Library-based, multiplexed strategy for mapping protein interaction networks via crosslinking Authors: Yuxin Lu, Shuting Meng, Yi Zhang, Fei Wang, Tingting Shi, Yijia Chi, Song Chen, Dongxin Zhao Published: 2026 ABSTRACT AbstractBackgroundProtein-protein interactions are fundamental to cellular function, yet resolving their interaction interfaces and dynamic behaviors in native biological contexts remains challenging, particularly for weak or transient interactions. Crosslinking strategies based on noncanonical amino acids offer an effective means to capture such interactions; however, traditional single-site incorporation provides limited coverage and may overlook critical interaction hotspots.ResultsBy employing a mutagenesis library, multiple interaction partners and cross-linking sites of a target protein can be simultaneously screened in a single experiment, without prior knowledge of its precise structural or functional features, enabling effective and unbiased analysis of its interaction network. In this study, we constructed an amber codon-scanning mutagenesis library of PSMD10, facilitating independent incorporation of the photocrosslinking ncAA p-azido-phenylalanine at multiple distinct residues. This approach allowed us to systematically interrogate and precisely map potential interaction regions across the protein surface. Coupled with crosslinking mass spectrometry, we identified multiple residues involved in intermolecular interactions, as well as previously unreported interaction partners, including T2FA, TBA1C, and ATRIP.ConclusionsThese findings expand our understanding of PSMD10-associated proteasome interactome, demonstrate a multiplexed strategy for in situ mapping of protein interaction interfaces with broad coverage, and offer a valuable platform for developing therapeutics that target protein-protein interactions.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12896-026-01181-7.Keywords: Scanning mutagenesis library, Protein-protein interactions, Amber codon scanning, Photo-crosslinking, Crosslinking mass spectrometry Introduction Protein-protein interactions (PPIs) play pivotal roles in almost all biological processes [1], but remain challenging to investigate under living conditions. Although the structure and function of individual proteins have been characterized in detail, their interaction network and dynamics, as well as interaction interface within a native cellular context remain largely elusive [2, 3]. Traditional methods to study protein-protein interactions, such as immunoprecipitation, pull-down-assays, yeast-two-hybrid approaches, and Förster resonance energy transfer (FRET)-based assays have proven powerful to elucidate protein interactomes, but these methods show a number of limitations [4]. For instance, high false positive identification rates, constriction to non-native environments, and loss of weak and/or transient interaction partners are common issues. Methods based on covalent bonding have been gaining strong momentum in recent years, as stable covalent linkages irreversibly lock interacting proteins and tolerate stringent processing conditions. These properties enable the identification of weak and transient interactions with improved specificity, reliability, and accuracy [5, 6]. One such approach involves genetically encoding photo-crosslinkers into proteins, which upon UV light-activation react with nearby residues to crosslink interacting partners [7–9]. Another approach employs small molecules to chemically crosslink specific type of residues (e.g., lysine) on protein surfaces [10]. When coupled with mass spectrometric analysis of the cross-linked complex, these methods provide protein structure and interaction information based on residue-specific reactivity [11]. For both photo-crosslinking and chemical crosslinking approaches, a dedicated, orthogonal aminoacyl-tRNA synthetase (AARS) and its cognate suppressor tRNA pair is required to incorporate ncAAs via amber stop codon suppression. This engineered system operates independently from the host’s native translational machinery, and has been well established in both prokaryotic and eukaryotic systems, including mammalian cells [12]. Although methods based on photo-crosslinkers and chemical crosslinkers have made progress in probing protein-protein interactions within native cellular environments, these strategies typically incorporate ncAAs at only a single residue. Therefore, these methods can only capture interaction partners and crosslinking information in the vicinity of the specific site, which necessitates the prior acquisition of accurate structural or functional data during experimental design. However, in many cases, it is difficult to precisely predict which ncAA incorporation sites will efficiently capture interacting proteins, due to limited knowledge of functionally critical or conformationally sensitive regions of the target protein. As a result, many interactions may be missed due to suboptimal experimental design. In recent years, with the continuous development of mutation library construction methods, library-based high-throughput screening technologies have been widely applied in various fields. For example, CRISPR screening libraries have played a key role in functional genomics, helping to uncover the relationship between genes and phenotypes [13–15]. Enzyme library screening, on the other hand, has been crucial in enzyme engineering and catalyst optimization, enabling the identification of enzymes with specific functions or improved properties through efficient screening [16]. Additionally, protein library screening and antibody library screening technologies have demonstrated tremendous potential in drug development and immunotherapy [17]. By combining mutation libraries with crosslinking technology and systematically screening for interaction information at different sites, this approach reduces the reliance on precise protein structural and functional data. It facilitates the comprehensive mapping of protein-protein interaction networks and provides deeper insights into the study of relevant biological processes. PSMD10 is a eukaryotic protein that functions as a subunit of the 19 S regulatory particle of the 26 S proteasome, thereby playing an essential role in the protein degradation machinery. Its structure is composed almost entirely of ankyrin repeat (ANK) domains, which endow the protein with strong protein-protein interaction capabilities [18]. Functionally, PSMD10 directly binds to the E3 ubiquitin ligase MDM2, facilitating the interaction between MDM2 and p53, enhancing p53 ubiquitination, and accelerating its proteasomal degradation, ultimately suppressing apoptosis [19, 20]. In addition, PSMD10 interacts directly with CDK4 and increases its phosphorylation efficiency toward Rb, thereby promoting cell-cycle progression [21]. With regard to autophagy regulation, PSMD10 engages in direct cytoplasmic interaction with ATG7 and modulates autophagy-related signaling pathways [22]. Given its interactions with multiple key proteins, PSMD10 is considered a protein of substantial therapeutic potential. Ongoing research targeting its interaction interfaces, particularly the hotspot regions, has highlighted PSMD10 as a promising candidate for small-molecule or peptide-based inhibitor development. The 4-azido-l-phenylalanine (Azi) is a commonly used azide-based photocrosslinker that contains a photolabile azido group. Under ultraviolet irradiation at approximately 365 nm, the azido group generates highly reactive nitrenes or nitroso radicals, which readily insert into adjacent C-H or N-H bonds to form stable covalent linkages. Using Azi as a crosslinker, Jingxiang Li et al. identified an interaction between NleE and 26 S Proteasome Regulatory Subunit 10 (PSMD10), revealing that they act together to suppress autophagy [23]. In this study, we employed a scanning mutagenesis library strategy in which amino acids across the selected regions of PSMD10 were individually replaced with amber codons, enabling incorporation of the ncAA Azi at multiple independent sites in a single experiment and subsequent covalent crosslinking. Coupled with crosslinking mass spectrometry, we performed library-based screening to identify PSMD10 interaction partners and map precise crosslinking sites in the native cellular environment. This approach substantially expands the coverage of interaction capture, allowing systematic interrogation of multiple potential functional regions on the protein surface, providing a more comprehensive and reliable delineation of protein interaction interfaces, and thereby increasing the likelihood of identifying authentic interaction partners. Methods Plasmids and antibodies All assays were performed with KAPA HiFi HotStart DNA Polymerase in a Bio-Rad S1000™ Thermal Cycler. The constructs EF1α-PSMD10(WT, XTAG)-HA-V5-P2A-Hygro have previously been described [24]. pIRE4-Azi was purchased from addgene (#105829). The coding sequence (CDS) of PSMD10 (678 bp, encoding 226 amino acids) was obtained from the NCBI RefSeq database (Gene ID: 5716). TAG mutations were designed and introduced within the amino acid region 37–72, and the corresponding oligonucleotide library was synthesized by GenScript. The synthesized oligonucleotide pool was then used as the template for PCR to generate the target fragments. The purified PCR products were assembled into the linearized EF1α-X-HA-V5-P2A-Hygro backbone using Gibson Assembly. The assembled constructs were subsequently transformed into competent cells, followed by colony screening and collection to obtain the final plasmid library. Antibodies against HA-tag (AE105), V5-tag (AE017) were purchased from ABclonal. Protein A/G magnetic beads (88802) were purchased from Thermo Fisher Scientific. Library construction The coding sequence (CDS) of PSMD10 (678 bp, encoding 226 amino acids) was obtained from the NCBI RefSeq database (Gene ID: 5716). Previous crosslinking assays identified residue 39 as a crosslinking site with E. coli proteins [23]. Accordingly, a focused library covering residues 37–41 was designed centered around this site. Residue 72 also showed a crosslinking band in the same experiments, prompting the inclusion of residues 70–74 for validation. In addition, residues 104 and 106, located at analogous positions in the next ankyrin repeat, were tested as exploratory targets based on sequence alignment. Due to the length limitation of synthetic oligonucleotides (108 bp maximum, encoding 36 amino acids), the full-length PSMD10 sequence was divided into seven consecutive 36-amino-acid fragments. The second fragment (residues 37–72) fell within the region of interest described above. TAG mutations were introduced within these regions, and the corresponding oligonucleotide library was synthesized by GenScript (Piscataway, NJ, USA). The oligonucleotide pool was PCR-amplified using forward and reverse primers with KAPA HiFi HotStart DNA Polymerase. The vector backbone (EF1α-X-HA-V5-P2A-Hygro) was linearized by double restriction digestion with EcoRI-HF and BsmBI-v2 at 37 °C for 3 h in NEBuffer r3.1. The purified backbone and amplified insert were assembled using Gibson Assembly at 50 °C for 30 min. The assembled product was electroporated into Stbl3 electrocompetent cells, recovered in SOC medium at 37 °C for 1 h, and plated on LB-agar plates containing 50 µg/mL ampicillin. After overnight incubation at 30 °C, colonies were collected and plasmid DNA was extracted using the FastPure EndoFree Plasmid Midi Kit. Library coverage and mutation accuracy were validated by Sanger sequencing and NGS. For complete details of the cloning procedures, including PCR cycling conditions and primer sequences for all sub-libraries, please refer to our previous work [24]. KAPA HiFi HotStart DNA Polymerase (07958927001) was purchased from Roche, EcoRI-HF (R3101V) and BsmBI-v2 (R0739S) were purchased from NEB; NEBuffer r3.1 (B7030S) was purchased from NEB, MultiF SeamLess Assembly Mix (RK21020) was purchased from ABconal, FastPure EndoFree Plasmid Midi Kit (DC205) was purchased from Vazyme. Transient transfection and photocrosslinking HEK293T cells were cultured in DMEM supplemented with 10% (v/v) fetal bovine serum (FBS) and 1% (v/v) penicillin/streptomycin (Gibco) and maintained in a humidified incubator at 37 °C with 5% CO₂. When cells reached approximately 70–80% confluency in 15 cm dishes, they were transfected with plasmids encoding the respective TAG constructs (EF1α-PSMD10(X-TAG)-HA-V5-P2A-Hygro,8 µg) and the orthogonal tRNA/aminoacyl-tRNA synthetase system plasmid (PIRE4-Azi,12 µg) at a 1.5:1 ratio using PEI according to the manufacturer’s instructions. Twelve hours post-transfection, the medium was replaced with fresh DMEM containing 1 mM Azi (HY-16714, MCE). Due to the photosensitivity of Azi, all procedures were performed under dark conditions. 48 h post-transfection, cells were harvested and collected by centrifugation. Each sample was derived from one 15 cm dish (~ 1.5 × 10⁷ cells). Cell pellets were resuspended in PBS and subjected to photo-crosslinking at 4 °C. For crosslinking, the cell suspension was placed under a UV lamp (365 nm) at a distance of 5 cm and irradiated for 25 min. The suspension was then resuspended, cells were collected by centrifugation, the supernatant was removed, and the cell pellets were used for subsequent protein extraction and analysis. Pulldown assays Cell pellets were resuspended in 1 mL of prechilled PBS and washed twice by pipetting to ensure thorough resuspension. Lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.5% Triton X-100) containing 2 × protease inhibitor (PI) was then added, and cells were lysed on ice for 30 min, with brief vortexing every 5 min. Following lysis, samples were centrifuged at 17,000 × g for 15 min, and the supernatant was transferred to a fresh 1.5 mL tube. SDS loading buffer (1× final concentration) was added, and samples were heated at 99.5 °C for 10 min. For SDS-PAGE, 20 µL of each sample was loaded per well. After electrophoresis, proteins were transferred to membranes, blocked, and probed with either anti-HA or anti-V5 antibodies for Western blot analysis. An appropriate amount of Mouse anti-HA-Tag monoclonal antibody was added to the remaining protein lysate and incubated overnight at 4 °C on a temperature-controlled rotary shaker. Subsequently, an appropriate amount of Protein A/G magnetic beads was pretreated by washing several times with pre-chilled lysis buffer followed by a final wash with ultrapure water. The pretreated beads were then added to the antigen–antibody complex and incubated at room temperature with rotation. After incubation, magnetic separation was performed to remove the supernatant, and the beads were washed again. The beads were then resuspended in lysis buffer containing protease inhibitors and rotated at room temperature, followed by the addition of SDS loading buffer and heating at 99.5 °C to fully dissociate the antigen–antibody complexes. A portion of the dissociated sample was subjected to SDS–PAGE, and after Western blot analysis, Coomassie Brilliant Blue staining was used to confirm the target band. The gel slice corresponding to the V5-tagged protein was excised after destaining, stored in ultrapure water at low temperature, and submitted for LC-MS/MS cross-linking mass spectrometry analysis. Mass spectrometry Protein samples after photo-crosslinking were analyzed by a specialized service provider (Jiahua Yaruibiotechnology Co., Ltd.). Samples were digested overnight (14–16 h) at 37 °C with trypsin, followed by desalting using C18 solid-phase extraction (SPE) columns to remove interfering small molecules. Peptide samples were analyzed using an UltiMate 3000 UPLC system coupled to an Orbitrap Exploris 480 high-resolution mass spectrometer (Thermo Scientific). Peptides were separated on a 75 μm i.d. × 20 cm C18 column and introduced into the Orbitrap Exploris 480 LC-MS system. The peptide separation gradient was set as follows: 0–2 min, 6% → 6.5% B; 2–37 min, 6.5% → 23% B; 37–54 min, 23% → 36% B; 54–54.5 min, 36% → 55% B; 54.5–55 min, 55% → 99% B; 55–60 min, 99% B for column equilibration, at a flow rate of 400 nL/min. Mass spectrometry data were acquired in data-dependent acquisition (DDA) mode. MS scans were collected over a mass range of 400–1200 m/z at a resolution of 60,000 at m/z 200, with an AGC target of 3 × 10⁶ and a maximum injection time of 50 ms. MS/MS scans were acquired with a cycle time of 1.5 s, AGC target of 1 × 10⁵, and maximum injection time of 50 ms. Mass spectrometry data were analyzed using pLink [25, 26] and Aixaa [27] software. Database searches included possible mutated protein sequences resulting from incorporation of the photo-crosslinker Azi into PSMD10, as well as the UniProt Homo sapiens protein database. A false discovery rate (FDR) of < 1% was applied. The precursor ion mass tolerance was set to 20 ppm, allowing up to two missed tryptic cleavages. The non-natural amino acid Azi was defined with a molecular formula of C9H7N2O1. Results To validate the crosslinking signals observed in the library-based screening, we performed single-site crosslinking assays for selected residues. Within the 37–72 region, residues D37, Q38, D39, S40, R41, D70, D71 and A72 were tested. As controls, residues outside this region (G73, W74, Q104, N105, G106, Y112) were also examined. The crystal structure of PSMD10 (PDB ID: 7VO6) was used for post-hoc visualization of these residues. Using Azi as a crosslinking probe, immunoblot analysis of the single-site mutation experiments revealed crosslinked bands in several mutants, indicating the presence of distinct crosslinking signals at variable positions. Further analysis showed that multiple TAG mutants exhibited crosslinked bands at different molecular weight positions, suggesting that PSMD10 may interact with multiple endogenous proteins at several sites. Notably, mutation of alanine at position 72 to TAG resulted in the strongest western blot (WB) signal, indicating that this position may have unique conformational properties or increased susceptibility to crosslinking. Similarly, substitution of Q38, S40 and N105 with a TAG codon also yielded clear and sharp crosslinked bands, suggesting efficient crosslinking at these sites (Fig. 1). After crosslinking, PSMD10 forms various high-molecular-weight complexes with endogenous proteins. The differential accessibility of V5 and HA epitopes in these complexes may result in non-identical band patterns between anti-V5 and anti-HA blots. Mapping the interactome of PSMD10 using photo-crosslinking. (A) Schematic representation of the molecular structure of Azi. (B) Structural model of the PSMD10 (PDB:7VO6), with residues 37–72 corresponding to the region harboring TAG codon mutations. (C, D) Immunoblot analysis reveals the intracellular interactions of PSMD10 point mutants. The crosslinking regions are marked by annotated lines beside the immunoblot. Full-length blots are presented in Supplementary Fig. 1 To generate the mutant library, we used an oligo-based cloning strategy (Fig. 2). This library comprises 36 different substitution constructs, each designed to replace a single amino acid within residues 37–72 of PSMD10 with a TAG codon, thereby enabling the site-specific incorporation of Azi. The corresponding 108 bp variable region was synthesized as an oligonucleotide pool and assembled into the expression vector in a pooled format. Next-generation sequencing (NGS) confirmed both the mutation coverage and positional accuracy, with the abundance of each TAG substitution varying less than 10-fold, meeting quality control criteria (Fig. 2B). The library was transfected into HEK293T cells along with plasmids encoding the Azi-specific aminoacyl-tRNA synthetase (aaRS) and tRNA, followed by treatment with 1 mM Azi to allow its incorporation. After UV-activated crosslinking, Azi-incorporated PSMD10 proteins were enriched from cell lysates using the fused 6×His and HA tags, as confirmed by western blot, and were subsequently subjected to crosslinking mass spectrometry (Fig. 2A). PSMD10 library construction, validation, and photo-crosslinking MS workflow. (A) The mutations in the library are concentrated in a 108 bp variable region, with 19 bp primer binding regions on both sides to ensure specific amplification of the target sequence. The variable region contains a TAG mutation for site-specific incorporation of Azi. The synthesized oligonucleotide pool was PCR-amplified and then cloned into a vector using Gibson assembly, resulting in the library plasmids. After adapter ligation, the library plasmids were analyzed by NGS to validate mutation coverage and accuracy. The validated library plasmids were co-transfected with Azi plasmids into HEK293T cells, incubated with 1 mM Azi, and crosslinked by UV activation. Purified proteins were separated by SDS-PAGE, and the target region was excised. The final samples were analyzed by crosslinking mass spectrometry. (B) The x-axis indicates the amino acid positions, and the y-axis represents the percentage of total NGS reads containing a TAG codon at each amino acid position, calculated as (reads at specific position / total TAG reads across all positions in the same replicate) × 100%. Data are shown as the mean of two independent replicates, with error bars representing standard deviation (SD) When the scanning library was used for crosslinking validation, crosslinked bands were observed in the library samples that were clearly distinct from those in the WT control (Fig. 3A). WB analysis of purified PSMD10 pull-down products revealed multiple crosslinked bands in the library samples, spanning molecular weights from approximately 55 to 115 kDa. Notably, a crosslinked band detected in the pull-down from the A72 single mutant was also present in the library sample, albeit at reduced intensity. This reduction is expected, as in the pooled library only ~ 1/36 of the incorporated Azi resides at position 72, while Azi incorporation at the remaining 35 positions can generate additional, distinct crosslinked products (Fig. 3B). The purified library pull-down products were subsequently excised from the gel and subjected to crosslinking mass spectrometry analysis. Identification of PSMD10 interaction partners by photo-crosslinking and purification. (A) Western blot analysis of crosslinked samples. Markers are indicated. The crosslinking regions are marked by annotated lines beside the immunoblot. WT and P-Lib samples are technical replicates (loaded in duplicate). Full-length blots are presented in Supplementary Fig. 2. (B) Western blot analysis of crosslink-purified samples. Markers are indicated. Red boxes indicate crosslinked bands. Full blots in Supplementary Fig. 2 Among the protein-protein interaction partners of PSMD10 identified by mass spectrometry, proteins with multiple crosslinking features were selected as candidate interactors due to their higher reliability. Three proteins, T2FA (UniProt ID: P35269), TBA1C (UniProt ID: Q9BQE3), and ATRIP (UniProt ID: Q8WXE1), were identified as being in spatial proximity to PSMD10 and their precise interaction sites were mapped in high-resolution mass spectrometry (Figs. 4, 5 and 6). Although these interactions have not been previously reported, the subcellular localizations of these proteins are consistent with that of PSMD10, suggesting the likelihood of their interaction. Notably, two distinct interaction pairs were identified for each protein, with the corresponding crosslinking sites on PSMD10 located in close proximity. This clustering suggests that the crosslinking events are reproducible and fall within the effective capture range of Azi. Together, these observations demonstrate the robustness of the approach, argue against random crosslinking, and further support the authenticity of the identified interactions. Crosslinking mass spectrometry analysis identifies specific interaction interfaces between PSMD10 and T2FA. When valine at position 64 or 66 of PSMD10 was mutated to TAG, crosslinking occurred with the amino acid at position 368 of T2FA. MS2 spectra are shown with b/y ions labeled. (Upper panel) In the absence of crosslinking, the m/z difference between the y5 and y7 ions would be approximately 245.27. The experimentally measured difference is 912.6 (indicated by arrows). (Lower panel) In the absence of crosslinking, the m/z difference between the y3 and y7 ions would be approximately 441.55. The experimentally measured difference is 1110.68 (indicated by arrows). These observed mass shifts, together with the consistent b/y ion series across multiple independent spectra, support the crosslinking site assignments Crosslinking mass spectrometry analysis identifies specific interaction interfaces between PSMD10 and TBA1C. When TAG was introduced at position 70 or 71 of PSMD10, crosslinking occurred within the RAV sequence (residues 373–375) of TBA1C. In the absence of crosslinking, the two linear peptides VQ-AR and UD-AG would have m/z values of 2235.546 and 1540.672, respectively. The experimentally detected precursor ion at m/z = 3761.1 matches the sum of the two peptides minus the mass loss from crosslinking (H + NH): 3761.1 ≈ 2235.546 + 1540.672 − 14 − 1 (indicated by arrows). The remaining b/y ions confirm the amino acid sequences of both peptides. Due to the lack of diagnostic fragment ions flanking a single residue, the crosslinking site is assigned at the peptide level rather than the residue level Crosslinking mass spectrometry analysis identifies specific interaction interfaces between PSMD10 and ATRIP. When aspartic acid at position 70 or 71 of PSMD10 was mutated to TAG, crosslinking occurred with the amino acid at position 296 of ATRIP. MS2 spectra are shown with b/y ions labeled. (Upper panel) In the absence of crosslinking, the b1 ion would have an expected m/z of ~ 188.2. The experimentally identified ion appears at m/z = 1266.68 (indicated by arrows). (Lower panel) In the absence of crosslinking, the m/z difference between the βb1 and b2 ions would be approximately 188.23. The experimentally measured difference is approximately 1267.67 (indicated by arrows). These observed mass shifts support the crosslinking site assignment Overall, these results demonstrate the power of our library-based photocrosslinking approach to capture and dissect multiple weak or transient protein-protein interactions in a single experiment. By integrating a site-saturation mutagenesis library with ncAA-mediated photocrosslinking, this strategy reduces the dependency on prior structural data, overcomes the limitations of single-site analysis, and expands the space for exploring protein interactions. Consequently, our approach enables more comprehensive interaction mapping, provides mechanistic insight into biological processes, and generates novel candidates for functional validation. Discussion Traditional approaches for studying PPIs have been instrumental in delineating interaction networks, yet they face substantial limitations in capturing weak or transient interactions within living cells. Photo-crosslinking strategies based on ncAAs offer a highly specific and stable solution; however, conventional implementations typically incorporate ncAAs at a single residue, resulting in incomplete coverage of interaction interfaces and the potential omission of functionally critical sites. To overcome this limitation, we constructed an amber codon-scanning mutagenesis library of PSMD10, enabling the incorporation of the photo-crosslinker Azi at multiple distinct positions and thereby achieving systematic interrogation of key structural regions across the protein surface. By integrating single-site mutant validation, library-scale crosslinking screens, and crosslinking mass spectrometry, we successfully captured the specific interaction network of PSMD10 within its native cellular environment and mapped multiple intermolecular crosslinking sites. We selected candidate interaction partners with multiple crosslinking features, including several previously unreported proteins (such as ATRIP, T2FA, and TBA1C). In the cross-linked library samples, multiple cross-linking sites were detected, further validating the effectiveness of combining the scanning mutation library strategy with covalent cross-linking experiments, and demonstrating the potential of this approach for screening of protein interaction cross-linking sites. We further examined the potential biological relevance of ATRIP, T2FA, and TBA1C as candidate interaction partners of PSMD10. Although direct interactions between PSMD10 and these proteins have not been previously reported, all three are involved in biological processes that are known to be associated with PSMD10. PSMD10 is a multifunctional protein involved in cell proliferation, proteasome regulation, and autophagy. TBA1C (α-tubulin) contributes to the cytoskeletal scaffolding of proteasomes, linking to PSMD10’s role in proteasome function. T2FA is a transcription factor involved in the RB1 pathway, which is consistent with PSMD10’s known function in cell cycle regulation via RB1. ATRIP is a key component of the DNA damage response pathway and has been implicated in proteasome crosstalk. The shared involvement of these proteins in PSMD10-related biological processes supports the possibility that they represent genuine interaction partners, warranting further experimental validation. Several lines of evidence from our data support that ATRIP, T2FA, and TBA1C are specific crosslinking targets rather than non-specific proximity hits. First, crosslinking was strictly dependent on the presence of the TAG mutation (U), as wild-type PSMD10 without U produced no crosslinked bands. Second, the crosslinks occurred at specific residue pairs (e.g., PSMD10 U70/U71 with the RAV sequence of TBA1C), rather than being randomly distributed across protein surfaces. Third, two independent TAG mutations (D70 and D71) both crosslinked to the same peptide region of TBA1C, providing mutual internal validation. Fourth, all crosslinked peptides were identified with high-confidence MS scores (> 0.99 for key identifications) and stringent 1% FDR filtering. Fifth, the overall background of non-specific crosslinking was low, as evidenced by the absence of common contaminant proteins. Based on these criteria, a list of the top 50 high-confidence PSMD10-interacting proteins is provided in Supplementary Table S1. Regarding protein abundance, ATRIP, T2FA, and TBA1C are relatively abundant proteins. Their higher abundance likely contributes to their preferential detection in our crosslinking experiments, as more abundant proteins have a higher probability of being crosslinked and identified by MS. This does not diminish the specificity of the identified interactions; rather, it explains why these particular candidates were captured among the broader proteome. Notably, known PSMD10 interactors such as MDM2, CDK4, and ATG7 were not detected in our crosslinking MS results. This is expected given the spatial constraint of the Azi crosslinker (~ 6–10 Å) and the fact that our TAG mutations were introduced within residues 37–72 of PSMD10. These known interactors likely bind to regions outside this fragment, such as the N-terminal or C-terminal domains, which are not captured by our library. The absence of these known interactors further supports the site-specific nature of our crosslinking approach and argues against non-specific capture of abundant or background proteins. We note that structural validation of the identified crosslinks, such as assessing solvent accessibility of the crosslinked residues and measuring spatial distances between crosslinked pairs using interaction prediction programs (e.g., ColabFold), would further strengthen the technical validity of the findings. These analyses will be pursued in future studies. Collectively, this study introduces multiplexed, broad-coverage strategy for mapping protein interaction interfaces and establishes a robust foundation for elucidating PSMD10’s biological functions in cell proliferation, autophagy regulation, and proteasome activity. This approach can be readily extended to other biologically important proteins and offers a promising technical platform for developing therapeutics targeting protein-protein interaction interfaces. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 High-resolution source file for figures 4 to 6 Acknowledgements Not applicable. Author contributions YL: Conceptualization, Investigation, Methodology, Validation, Writing- original draft, Writing-review and editing. SM: Methodology, Validation, Writing-review and editing. YZ: Methodology, Writing-review and editing. FW: Conceptualization, Resources, Writing-review and editing. TS: Conceptualization, Resources, Writing-review and editing. YC: Software, Formal Analysis, Visualization, Writing-review and editing. SC: Conceptualization, Project administration, Supervision, Funding acquisition, Writing-review and editing. DZ: Conceptualization, Project administration, Supervision, Funding acquisition, Writing-review and editing. Funding This research was funded by the Strategic Priority Research Program of the Chinese Academy of Sciences, grants No. XDB1260302 and XDB0830300, Innovative Drug Research and Development National Science and Technology Major Project, grant 2025ZD1804100, and the National Natural Science Foundation of China, grant 82373972. Data availability The mass spectrometry proteomics data generated in this study have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the iProX partner repository [1, 2] with the dataset identifier PXD075032. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Song Chen, Email: sc43@sanger.ac.uk. Dongxin Zhao, Email: zhaodongxin@simm.ac.cn. References Associated Data Supplementary Materials Supplementary Material 1 High-resolution source file for figures 4 to 6 Data Availability Statement The mass spectrometry proteomics data generated in this study have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the iProX partner repository [1, 2] with the dataset identifier PXD075032.