate target proteins or cause cell death. Since singlet oxygen is a highly reactive substance with a theoretical diffusion distance of about 70 nm, spatially limited oxidation around the photosensitizer can be controlled. Based on this concept, we decided to use singlet oxygen to achieve close labeling of protein complexes in living cells. We have developed a PDPL chemoproteomic approach to fulfill four functions: (1) to catalyze the generation of active singlet oxygen similar to the PL enzymatic approach; (2) provide time-resolved labeling upon light initiation; (3) by alteration (4) Avoid using endogenous cofactors (such as biotin) to reduce background, or use highly perturbing exogenous reagents (such as peroxides) to minimize cell exposure to environmental stress.
Photosensitizers can be divided into two categories including small molecular weight fluorophores (eg rose bengal, methylene blue)22 and genetically encoded small proteins (eg miniSOG, KillerRed)23. To achieve a modular design, we developed the first generation PDPL platform by adding photosensitizer (PS) proteins to POI24,25 (Figure 1a). When irradiated with blue light, singlet oxygen oxidizes proximal nucleophilic amino acid residues, resulting in an umpolung polarity that is electrophilic and can further react with amine probe nucleophiles16,17. The probe is designed with an alkyne handle to allow click chemistry and pull down for LC/MS/MS characterization.
Schematic illustration of labeling of protein complexes mediated by miniSOG. When exposed to blue light, cells expressing miniSOG-POI generate singlet oxygen, which modifies interacting proteins but not non-binding proteins. Intermediate products of photooxidation are intercepted by relay labels of the amine chemical probe to form covalent adducts. The alkynyl group on the chemistry probe allows click chemistry for enrichment by pull-down followed by LC-MS/MS quantitation. b Chemical structure of amine probes 1-4. c Representative fluorescent gel analysis of mitochondrial localized miniSOG-mediated proteomic markers using probes 1-4 and relative quantification based on gel densitometry. The signal-to-background ratio of chemical probes was assessed using negative control experiments excluding blue light or using HEK293T cells without miniSOG expression. n = 2 biologically independent samples. Each dot represents a biological replica. d Representative detection and quantification of PDPL using optimized probe 3 in the presence or absence of the indicated PDPL components like c. n = 3 biologically independent samples. Each dot represents a biological replica. Centerlines and whiskers represent the mean and ± standard deviation. CBB: Coomassie Brilliant Blue. e Confocal imaging of singlet oxygen with far-red Si-DMA stain. Scale bar: 10 µm. Gel imaging and confocal experiments were independently repeated at least twice with similar results.
We first tested the ability of the mature photosensitizers miniSOG26 and KillerRed23, stably expressed in HEK293T, to mediate propargylamine labeling of the proteome as a chemical probe (Supplementary Fig. 1a). Gel fluorescence analysis showed that whole proteome labeling was achieved using miniSOG and blue light irradiation, while no visible labeling product was observed with KillerRed. To improve the signal-to-background ratio, we then tested a set of chemical probes containing aniline (1 and 3), propylamine (2), or benzylamine (4). We observed that HEK293T cells themselves had a higher background signal compared to no blue light, possibly due to the endogenous riboflavin photosensitizer, flavin mononucleotide (FMN) 27 . Aniline-based chemical probes 1 and 3 gave better specificity, with HEK293T stably expressing miniSOG in mitochondria displaying an >8-fold increase in signal for probe 3, while probe 2 used in the RNA-labeling method CAP-seq only displaying ~2.5-fold signal increase, likely due to different reactivity preferences between RNA and protein (Fig. 1b, c). Aniline-based chemical probes 1 and 3 gave better specificity, with HEK293T stably expressing miniSOG in mitochondria displaying an >8-fold increase in signal for probe 3, while probe 2 used in the RNA-labeling method CAP-seq only displaying ~2.5- fold signal increase, likely due to different reactivity preferences between RNA and protein (Fig. 1b, c). Aniline-based chemical probes 1 and 3 showed better specificity: HEK293T, which stably expresses miniSOG in mitochondria, shows more than an 8-fold increase in signal for probe 3, while probe 2, used in the CAP-seq RNA labeling method, only shows ~2.5-fold signal increase, probably due to different reactivity preferences between RNA and protein (Fig. 1b, c).基于苯胺的化学探针1 和3 具有更好的特异性，HEK293T 在线粒体中稳定表达miniSOG，探针3 的信号增加> 8 倍，而用于RNA 标记方法CAP-seq 的探针2 仅显示~2.5-倍信号增加，可能是由于RNA 和蛋白质之间的不同反应偏好（图1b，c）。基于 苯胺 的 化学 探针 1 和 3 具有 更 的 特异性 ， hek293t 在 粒体 中 稳定 表达 minisog ， 探针 3 的 信号 增加 增加> 8 倍 而 于 于 rna 标记 cap-eq 的 2 仅 ~ ~ ~ ~ 2.5-倍信号增加，可能是由于RNA Aniline-based chemical probes 1 and 3 had better specificity, HEK293T stably expressed miniSOG in mitochondria, and probe 3 had over 8-fold increase in signal, while probe 2 for the CAP-seq RNA labeling method showed only ~2.5 -fold increase. in the signal, probably due to different reaction preferences between RNA and protein (Fig. 1b, c). In addition, probe 3 isomers and hydrazine probes (probes 5, 6, 7) were tested, confirming the optimization of probe 3 (Supplementary Fig. 1b,c). Similarly, in-gel fluorescence analysis revealed other optimized experimental parameters: irradiation wavelength (460 nm), chemical probe concentration (1 mM), and irradiation time (20 min) (Supplementary Fig. 2a–c). Omitting any component or step in the PDPL protocol resulted in significant signal reversal to the background (Fig. 1d). Notably, protein labeling was significantly reduced in the presence of sodium azide or trolox, which are known to quench singlet oxygen. The presence of D2O, which is known to stabilize singlet oxygen, enhances the labeling signal. To investigate the contribution of other reactive oxygen species to labeling, mannitol and vitamin C were added to establish hydroxyl and superoxide radical scavengers, respectively, 18, 29, but they were not found to reduce labeling. Addition of H2O2, but not illumination, did not result in labeling (Supplementary Fig. 3a). Fluorescence singlet oxygen imaging with Si-DMA probes confirmed the presence of singlet oxygen in the HEK293T-miniSOG wire, but not in the original HEK293T wire. In addition, mitoSOX Red could not detect superoxide production after illumination (Fig. 1e and Supplementary Fig. 3b) 30. These data strongly suggest that singlet oxygen is the main reactive oxygen species responsible for subsequent proteomic labeling. Cytotoxicity of PDPL was assessed including blue light irradiation and chemical probes, and no significant cytotoxicity was observed (Supplementary Fig. 4a).
In order to study the labeling mechanism and enable proteomic identification of protein complexes using LC-MS/MS, we first need to determine which amino acids are modified and the delta mass of probe labels. Methionine, histidine, tryptophan and tyrosine have been reported to be modified by singlet oxygen14,15. We integrate the TOP-ABPP31 workflow with the unbiased open search provided by the FragPipe computing platform based on MSFragger32. After singlet oxygen modification and chemical probe labeling, click chemistry was performed using a biotin reduction label containing a cleavable linker, followed by neutravidin stretching and trypsin digestion. The modified peptide, still bound to the resin, was photocleaved for LC-MS/MS analysis (Figure 2a and Supplementary Data 1). A large number of modifications occurred throughout the proteome with over 50 peptide map (PSM) matches listed (Fig. 2b). Surprisingly, we only observed modification of histidine, probably due to the higher reactivity of oxidized histidine towards aniline probes than other amino acids. According to the published mechanism of histidine oxidation by singlet oxygen,21,33 the proposed delta-mass structure of +229 Da corresponds to the adduct of probe 3 with 2-oxo-histidine after two oxidations, while +247 Da is the hydrolysis product of +229 Da (Supplementary Fig. 5). The evaluation of the MS2 spectrum showed a high reliability of identification of most of the y and b ions, including the identification of modified fragment ions (y and b) (Fig. 2c). Context analysis of the local sequence of PDPL-modified histidines revealed a moderate motif preference for small hydrophobic residues at ±1 positions (Supplementary Fig. 4b). On average, 1.4 histidines were identified per protein, and the sites of these markers were determined by solvent accessible surface area (SASA) and relative solvent availability (RSA) analysis (Supplementary Fig. 4c,d).
An unbiased workflow for studying residual selectivity using the FragPipe computing platform powered by MSFragger. Cleavable linkers are used in Click chemistry to allow photocleavage of modified peptides from streptavidin resin. An open search was launched to identify numerous modifications, as well as relevant remnants. b Assign the mass of modifications occurring throughout the proteome. Peptide mapping PSM. c MS2 spectral annotation of histidine sites modified with probe 3. As a representative example, a covalent reaction with probe 3 added +229.0938 Da to the modified amino acid. d Mutation assay used to test for PDPL markers. PRDX3 (H155A, H225A) and PRDX1 (H10A, H81A, H169A) were transfected with wild-type plasmids for anti-Flag detection. e The synthetic peptide was reacted with purified miniSOG in the presence of probe 3 and the corresponding products with Δm +247 and +229 were noted in the LC-MS spectrum. f In vitro protein-to-protein interactions modeled with miniSOG-6xHis-tag and anti-6xHis antibody. Antibiotin (streptavidin-HRP) and anti-mouse Western blot analysis of miniSOG-6xHis/anti-6xHis antibody complexes labeled with probe 3, depending on the time of exposure to light. Labels for individual proteins are expressed in the corresponding molecular weight: LC antibody light chain, HC antibody heavy chain. These experiments were independently repeated at least twice with similar results.
For biochemical verification of the labeling site, PRDX3 and PRDX1 identified by mass spectrometry were changed from histidine to alanine and compared to wild type in transfection assays. The PDPL results showed that the mutation significantly reduced labeling (Fig. 2d). Meanwhile, the peptide sequences identified in the open search were synthesized and reacted in vitro with purified miniSOG in the presence of probe 3 and blue light, yielding products with a mass shift of +247 and +229 Da when detected by LC-MS (Fig. 2e). ). To test whether interacting proximal proteins could be labeled in vitro in response to miniSOG photoactivation, we designed an artificial proximity assay by interaction between the miniSOG-6xHis protein and an anti-His monoclonal antibody in vitro (Figure 2f). In this assay, we expected proximal labeling of antibody heavy and light chains with miniSOG. In fact, anti-mouse (recognizing the heavy and light chains of anti-6xHis-labeled antibody) and streptavidin Western blots showed strong biotinylation of the heavy and light chains. Notably, we noticed miniSOG autobiotinylation due to the 6xHis tag and cross-links between light and heavy chains, which may be related to the previously described gap between lysine and 2-oxo-histidine proximal response. In conclusion, we conclude that PDPL modifies histidine in a proximity dependent manner.
Our next goal was to characterize the subcellular proteome to test the specificity of in situ labeling. Therefore, we stably expressed miniSOG in the nucleus, mitochondrial matrix, or outer ER membrane of HEK293T cells (Fig. 3a). Gel fluorescence analysis revealed abundant labeled bands at three subcellular locations as well as different labeling patterns (Fig. 3b). Fluorescence imaging analysis showed high specificity of PDPL (Fig. 3c). The PDPL workflow was followed by click reactions with rhodamine dyes to delineate subcellular proteomes using fluorescence microscopy, and PDPL signals were colocalized with DAPI, mitochondrial trackers, or ER trackers, confirming the high fidelity of PDPL. For the three organelle locations, a side-by-side comparison of PDPL with TurboID using avidin western blot showed that PDPL was labeled more specifically compared to their respective controls. Under PDPL conditions, more labeled bands appeared, indicating more PDPL-labeled proteins (Supplementary Fig. 6a-d).
a Schematic representation of miniSOG-mediated organelle-specific proteome labeling. miniSOG targets the mitochondrial matrix via fusion to the N-terminal 23 amino acids of human COX4 (mito-miniSOG), the nucleus via fusion to H2B (nucleus-miniSOG), and Sec61β via the cytoplasmic side of the ER membrane (ER-miniSOG). Indications include gel imaging, confocal imaging, and mass spectrometry. b Representative gel images of three organelle-specific PDPL profiles. CBB Coomassie Brilliant Blue. c Representative confocal images of HEK293T cells stably expressing miniSOG with different subcellular localizations detected by antibody labeled V5 (red). Subcellular markers are used for mitochondria and ER (green). The PDPL workflow includes the detection of miniSOG (yellow) labeled subcellular proteomes using Cy3-azide click chemistry. Scale bar: 10 µm. d Volcanic plots of PDPL-tagged proteomes in various organelles quantified by unlabeled quantification (n = 3 independent biological experiments). Two-tailed Student’s t-test was used on volcano plots. HEK293T wild type was used as a negative control. Significantly changed proteins are highlighted in red (p < 0.05 and >2-fold ion intensity difference). Significantly changed proteins are highlighted in red (p < 0.05 and >2-fold ion intensity difference). Значительно измененные белки выделены красным цветом (p < 0,05 и >2-кратная разница в интенсивности ионов). Significantly altered proteins are highlighted in red (p < 0.05 and >2-fold difference in ion intensity).显着变化的蛋白质以红色突出显示（p < 0.05 和> 2 倍离子强度差异）。显着变化的蛋白质以红色突出显示（p < 0.05和> 2 Значительно измененные белки выделены красным цветом (p < 0,05 и > 2-кратная разница в ионной силе). Significantly altered proteins are highlighted in red (p < 0.05 and > 2-fold difference in ionic strength). Related proteins important for HEK293T-miniSOG but not important for HEK293T are shown in green. e Analysis of the specificity of proteomic datasets from experiments d. The total number of statistically significant proteins in each organelle (red and green dots) is marked at the top. Histograms show proteins localized in organelles based on MitoCarta 3.0, GO analysis and A. Ting et al. people. Separate datasets for mitochondria, nuclei, and ER. These experiments were independently repeated at least twice with similar results. Raw data are provided in the form of raw data files.
Encouraged by the gel and imaging results, label-free quantification was used to quantify the identified proteome in each organelle (Supplementary Data 2). Untransfected HEK293T was used as a negative control to subtract background markers. Volcano plot analysis displayed significantly enriched proteins (p < 0.05 and >2-fold ion intensity) as well as singleton proteins that are only present in miniSOG-expressing lines (Fig. 3d red and green dots). Volcano plot analysis displayed significantly enriched proteins (p < 0.05 and >2-fold ion intensity) as well as singleton proteins that are only present in miniSOG-expressing lines (Fig. 3d red and green dots). Анализ графика вулкана показал значительно обогащенные белки (p <0, 05 и > 2-кратная интенсивность ионов), а также одиночные белки, которые присутствуют только в линиях, экспрессирующих miniSOG (рис. 3d, красные и зеленые точки). Volcano plot analysis showed significantly enriched proteins (p<0.05 and >2-fold ion intensity) as well as single proteins that are only present in miniSOG-expressing lines (Fig. 3d, red and green dots).火山图分析显示出显着富集的蛋白质（p < 0.05 和>2 倍离子强度）以及仅存在于miniSOG 表达系中的单一蛋白质（图3d 红色和绿色点）。火山图 分析 显示 出 显着 的 蛋白质 （p <0.05 和> 2 倍 离子 强度） 仅 存 在于 在于 minisog 表达 系 的 单一 蛋白质 （图 3d 红色 绿色点。。。）））））））））） Анализ графика вулкана выявил значительно обогащенные белки (p <0, 05 и> 2x ионная сила), а также отдельные белки, присутствующие только в экспрессионной линии miniSOG (красные и зеленые точки на рис. 3d). Volcano plot analysis revealed significantly enriched proteins (p<0.05 and >2x ionic strength) as well as single proteins only present in the miniSOG expression line (red and green dots in Fig. 3d). Combining these data, we identified 1364, 461, and 911 statistically significant nuclear, mitochondrial, and ER outer membrane proteins, respectively. To analyze the accuracy of organelle-localized PDPL, we used MitoCarta 3.0, Gene Ontology (GO) analysis, and A. Ting et al. a data set8 was used for mitochondria, nucleus, and ER to test the organelle specificity of the detected proteins, corresponding to an accuracy of 73.4, 78.5, and 73.0% (Fig. 3e). The specificity of PDPL confirms that PDPL is an ideal tool for identifying organelle-specific proteomes. Notably, submitochondrial analysis of identified mitochondrial proteins showed that the trapped proteome was mainly distributed in the matrix and inner membrane (226 and 106, respectively), accounting for 91.7% (362) of the total number of identified mitochondrial proteins. a high level of PDPL was additionally confirmed (Supplementary Fig. 7a). Similarly, subnuclear analysis showed that the captured proteome was mainly distributed in the nucleus, nucleoplasm, and nucleolus (Supplementary Fig. 7b). Nuclear proteomic analysis with a nuclear localization signal peptide (3xNLS) showed similar accuracy to the H2B construct (Supplementary Fig. 7c–h). To determine the specificity of the PDPL marker, nuclear laminin A was chosen as a more discretely localized POI7 trap. PDPL identified 36 significantly enriched proteins, of which 12 proteins (30.0% including lamin A) were well-characterized lamin A interacting proteins annotated by the String database, with a higher percentage than the BioID method (122 proteins) 28 of 28. , 22.9%) 7. Our method identified fewer proteins, possibly due to limited labeling areas, which was made possible by more active singlet oxygen. GO analysis showed that the identified proteins were mainly located in the nucleoplasm (26), nuclear membrane (10), nuclear membrane (9), and nuclear pores (5). Collectively, these nuclear-localized proteins accounted for 80% of the enriched proteins, further demonstrating the specificity of PDPL (Supplementary Fig. 8a–d).
Having established the ability of PDPL to perform proximity marking in organelles, we then tested whether PDPL could be used to analyze POI binding partners. In particular, we sought to define PDPL analysis of cytosolic proteins, which are considered more difficult targets than their membrane-localized counterparts due to their highly dynamic nature. The bromodomain and extraterminal (BET) protein BRD4 has attracted our attention for its key role in various diseases 35, 36 . The complex formed by BRD4 is a transcriptional coactivator and an important therapeutic target. By regulating the expression of c-myc and Wnt5a transcription factors, BRD4 is thought to be a key determinant of acute myeloid leukemia (AML), multiple myeloma, Burkitt’s lymphoma, colon cancer and inflammatory diseases37,38. In addition, some viruses target BRD4 to regulate viral and cellular transcription, such as papillomavirus, HIV, and SARS-CoV-236,39.
To map the BRD4 interaction using PDPL, we combined miniSOG with a short N- or C-terminal isoform of BRD4. Proteomic results revealed a high degree of overlap between the two constructs (Supplementary Fig. 9a). The nuclear proteome identified with miniSOG-H2B covers 77.6% of the proteins interacting with BRD4 (Supplementary Fig. 9b). Then, different times of illumination (2, 5, 10, 20 min) were used to adjust the marker radius (Fig. 4a and supplementary data 3). We conclude that at shorter photoperiods, PDPL will primarily label direct binding partners, while longer periods will include proteins identified during shorter photoactivation periods as well as indirect targets in labeling complexes. In fact, we found strong overlap between adjacent time points (84.6% for 2 and 5 min; 87.7% for 5 and 10 min; 98.7% for 10 and 20 min) (Fig. 4b and Supplementary Fig. 9c ). In all experimental groups, we found not only BRD4 self-labeling, but several known targets such as MED1, CHD8, BICRA, NIPBL, SMC1A, and HMGB1 annotated in the string database. The ionic strength of these targets is proportional to the exposure time (Fig. 4c and Supplementary Fig. 9d). GO analysis of the proteins identified in the 2-minute group showed that the identified proteins were localized in the nucleus and were involved in chromatin remodeling and RNA polymerase function. The molecular function of the protein was enriched in chromatin binding or transcriptional coactivation, consistent with BRD4 function (Fig. 4d). String database-enabled protein interaction analysis revealed a first level of indirect interactions between BRD4 and HDAC family interacting complexes such as SIN3A, NCOR2, BCOR, and SAP130 (Fig. 4e and Supplementary Fig. 9e), consistent with BRD4 and HDAC binding acetylated histones . . In addition, representative targets identified by LC-MS/MS, including Sin3A, NSUN2, Fus, and SFPQ, were confirmed by Western blotting (Fig. 4f). Recently, the short isoform of BRD4 has been reported to form nuclei with liquid-liquid phase separation (LLPS) properties. The RNA binding proteins Fus and SFPQ mediate the LLPS of various cellular processes and have been identified here as unrecorded BRD4 binding proteins. The interaction between BRD4 and SFPQ was confirmed by co-immunoprecipitation (co-IP) experiments (Figure 4g), suggesting another mechanism for BRD4-mediated liquid-liquid phase separation deserving further investigation. Taken together, these results suggest that PDPL is an ideal platform for identifying known BRD4 interacting as well as unknown binding proteins.
a Schematic representation of miniSOG-mediated BRD4 proximity marking, exposure times: 2, 5, 10, and 20 min. b Overlap of proteins identified at different illumination times. Protein enrichment identified in HEK293T-miniSOG-BRD4 was statistically significant compared to wild-type HEK293T. c Ion intensity when quantifying unlabeled representative known BRD4-binding proteins during the specified exposure time. n = 3 biologically independent samples. Data are presented as mean ± standard deviation. d Gene ontological analysis (GO) of proteins identified in the 2-minute group. The first ten GO terms are listed. Bubbles are colored according to the GO term category, and bubble size is proportional to the number of proteins found in each term. e String analysis of proteins interacting with BRD4. The yellow circles are direct glue and the gray circles are the first layer of indirect glue. The red lines represent the experimentally determined interactions and the blue lines represent the predicted interactions. f Representative BRD4 binding targets identified in LC-MS/MS were verified by Western blotting. g Co-immunoprecipitation experiments confirm the interaction between SFPQ and BRD4. These experiments were independently repeated at least twice with similar results. Raw data are provided in the form of raw data files.
In addition to identifying unregistered POI-associated targets, we hypothesize that PDPL will be suitable for identifying substrates for enzymes, which would require the characterization of indirect binding proteins in large complexes to annotate unregistered substrates. Parkin (encoded by PARK2) is an E3 ligase and mutations in parkin are known to cause autosomal recessive juvenile Parkinson’s disease (AR-JP)42. In addition, parkin has been described as essential for mitophagy (mitochondrial autophagy) and removal of reactive oxygen species. However, although several parkin substrates have been identified, the role of parkin in this disease remains unclear. To annotate its uncharacterized substrates, PDPL was tested by adding miniSOG to the N- or C-terminus of parkin. Cells were treated with the carbonyl cyanide proton transporter m-chlorophenylhydrazone (CCCP) to activate parkin via the PINK1-Parkin pathway. Compared to our BRD4 PDPL results, parkin N-terminus fusion revealed a larger set of target proteins, although it covered a larger portion of the C-terminus (177 out of 210) (Figure 5a,b and Supplementary Data 4). the result is consistent with reports that N-terminal tags can aberrantly activate Parkin44. Surprisingly, there were only 18 overlapping proteins in our data with published AP-MS results for Parkin43, likely due to differences between cell lines and proteomics workflows. In addition to four known proteins (ARDM1, HSPA8, PSMD14, and PSMC3) identified by two methods (Fig. 5c)43. To further validate the results of LC-MS/MS, PDPL treatment and subsequent Western blotting were used to compare the results of the HEK293T parent cell assay and the stable N-terminal parkin line. Previously unknown targets CDK2, DUT, CTBP1, and PSMC4 were tested with a known binder, DNAJB1 (Fig. 5d).
Volcano plot of parkin-interacting proteins in HEK293T cells with stably expressed miniSOG fused to the N- or C-terminus of parkin (n = 3 independent biological experiments). Two-tailed Student’s t-test was used on volcano plots. HEK293T was used as a negative control. Significantly changed proteins are highlighted in red (p < 0.05 and >2-fold ion intensity difference). Significantly changed proteins are highlighted in red (p < 0.05 and >2-fold ion intensity difference). Значительно измененные белки выделены красным цветом (p < 0,05 и >2-кратная разница в интенсивности ионов). Significantly altered proteins are highlighted in red (p < 0.05 and >2-fold difference in ion intensity).显着变化的蛋白质以红色突出显示（p < 0.05 和> 2 倍离子强度差异）。显着变化的蛋白质以红色突出显示（p < 0.05和> 2 Значительно измененные белки выделены красным цветом (p < 0,05 и > 2-кратная разница в ионной силе). Significantly altered proteins are highlighted in red (p < 0.05 and > 2-fold difference in ionic strength). Related proteins important for HEK293T-miniSOG but not important for HEK293T are shown in green. b Venn diagram showing overlapping proteins between N-terminal and C-terminal constructs. N-terminal tags can aberrantly activate parkin and result in more recognizable proteins. c Venn diagram showing overlapping proteins between PDPL and AP-MS. Known interactors are listed, including 4 of 18 overlapping proteins and 11 of 159 proteins specifically identified in PDPL. d Representative targets identified by LC-MS/MS were verified by Western blotting. e Ssu72 and SNW1 were identified as unregistered parkin substrates. These FLAG-tagged protein plasmids were transfected into HEK293T and HEK293T-Parkin-miniSOG followed by CCCP treatment at various time points. The degradation was more pronounced in the Parkin overexpression line. f Using the proteasome inhibitor MG132, it was confirmed that the degradation process of Ssu72 and SNW1 is mediated by proteasome-ubiquitination. These experiments were independently repeated at least twice with similar results. Raw data are provided in the form of raw data files.
Notably, the proteins identified by PDPL must include parkin-binding proteins and their substrates. To detect unregistered parkin substrates, we selected seven identified proteins (PUF60, PSPC1, UCHL3, PPP1R8, CACYBP, Ssu72 and SNW1) and transfected plasmids to expose these genes to normal HEK293T and stably express miniSOG-Parkin’s HEK293T followed by CCCP treatment. The levels of Ssu72 and SNW1 proteins were significantly reduced in the stable miniSOG-Parkin line (Fig. 5e). Treatment with CCCP for 12 hours resulted in the most significant degradation of both substrates. To investigate whether the degradation of Ssu72 and SNW1 is regulated by proteasome-ubiquitination, the proteasome inhibitor MG132 was added to inhibit proteasome activity, and in fact we found that their degradation process was inhibited (Fig. 5f). Additional non-substrate targets were confirmed as Parkin interactors using Western blotting (Supplementary Fig. 10), which showed consistent results with LC-MS/MS. In conclusion, the integration of the PDPL workflow with target protein transfection verification allows the identification of unregistered E3 ligase substrates.
We have developed a common proximity marking platform that allows you to identify in space and time interacting POIs. The platform is based on the miniSOG photosensitizer protein, which is only about 12 kDa, less than half the size of the mature APEX2 enzyme (27 kDa) and one third the size of TurboID (35 kDa). The smaller size should greatly expand the range of applications for studying small protein interactomes. Further exploration of additional photosensitizers, whether genetically encoded proteins or small molecules, is needed to increase the quantum yield of singlet oxygen and expand the sensitivity of this approach. For the current version of miniSOG, high temporal resolution can be achieved using blue illumination to activate proximity markers. In addition, longer exposure time released a larger “cloud” of singlet oxygen, resulting in modification of more distal histidine residues, increased labeling radius, and the ability to fine-tune the PDPL spatial resolution. We also tested seven chemical probes to increase the signal-to-background ratio and explored the molecular mechanism behind this approach. The TOP-ABPP workflow combined with unbiased open search confirmed that modifications occurred only in histidines and no consistent microenvironment was observed for increased histidine modifications, except for a moderate preference for histidines in the loop region.
PDPL has also been used to characterize subcellular proteomes with proteome specificity and coverage at least comparable to other proximity labeling and organelle-specific chemical probe methods. Proximity markers have also been successfully used to characterize the surface, lysosomal, and secretome-associated proteomes46,47. We believe that PDPL will be compatible with these subcellular organelles. In addition, we challenged PDPL by identifying targets for cytosolic protein binding that are more complex than membrane bound proteins due to their dynamic properties and involvement in more temporal interactions. PDPL was applied to two proteins, the transcriptional coactivator BRD4 and the disease-associated ligase E3 Parkin. These two proteins were chosen not only for their fundamental biological functions, but also for their clinical relevance and therapeutic potential. For these two POIs, well-known binding partners as well as unregistered targets were identified. Notably, the phase separation-associated protein SFPQ was confirmed by co-IP, which may indicate a new mechanism by which BRD4 (short isoform) regulates LLPS. At the same time, we believe that the identification of Parkin substrates is a scenario in which the identification of indirect adhesives is required. We identified two unidentified parkin substrates and confirmed their degradation along the ubiquitination-proteasome pathway. Recently, a mechanism-based trapping strategy has been developed to detect hydrolase substrates by trapping them with enzymes. Although this is a very powerful method, it is not suitable for the analysis of substrates involved in the formation of large complexes and requires the formation of covalent bonds between the enzyme and the substrate. We expect that PDPL can be extended to study other protein complexes and enzyme families, such as the deubiquitinase and metalloprotease families.
A new form of miniSOG, called SOPP3, has been developed with improved singlet oxygen production. We compared miniSOG with SOPP3 and found improved marking performance, although the signal-to-noise ratio remained unchanged (Supplementary Fig. 11). We hypothesized that optimization of SOPP3 (eg, through directed evolution) would lead to more efficient photosensitizer proteins that require shorter light times and thus allow more dynamic cellular processes to be captured. Notably, the current version of PDPL is limited to the cellular environment as it requires blue light illumination and cannot penetrate deep tissues. This feature precludes its use in animal model studies. However, the combination of optogenetics with PDPL could provide an opportunity for animal research, especially in the brain. In addition, other engineered infrared photosensitizers also remove this limitation. Research is currently underway in this area.
The HEK293T cell line was obtained from ATCC (CRL-3216). The cell line tested negative for mycoplasma infection and was cultured in DMEM (Thermo, #C11995500BT) supplemented with 10% fetal bovine serum (FBS, Vistech, #SE100-B) and 1% penicillin/streptomycin (Hyclone, #SV30010). grown in.
3-Aminophenylene (sample 3) and (4-ethynylphenyl)methanamine (sample 4) were purchased from Bidepharm. Propylamine (probe 2) was purchased from Energy-chemicals. N-(2-Aminophenyl)pent-4-ynamide (probe 1) was synthesized according to published methods.
Supplementary Table 1 lists the genetic constructs used in this study. The miniSOG and KillerRed sequences were cloned from a gift plasmid from P. Zou (Peking University). The mitochondrial matrix targeting sequence was derived from the 23 N-terminal amino acids of COX4 and cloned into the indicated vectors using a Gibson assembly (Beyotime, #D7010S). To target the membrane and nucleus of the endoplasmic reticulum, SEC61B human DNA (NM_006808.3) (NEB, #M0491L) amplified by PCR from a cDNA library of HEK293T cells, and H2B DNA (donated by D. Lin, Shenzhen Bay Laboratory) and cloned , as mentioned above. Unless otherwise indicated, other protein genes used for transfection and construction of stable cell lines were PCR amplified from the HEK293T cell cDNA library. G3S (GGGS) and G4S (GGGGS) were used as linkers between the bait protein and miniSOG. A V5 epitope tag (GKPIPNPLLGLDST) was added to these fusion constructs. For expression in mammals and to establish a stable cell line, the miniSOG fusion construct was subcloned into the pLX304 lentiviral vector. For bacterial expression, miniSOG was cloned into the pET21a vector labeled 6xHis at the C-terminus.
HEK293T cells were seeded at 2.0 x 105 cells per well in six-well plates and transfected 24 hours later with recombinant lentiviral plasmids (2.4 μg pLX304) and viral packaging plasmids (1.5 μg psPAX2 and 1.2 μg pMD2.G) with using Lipo8000 (Beyotime, #C0533), about 80% fusion. After overnight transfection, the medium was changed and incubated for another 24 hours. The collection of the virus was carried out after 24, 48 and 72 hours. Prior to infection of the target cell lines, the viral medium was filtered through a 0.8 μm filter (Merck, #millex-GP) and polybrene (Solarbio, #H8761) was added to a concentration of 8 μg/ml. After 24 hours, the cells were allowed to recover by changing the medium. Cells were selected using 5 μg/ml blasticidin (Solarbio, #3513-03-9) for the first three passages as a lower stringent selection. Then used 20 μg/ml as a more stringent regimen for the next three passages.
Cells were seeded in 12-well chambers (Ibidi, #81201) at a density of approximately 20,000 cells per well. To improve adhesion of HEK293T cells, add 50 µg/ml fibronectin (Corning, #356008) diluted in phosphate buffered saline (PBS, Sangon, #B640435) at 37°C. The chambers were pretreated for 1 hour and then removed with PBS. After 24 h, cells were washed once with PBS, incubated with 1 mM probe 3 in fresh Hanks’ balanced salt solution (HBSS, Gibco, #14025092) for 1 h at 37°C, and then incubated with a blue LED (460 nm). ) were irradiated for 10 min at room temperature. After that, the cells were washed twice with PBS and fixed with 4% formaldehyde in PBS (Sangon, #E672002) for 15 minutes at room temperature. Excess formaldehyde was removed from fixed cells by washing three times with PBS. Cells were then permeabilized with 0.5% Triton X-100 (Sangon, #A600198) in PBS and washed 3 times with PBS. Then remove the chamber and add to each sample 25 µl of a click reaction mixture containing 50 µM Cy3-azide (Aladdin, #C196720), 2 mM CuSO4 (Sangon, #A603008), 1 mM BTTAA (Confluore, #BDJ-4 ) and 0.5 mg/ml sodium ascorbate (Aladdin, no. S105024) and incubated for 30 minutes at room temperature. After a snap reaction, cells were washed six times with PBS containing 0.05% Tween-20 (Sangon, #A600560) (PBST) and then blocked with 5% BSA (Abcone, #B24726) in PBST for 30 minutes at room temperature.
For colocalization immunostaining, cells were incubated with primary antibodies according to the indicated conditions: mouse anti-V5 tag mAb (1:500, CST, #80076), rabbit anti-Hsp60 mAb (1:1000), ABclonal, #A0564), rabbit polyclonal anti-calnexin antibody (1:500, Abcam, #ab22595) or rabbit anti-lamin A/C monoclonal antibody (1:500; CST, #2032) at 4 °C overnight. After washing 3 times with PBST, cells were incubated with secondary antibodies: goat anti-rabbit Alexa Fluor 488 (Thermo, #A11034) diluted 1:1000, goat anti-mouse Alexa Fluor 594 (CST, #8889) diluted 1:1000. dilution Dilute at room temperature for 30 minutes. Cells were then washed 3 times with PBST and counterstained with DAPI (Thermo, #D1306) in PBS for 10 minutes at room temperature. After 3 washes with PBS, cells were sealed in 50% glycerol (Sangon, #A600232) in PBS for imaging. Immunofluorescent images were obtained using a ZEISS LSM 900 Airyscan2 confocal microscope and ZNE 3.5 software.
For singlet oxygen fluorescent imaging, cells were washed twice with Hanks HEPES buffer before adding 100 nM Si-DMA in Hanks HEPES buffer (DOJINDO, #MT05). After exposure to light, the cells were incubated in a CO2 incubator at 37°C for 45 minutes. Cells were then washed twice with Hanks’ HEPES buffer and counterstained with Hoechst in Hanks’ HEPES buffer for 10 minutes at room temperature and visualized using a ZEISS LSM 900 confocal microscope. , #M36008) in HBSS buffer containing calcium and magnesium. After exposure to light or doxorubicin (MCE, #HY-15142A), cells were incubated in a CO2 incubator at 37° C. for 10 minutes, washed twice with HBSS buffer, and incubated with Hoechst in HBSS buffer at room temperature. minutes. Doxorubicin was used as a positive probe control where cells were treated with 20 μM doxorubicin in HBSS containing 1% BSA for 30 min. Immunofluorescent images were obtained using a Zeiss LSM 900 confocal microscope.
HEK293T cells stably expressing mito-miniSOG were seeded at a density of approximately 30% in 15 cm dishes. After 48 hours, when ~80% confluence was reached, the cells were washed once with PBS, incubated with 1 mM Probe 3 in fresh HBSS buffer in for 1 hour at 37°C and then illuminated with a blue LED for 10 minutes at room temperature. . Thereafter, the cells were washed twice with PBS, scraped and resuspended in ice-cold PBS buffer containing EDTA-free protease inhibitors (MCE, #HY-K0011). Cells were lysed by sonicating the tip for 1 minute (1 second on and 1 second off at 35% amplitude). The resulting mixture was centrifuged at 15,871 xg for 10 min at 4°C to remove debris, and the supernatant concentration was adjusted to 4 mg/mL using a BCA protein assay kit (Beyotime, #P0009). Combine 1 ml of the above lysate with 0.1 mM photodegradable biotin azide (Confluore, #BBBD-14), 1 mM TCEP (Sangon, #A600974), 0.1 mM TBTA ligand (Aladdin, #T162437), and 1 mM CuSO4 incubator with bottom rotation up for 1 hour at room temperature. After a snap reaction, add the mixture to the pre-mixed solution (MeOH:CHCl3:H2O = 4 ml:1 ml:3 ml) in a 10 ml glass vial. The samples were mixed and centrifuged at 4500 g for 10 minutes at room temperature. The lower and upper solutions were discarded, the precipitate was washed twice with 1 ml of methanol and centrifuged at 15871×g for 5 min at 4°C. Add 1 ml of 8 M urea (Aladdin, no. U111902) in 25 mM ammonium bicarbonate (ABC, Aladdin, no. A110539) to dissolve the precipitate. Samples were reconstituted with 10 mM dithiothreitol (Sangon, #A100281 in 25 mM ABC) for 40 minutes at 55°C followed by the addition of 15 mM fresh iodoacetamide (Sangon, #A600539) at room temperature in the dark. Alkylation within 30 minutes. . An additional 5 mM dithiothreitol was added to stop the reaction. Prepare approximately 100 µl NeutrAvidin agarose beads (Thermo, #29202) for each sample by washing 3 times with 1 ml PBS. The above proteome solution was diluted with 5 ml PBS and incubated with pre-washed NeutrAvidin agarose beads for 4 hours at room temperature. The beads were then washed 3 times with 5 ml PBS containing 0.2% SDS (Sangon, #A600485), 3 times with 5 ml PBS containing 1M urea, and 3 times with 5 ml ddH2O. The beads were then harvested by centrifugation and resuspended in 200 μl of 25 mM ABC containing 1 M urea, 1 mM CaCl 2 (Macklin, #C805228) and 20 ng/μl trypsin (Promega, #V5280). Trypsinize overnight at 37°C with rotation. The reaction was stopped by adding formic acid (Thermo, # A117-50) until the pH reached 2-3. The beads were washed 3 times with 1 ml of PBS containing 0.2% SDS, 3 times with 1 ml of PBS containing 1 M urea, and then 3 times with 1 ml of distilled water. The modified peptides were released by light lysis (365 nm) for 90 min using 200 μl of 70% MeOH. After centrifugation, the supernatant was collected. The beads were then washed once with 100 μl of 70% MeOH and the supernatants were pooled. Samples were dried in a Speedvac vacuum concentrator and stored at -20°C until analysis.
To identify and quantify singlet oxygen modified peptides, samples were redissolved in 0.1% formic acid and 1 μg of peptides were analyzed using an Orbitrap Fusion Lumos Tribrid mass spectrometer equipped with a nano ESI source from Tune and Xcalibur from vendor software 4.3. Samples were separated on a 75 µm × 15 cm internally packed capillary column with 3 µm C18 material (ReproSil-pur, #r13.b9.) and connected to an EASY-nLC 1200 UHPLC system (Thermo). The peptides were separated by linear 95 minute gradient chromatography from 8% solvent B to 50% solvent B (A = 0.1% formic acid in water, B = 0.1% formic acid in 80% acetonitrile), then linearly increased to 98% B min in 6 min at a flow rate of 300 nl/min. Orbitrap Fusion Lumos collects data alternately between full MS scan and MS2 scan depending on the data. The sputtering voltage was set to 2.1 kV and the temperature of the ion transport capillary was 320°C. MS spectra (350-2000 m/z) were collected with a resolution of 120,000, AGC 4 × 105, and a maximum input time of 150 ms. The 10 most common multiply charged precursors in each full scan were fragmented using HCD with a normalized collision energy of 30%, a quadrupole isolation window of 1.6 m/z, and a resolution setting of 30,000. An AGC target for tandem mass spectrometry using 5×104 and maximum input time 150 ms. The dynamic exception is set to 30 seconds. Unassigned ions or those with a charge of 1+ and >7+ were rejected for MS/MS. Unassigned ions or those with a charge of 1+ and >7+ were rejected for MS/MS. Неназначенные ионы или ионы с зарядом 1+ и >7+ были отклонены для МС/МС. Unassigned ions or ions with a charge of 1+ and >7+ were rejected for MS/MS.未指定的离子或电荷为1+ 和>7+ 的离子被拒绝用于MS/MS。未指定的离子或电荷为1+ 和>7+ 的离子被拒绝用于MS/MS。 Неуказанные ионы или ионы с зарядами 1+ и >7+ были отклонены для МС/МС. Unspecified ions or ions with charges of 1+ and >7+ were rejected for MS/MS.
The raw data is processed using the FragPipe computing platform based on MSFragger. Mass biases and corresponding amino acids were determined using an open search algorithm with a precursor mass tolerance of -150 to 500 Da. Modified peptides were then identified using histidine modifications with mass gains of +229.0964 and +247.1069 Da in PD (Proteome Discoverer 2.5, Thermo).
Cells stably expressing the fused miniSOG gene were plated in 6 cm dishes. Upon reaching ~80% confluence, cells were washed once with HBSS (Gibco, #14025092), then incubated with chemical probes in HBSS for 1 hour at 37°C and illuminated with blue light. 10W LED for 20 minutes at room temperature. To determine which type of reactive oxygen species is involved in PDPL, 0.5 mM vitamin C (MCE, #HY-B0166), 5 mM Trolox (MCE, #HY-101445), D2O (Sigma, #7789-20-0) , 100 mM mannitol (Energy Chemical, #69-65-8), 100 μM H2O2, 10 mM NaN3 were added to the cells as supplements. After washing with cold PBS, cells were scraped, collected in 1.5 ml centrifuge tubes, and sonicated with a tip for 1 min in 200 μl of PBS with 1x protease inhibitor without EDTA (1 s and 1 s without, amplitude 35%). The resulting mixture was centrifuged at 15,871 × g for 10 min at 4 °C and the supernatant concentration was adjusted to 1 mg/mL using a BCA protein assay kit. Approximately 50 µl of the above lysate was incubated with 0.1 mM rhodamine azide (Aladdin, no. T131368), 1 mM TCEP, 0.1 mM TBTA ligand, and 1 mM CuSO4 for 1 hour at room temperature with rotation from bottom to top. After the click reaction, precipitation with acetone was carried out by adding 250 μl of pre-chilled acetone to the samples, incubating at -20°C for 20 min and centrifuging at 6010×g for 10 min at 4°C. Collect the pellet and boil in 50 µl of 1x Laemmli’s buffer for 10 min at 95 °C. Samples were then analyzed on SDS-PAGE long gels and visualized using the Bio-rad ChemiDoc MP Touch imaging system with Image Lab Touch software.
Expression and purification of the recombinant miniSOG-6xHis protein was performed as previously described. Briefly, E. coli BL21(DE3) cells (TransGen, #CD701-02) were transformed with pET21a-miniSOG-6xHis and protein expression was induced with 0.5 mM IPTG (Sangon, #A600168). After cell lysis, proteins were purified using Ni-NTA agarose beads (MCE, no. 70666), dialyzed against PBS, and stored at –80°C.
For an antibody-based in vitro label proximity assay, mix 100 μM purified miniSOG, 1 mM probe 3, and 1 μg anti-label mouse monoclonal antibody (TransGen, #HT501-01) in PBS to a total reaction volume of 50 μl. . The reaction mixture was irradiated with blue LED light for 0, 2, 5, 10, and 20 min at room temperature. The mixture was incubated with 0.1 mM biotin-PEG3-azide (Aladdin, #B122225), 1 mM TCEP, 0.1 mM TBTA ligand, and 1 mM CuSO4 for 1 h at room temperature on an upward motion shaker. After a snap reaction, add 4x Laemmli’s buffer directly to the mixture and boil at 95°C for 10 min. Samples were analyzed on SDS-PAGE gels and analyzed by western blotting with streptavidin-HRP (1:1000, Solarbio, #SE068).
A histidine-containing synthetic peptide with C-terminal amidation (LHDALDAK-CONH2) was used to analyze nearby peptide-based in vitro labeling. In this assay, 100 μM purified miniSOG, 10 mM probe 3 and 2 μg/ml synthetic peptide were mixed in PBS in a total reaction volume of 50 μl. The reaction mixture was irradiated with blue LED light for 1 hour at room temperature. One microliter of sample was analyzed using an LC-MS system (Waters, SYNAPT XS Ions Mobility Time-of-Flight mass spectrometer with MassLynx spectrum analysis software).
HEK293T cells stably expressing the miniSOG fusion gene were seeded in 10 cm dishes for lines with different organelle localization (Mito, ER, Nucleus) and 15 cm dishes for Parkin-miniSOG and BRD4-miniSOG lines. Upon reaching ~90% confluence, the cells were washed once with HBSS, then incubated with probe 3 in HBSS for 1 hour at 37°C and illuminated with a 10 W blue LED at room temperature. For non-contact labeling of Parkin, 10 µM proton carbonyl cyanide carrier m-chlorophenylhydrazone CCCP (Solarbio, #C6700) with probe 3 in HBSS was added for 1 hour at 37°C. The cell lysis, click chemistry, reduction and alkylation steps were the same as described above, except that 2 mg of lysate was added and biotin PEG3 azide was used in the click reaction instead of photodegradable biotin azide. After enrichment, the beads were washed 3 times with 5 ml of PBS containing 0.2% SDS, 3 times with 5 ml of PBS containing 1 M urea, and 3 times with 5 ml of PBS. Thereafter, 2 µg trypsin was added to 300 µl 25 mM ABC containing 1 M urea to cleave the protein overnight at 37°C. The reaction was stopped by adding formic acid until a pH of 2-3 was reached. After trypsinization on beads, the peptide solution was desalted using a SOLAµ HRP column (Thermo, #60209-001) and dried in a Speedvac vacuum concentrator. Peptides were redissolved in 0.1% formic acid and 500 ng of peptides were analyzed using an Orbitrap Fusion Lumos Tribrid mass spectrometer equipped with the nano-ESI source described above. Peptides were separated on commercial RP-HPLC precolumns (75 μm x 2 cm) (Thermo, no. 164946) and analytical RP-HPLC columns (75 μm x 25 cm) (Thermo, no. 164941), both filled with 2 μm. gradient from 8% to 35% ACN in 60 minutes, then linearly increased to 98% B in 6 minutes at a flow rate of 300 Nl/min. MS spectra (350-1500 m/z) were collected with a resolution of 60,000, AGC 4 × 105, and a maximum input time of 50 ms. Selected ions were sequentially fragmented by HCD in 3 s cycles with a normalized collision energy of 30%, a quadrupole isolation window of 1.6 m/z, and a resolution of 15000. A 5 × 104 tandem mass spectrometer AGC target and a maximum injection time of 22 ms were used. The dynamic exclusion is set to 45 seconds. Unassigned ions or those with a charge of 1+ and >7+ were rejected for MS/MS. Unassigned ions or those with a charge of 1+ and >7+ were rejected for MS/MS. Неназначенные ионы или ионы с зарядом 1+ и >7+ были отклонены для МС/МС. Unassigned ions or ions with a charge of 1+ and >7+ were rejected for MS/MS.未指定的离子或电荷为1+ 和>7+ 的离子被拒绝用于MS/MS。未指定的离子或电荷为1+ 和>7+ 的离子被拒绝用于MS/MS。 Неуказанные ионы или ионы с зарядами 1+ и >7+ были отклонены для МС/МС. Unspecified ions or ions with charges of 1+ and >7+ were rejected for MS/MS.
The sample preparation steps up to the enrichment of the NeutrAvidin beads were the same as in the LC-MS/MS analysis described above. Approximately 50 μg of lysate was used as input for loading control and 2 mg of lysate was used for click reactions. After enrichment and washing with neutravidin, the bound proteins were eluted by adding 50 μl of Laemmli’s buffer to the agarose resin beads and boiling at 95° C. for 5 minutes. Control load input and bead enriched samples were analyzed by SDS-PAGE and transferred to PVDF membranes (Millipore, #ISEQ00010) by standard Western blot methods. The membranes were blocked with 5% skim milk (Sangon, #A600669) in TBS containing 0.1% tween-20 (TBST) and incubated sequentially with primary and secondary antibodies. Primary antibodies were diluted 1:1000 in 5% skim milk in TBST and incubated overnight at 4°C. Secondary antibodies were used in a ratio of 1:5000 and incubated for 1 hour at room temperature. The membranes were visualized by chemiluminescence using the Chemidoc MP imaging system. All uncut scans of blots and gels in the figure are presented as raw data.
Primary antibodies used in this study included rabbit anti-SFPQ monoclonal antibody (CST, no. 71992), rabbit anti-FUS monoclonal antibody (CST, no. 67840), rabbit anti-NSUN2 polyclonal antibody (Proteintech, no. 20854-1-AP), rabbit anti-mSin3A polyclonal antibody (Abcam, #ab3479), mouse anti-tag monoclonal antibody (TransGen, #HT201-02), mouse anti-β-actin monoclonal antibody (TransGen, #HC201-01), rabbit anti-CDK2 monoclonal antibody (ABclonal, #A0094), rabbit monoclonal antibody to CTBP1 (ABclonal, #A11600), rabbit polyclonal antibody to DUT (ABclonal, #A2901), rabbit polyclonal antibody to PSMC4 (ABclonal, #A2505), rabbit anti-DNAJB1 polyclonal antibody (ABclonal, # A5504). These antibodies were used at a 1:1000 dilution in 5% skim milk in TBST. The secondary antibodies used in this study included anti-rabbit IgG (TransGen, #HS101-01), anti-mouse IgG (TransGen, #HS201-01) at a 1:5000 dilution.
To further investigate whether BRD4 interacts with SFPQ, stable HEK293T and BRD4-miniSOG cells overexpressing HEK293T were plated in 10 cm dishes. Cells were washed with cold PBS and lysed in 1 ml Pierce IP lysis buffer (Thermo Fisher, #87787) with EDTA-free protease inhibitor for 30 minutes at 4°C. After that, the lysates were collected in 1.5 ml centrifuge tubes and centrifuged at 15,871 xg for 10 min at 4°C. The supernatant was harvested and incubated with 5 µg of anti-V5 labeled mouse monoclonal antibody (CST, #80076) overnight at 4°C. Wash approximately 50 µl of protein A/G magnetic beads (MCE, #HY-K0202) twice with PBS containing 0.5% Tween-20. Then the cell lysates were incubated with magnetic beads for 4 hours at 4°C with rotation from bottom to top. Then the beads were washed four times with 1 ml of PBST buffer and boiled at 95°C for 5 min. Samples were analyzed on SDS-PAGE gels and transferred to PVDF membranes using standard Western blot methods. The membranes were blocked in 5% skim milk in TBST and incubated sequentially with primary and secondary antibodies. Primary Antibody Rabbit anti-SFPQ monoclonal antibody (CST, #71992) was used at a ratio of 1:1000 in 5% skim milk in TBST and incubated overnight at 4°C. Anti-rabbit IgG was used at a ratio of 1:5000 and incubated for 1 hour at room temperature. The membranes were visualized by chemiluminescence using the Chemidoc MP imaging system.
All structures used for the Solvent Accessible Surface Area (SASA) analysis were obtained from the Protein Data Bank (PDB)52 or the AlphaFold Protein Structure Database53. Absolute SASA was calculated for each residue using the FreeSASA program. Only complete and unambiguous SASA data for labeled histidine and its neighbors were used to obtain the mean SASA for each structure. The relative solvent accessibility (RSA) for each histidine was calculated by dividing the absolute SASA value by the empirical maximum possible residue surface area available to the solvent. All histidines were then classified as hidden if the mean RSA was below 20%, otherwise exposed56.
Raw files obtained in DDA mode were searched using Proteome Discoverer (v2.5) or MSfragger (Fragpipe v15.0) in the appropriate SwissProt verified protein database containing common contaminants. The peptides required complete trypsin with two missing cleavage sites, carbamoyl methylation as a fixed modification and methionine oxidation as a dynamic modification. Precursor and fragment weight tolerances were set to 10 ppm and 0.02 Da (MS2 Orbitrap), respectively. Contaminant hits were removed, and proteins were filtered to obtain a false discovery rate of <1%. Contaminant hits were removed, and proteins were filtered to obtain a false discovery rate of <1%. Попадания загрязняющих веществ были удалены, а белки отфильтрованы, чтобы получить коэффициент ложного обнаружения <1%. Contaminant hits were removed and proteins filtered to give a false detection rate of <1%.去除污染物命中，过滤蛋白质以获得<1%的错误发现率。 <1%的错误发现率。 Попадания загрязняющих веществ были удалены, а белки отфильтрованы для достижения уровня ложных обнаружений <1%. Contaminant hits were removed and proteins filtered to achieve a false positive rate of <1%. For quantitative analysis without the use of labels, normalized protein content from three biological repeats was used. Protein subcellular localization analysis was performed using Gene Ontology (GO) analysis from DAVID Bioinformatics Resources, MitoCarta 3.0 and databases compiled and published by the Alice Ting group. The volcano map was obtained from Perseus (v1.6.15.0). Protein abundance fold changes were tested for statistical significance using a two-sided t-test, and protein hits were identified with abundance change >2 (unless otherwise stated) and p value <0.05. Protein abundance fold changes were tested for statistical significance using a two-sided t-test, and protein hits were identified with abundance change >2 (unless otherwise stated) and p value <0.05. Изменения кратности содержания белка были проверены на статистическую значимость с использованием двустороннего t-критерия, и совпадения белков были идентифицированы с изменением содержания> 2 (если не указано иное) и значением p <0,05. Protein content fold changes were tested for statistical significance using a two-tailed t-test, and protein matches were identified with content change >2 (unless otherwise noted) and a p value <0.05.使用双边t 检验测试蛋白质丰度倍数变化的统计显着性，并确定蛋白质命中的丰度变化> 2（除非另有说明）和p 值<0.05。使用 双边 t 检验 测试 蛋白质 倍数 变化 的 统计 显着性 并 确定 蛋白质 命 中 的 丰度 变化> 2 （另 有 说明） 和 p 值 <0.05。 Статистическую значимость кратных изменений содержания белка проверяли с использованием двустороннего t-критерия, а совпадения белков определяли для изменений содержания >2 (если не указано иное) и p-значений <0,05. Statistical significance of fold changes in protein content was tested using a two-tailed t-test, and protein matches were determined for content changes >2 (unless otherwise indicated) and p-values ​​<0.05. Protein interaction analysis was performed using GO analysis along with the String database.
Three biological replicates were carried out with similar results. Statistical analysis was done with GraphPad Prism (GraphPad software) and volcano plots were generated with Perseus (v1.6.15.0). To compare the two groups, p-values ​​were determined using a two-tailed Student’s t-test. Only singleton proteins identified at least twice in the experimental group were included in the volcano plots, and the corresponding missing values ​​in the control group were replaced with Perseus from a normal distribution so that the p-value could be calculated. Error bars represent the mean ± standard deviation. In proteomic analyzes for statistical analysis, the abundance of proteins that appeared in at least two biological replicates was retained. Statistical methods were not used to pre-determine the sample size. The experiments are not random. The researchers were not blind to the tasks during the experiment and evaluation of the results.
For more information on study design, see the Nature Research Report abstract linked to this article.
The mass spectrometry data obtained in this study was submitted to the ProteomeXchange Consortium via the iProX57 partner repository under dataset ID PXD034811 (PDPL-MS dataset). Raw data are provided in the form of raw data files. This article provides the original data.
Gingras, AC, Abe, KT & Raught, B. Getting to know the neighborhood: using proximity-dependent biotinylation to characterize protein complexes and map organelles. Gingras, AC, Abe, KT & Raught, B. Getting to know the neighborhood: using proximity-dependent biotinylation to characterize protein complexes and map organelles. Gingras, A.S., Abe, K.T. and Raut, B. Familiarity with surroundings: using proximity-dependent biotinylation to characterize protein complexes and map organelles. Gingras, AC, Abe, KT & Raught, B. 了解邻里：使用邻近依赖性生物素化来表征蛋白质复合物并绘制细胞器图。 Gingras, AC, Abe, KT & Raught, B. Understanding the neighborhood: use the neighborhood’s dependence on biological life. Gingras, A.S., Abe, K.T. and Raut, B. Understanding proximity: characterization of protein complexes and mapping of organelles using proximity-dependent biotinylation. Current. My opinion. Chemical. biology 48, 44–54 (2019).
Geri, J. B. et al. Mapping the microenvironment by transferring Dexter energy to immune cells. Science 367, 1091–1097 (2020).
Hertling, E.L. et al. Two-proteome scale networks detect cell-specific remodeling of the human interactome. Cells 184, 3022–3040.e3028 (2021).