High-Ratio Partial Proteolysis with Carrier Proteome (HOLSER) Enables Global Structure Profiling and Site-Resolved Elucidation of Ligand–Protein Interactions Authors: Xuepei Zhang, Hassan Gharibi, Bohdana Sokolova, Zhaowei Meng, Hezheng Lyu, Amir Ata Saei, Massimiliano Gaetani, Roman A. Zubarev Published: 2026 ABSTRACT Understanding how cellular proteins interact with their environment, including endogenous and exogenous molecules, is critical for elucidating mechanisms of cellular regulation and drug action. Partial proteolysis-based techniques offer peptide-level resolution of ligand-induced conformational changes but are limited by modest proteome coverage and depth, as well as sensitivity to the experimental conditions. To overcome these limitations, we developed high-ratio partial proteolysis with carrier proteome (HOLSER), an efficient workflow that features an extended digestion time for reduced peptide yield variability as well as tandem mass tag multiplexing that includes full digests for enhanced proteome depth and sequence coverage, as well as higher precision of peptide abundance measurements. We demonstrate HOLSER capabilities of probing structural changes on the scale of specific binding sites for kinase target mapping, individual protein domains for structural mapping of the FKBP–mTOR complex in response to rapamycin, as well as global proteome structure profiling. Introduction Characterizing protein native structures and their changes induced by protein–ligand interactions is essential for understanding cellular signaling, the molecular basis of disease, and drug mechanisms. Traditional approaches such as thermal shift assays , and affinity-based pull-downs − often lack the resolution or throughput required to profile complex proteomes under physiologically relevant conditions. In recent years, limited proteolysis-mass spectrometry (LiP-MS) − and the peptide-centric local stability assay (PELSA) have emerged as powerful tools to detect ligand-induced conformational changes at the peptide level. These profiling techniques rely on limited proteolysis of native proteins, where ligand-stabilized regions resist enzymatic cleavage with detection of the ligand-induced changes via quantitative mass spectrometry. The most recent such approach, PELSA, uses lower-grade trypsin at a high enzyme–substrate ratio (1:2) in combination with a very short digestion time. PELSA opened the way for scalable structural analysis at the proteome level and demonstrated spectacular results in selected cases. But despite these significant promises, the technique possesses some critical limitations. Due to the shallow digestion, PELSA exhibits modest proteome depth and limited sequence coverage, which significantly hampers the statistical power and generality of the analysis. Due to the short, 1 min digestion time, the approach may suffer from stochastic quantification variability, particularly for low-abundance proteins and transient interactions. This is exacerbated by the transition from peptides to the protein level, critical for drug target elucidation, hinging in PELSA on the detection and correct quantification of a single “best” peptide. These shortcomings limit the applicability of this otherwise highly promising technique in high-throughput ligand screening and domain-level interaction mapping. The recently developed high-throughput PELSA (HT-PELSA), a scaled-up adaptation of the original PELSA approach, extends its analytical capabilities. However, the inherent limitation of low sequence coverage due to restricted proteolysis remains unaddressed. To address these challenges, we developed HOLSERan optimized workflow for ligand-induced stability profiling in cell lysate under native conditions (Figure A). Inspired by PELSA to use a high enzyme-to-substrate (E/S) ratio of 1:1 (w/w) and lower-grade trypsin, HOLSER features several critical differences. First, the digestion time is extended from one to 4 min to improve cleavage uniformity and reduce stochastic variability of the peptide yield. Then, instead of the label-free analysis that suffers significantly from the random fluctuations of electrospray current, HOLSER employs the more precise tandem mass tag (TMT)-based quantification. Also, data-dependent acquisition is used that is less prone to false positive results , than the data-independent acquisition (DIA) employed in PELSA. Importantly, in the same TMT set, partial digests in HOLSER are multiplexed with “carrier” channels containing a digest obtained in full (8 h long) proteolysis. As in single-cell proteomics, the main purpose of the carrier channel is to boost sensitivity and detection probability, especially for low-abundance peptides from protected regions, and thus increase the analysis depth (number of quantified peptides and proteins) and average protein sequence coverage. Also, comparison of the peptide abundances in the partial digest versus the full digest provides the “accessibility factor” that reflects protein tertiary structure even in the absence of a ligand. With TMT, improved quantitative precision allowed the number of biological replicates to be reduced from four to three without apparent loss of statistical power. While the current workflow requires additional LC-MS time because of deep fractionation, TMT multiplexing improves sample throughput and offers increased scalability as multiplexing capacity expands. Finally, instead of a single “best” peptide, HOLSER uses several peptides for determining the target protein, with their p-values combined using Fisher’s formula. This provides robust and statistically viable target identification. Overview of experimental workflow and TMT labeling. (A) Workflow of HOLSER. E/S ratio (w/w). (B) TMT labeling scheme showing the assignment of unique tags to samples from different experimental groups and the inclusion of carrier channels. (C) Box plot showing relative peptide abundance for each TMT channel normalized to the carrier (dashed line at 100%). In this proof-of-principle study, we apply HOLSER to verify known ligand–protein interactions with high spatial resolution. We benchmark HOLSER performance against PELSA in global proteome profiling, staurosporine kinase target deconvolution, and targeted analysis of the FKBP–mTOR complex under rapamycin treatment. The results confirm that, as expected, HOLSER significantly improves proteome depth, protein sequence coverage, and the robustness of ligand-induced stabilization, additionally allowing for domain-level global structural mapping. Experimental Section Cell Culture HeLa cells were purchased from ATCC. Cells were grown at 37 °C in Dulbecco’s Modified Eagle’s Medium (DMEM, Thermo Fisher Scientific, Cat. 11594446) supplemented with 10% heat-inactivated fetal bovine serum (FBS, Thermo Fisher Scientific, Cat. 11560636) and 1% penicillin/streptomycin (Gibco, Cat. 15140–122) in a humidified atmosphere containing 5% CO2. HOLSER Assay HeLa cells were collected and washed with PBS twice. The cells were resuspended in 20 mM EPPS (Merck, E9502) buffer at pH 8.2 containing protease inhibitors (Thermo Fisher Scientific, Cat. 78439) and were subjected to three freeze/thaw cycles to lyse the cells. The protein concentration was measured using a rapid protein quantification assay kit (Thermo Fisher Scientific, Pierce 660 nm) according to the manufacturer’s protocol. Then the lysate was diluted to 1 mg/mL with the lysis buffer (EPPS buffer at pH 8.2). 50 μL of cell lysates were incubated with 5 μL of the ligand solution in DMSO or DMSO only for 30 min at 25 °C in three replicates. The final concentration of the ligand was 10 μM for MTX, 10 μM for staurosporine, and 2 μM for rapamycin. Limited trypsinolysis was carried out at an enzyme-to-substrate ratio of 1:1 (w/w) for a total duration of 4 min. Specifically, 10 μL of trypsin (Sigma-Aldrich, T1426) 5 mg/mL stock solution was added to 50 μg of protein sample. The mixture was gently and thoroughly mixed for 30 s using a multichannel pipet to ensure even distribution. The samples were then incubated at 25 °C for 3.5 min in a ThermoMixer F1.5 (Eppendorf), being shaken at 1000 rpm. For carrier proteome samples, trypsin digestion was carried out for 8 h. Proteolysis was stopped by adding 165 μL of 8 M guanidine hydrochloride (GdmCl) in 20 mM EPPS buffer (pH 8.2), resulting in a final concentration of 6 M GdmCl. To alkylate free cysteine residues via carbamidomethylation, tris­(2-carboxyethyl)­phosphine (TCEP) was added to a final concentration of 10 mM, followed by incubation at 95 °C for 5 min. Subsequently, iodoacetamide (IAA) was added to a final concentration of 40 mM. The samples were then diluted to a final concentration of 1 M guanidine hydrochloride (GdmCl) using 20 mM EPPS buffer (pH 8.2). The peptide samples were labeled using 18 TMTpro reagents (Thermo Fisher Scientific, Cat. A52045) according to the manufacturer’s instructions. Three carrier channels were assigned to the lowest-mass reporter channels, followed by two empty channels to create a buffer region before the partial proteolysis samples. The labeling reaction was quenched by 0.5% hydroxylamine (Thermo Fisher Scientific, Cat. 15225753). After pooling together the individually labeled samples, the peptides were acidified by trifluoroacetic acid (TFA) to a final concentration of 1% and desalted using a C18 Sep-Pak cartridge. The cartridge was first conditioned by washing twice with 1 mL of 100% acetonitrile (ACN), followed by two washes with 1 mL of 0.1% TFA to equilibrate the resin. Acidified peptide samples were then loaded onto the cartridge and washed twice using 5% ACN with 0.1% formic acid (FA). Peptides were eluted from the cartridge using 50% ACN with 0.1% FA in two steps of 300 μL each. The eluates were collected and subsequently dried. Prior to LC-MS/MS analysis, dried peptides were reconstituted and fractionated by high-performance liquid chromatography with a Dionex Ultimate 3000 UPLC system (Thermo Fisher Scientific) as previously described. , The generated 96 fractions were concatenated into 48 by pooling together every fourth fraction. For LC-MS/MS analysis, 1 μg of each pooled fraction was injected and analyzed. NanoLC-MS/MS Analysis NanoLC-MS/MS analyses were performed on an Orbitrap Exploris 480 mass spectrometer equipped with an EASY electrospray ionization source and connected online with an Ultimate 3000 nanoflow UPLC system (all from Thermo Fisher Scientific). The samples were preconcentrated and desalted online using a PepMap C18 nanotrap column (length-2 cm; inner diameter-75 μm; particle size-3 μm; pore size-100 Å; Thermo Fisher Scientific) with a flow rate of 3 μL/min for 5 min. Peptide separation was performed on an EASY-Spray C18 reversed-phase nanoLC column (Acclaim PepMap RSLC; length-50 cm; inner diameter-2 μm; particle size-2 μm; pore size-100 Å; Thermo Fisher Scientific) at 55 °C and a flow rate of 300 nL/min. Peptides were separated using a binary solvent system consisting of 0.1% (v/v) FA and 2% (v/v) ACN (solvent A) as well as 98% ACN (v/v) and 0.1% (v/v) FA (solvent B). The peptides were eluted with a gradient of 3–26% B in 97 min, and 26–95% B in 9 min. Subsequently, the analytical column was washed with 95% B for 5 min before re-equilibration with 3% B. The mass spectrometer was operated in a data-dependent acquisition (DDA) mode. A survey mass spectrum (from m/z 375 to 1500) was acquired in the Orbitrap analyzer at a nominal resolution of 120,000. The automatic gain control (AGC) target was set as 100% standard, with the maximum injection time of 50 ms. In a 3 s DDA cycle, the most abundant ions with charge states 2+ to 7+ were isolated, fragmented using HCD MS/MS with 33% normalized collision energy, and detected in the Orbitrap analyzer at a nominal mass resolution of 50,000. The AGC target for MS/MS was set as 250% standard with a maximum injection time of 100 ms, whereas dynamic exclusion was set to 45 s with a 10 ppm mass window. HDX-MS Assay Recombinant KPNA2 was kindly provided by Saei’s lab at Karolinska Institutet. Recombinant FKBP3 protein was obtained from Elabscience (PKSH032876). For HDX labeling, 4 μL of KPNA2 (5 μM) or FKBP3 (10 μM) was mixed with 36 μL of stock buffer prepared in D2O (Sigma, 151882) and incubated at 10 °C for 30 s. The buffer composition for KPNA2 was 20 mM Tris-HCl, 50 mM NaCl, 2.5 mM DTT, pH 8.0, whereas for FKBP3 it was 20 mM Tris-HCl, 1 mM DTT, pH 8.0. After adding 1 volume of ice-cold 1.8 M guanidine hydrochloride in 0.8% formic acid, the samples were analyzed in an automated HDX-MS system (TRAJAN) in which injected samples were automatically digested, cleaned, and separated at 4 °C. Samples were digested using an Enzymeate BEH pepsin column (Waters) followed by a 3 min desalting step using 0.1% FA at 100 μL/min. Peptic peptides were then separated by a 21 mm I.D × 50 mm length C18/3 μm column using a 21 min/10%–90% acetonitrile gradient in 0.1% formic acid at 40 μL/min. An Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific) operated at 60,000 resolution at m/z 200 was used for analysis. Raw files were searched against the protein databases containing the KPNA2 or FKBP3 amino acid sequence using the MASCOT search engine. The HDExaminer software (Sierra Analytics) was used to process all HDX-MS data. Proteomics Data Analysis LC-MS/MS data were analyzed using Proteome Discoverer version 3.1 (Thermo Scientific). Database searches were performed against the UniProt human reference proteome (Proteome ID: UP000005640_9606; downloaded in May 2025), containing 20,663 reviewed protein entries, for peptide and protein identification and quantification. TMT reporter ion abundances were normalized to the total abundance in a given TMT channel. Statistical Analysis Statistical significance of peptide abundance changes between treated and untreated samples was assessed using a two-sided, unpaired Welch’s t test, with p-values (p p) calculated across biological triplicates. The peptide data were translated to the protein level by selecting the top three peptides based on a combined score calculated as the absolute value of the log2 fold change in the treated sample versus the DMSO control multiplied by -log10­(p p). The fold change for the protein was calculated as the average of those for the selected three peptides, and the protein p-value was obtained from the p p values by combining them using Fisher’s formula. All statistical analyses were conducted using R (version 4.5.0). A p-value threshold of <0.05 was considered statistically significant. Ligand–protein Interaction Mapping The crystal structures of the studied protein–ligand complexes were retrieved and visualized in PyMOL (Schrödinger, LLC). The ligand and interacting residues were both shown in a stick representation, with the ligand colored brown and the residues colored cyan. An embedded Python script was used to identify residues within the interaction interface. The script generated a tab-delimited file listing the interacting residues together with their chain ID, residue number, interaction type, and distance to the ligand. Accessibility Analysis Peptides with a coefficient of variation (CV) below 30% were selected for accessibility analysis. For HDX-MS, per-residue deuterium uptake was calculated by averaging the uptake values of all peptides covering each residue after 30 s of exchange. For HOLSER, the log2 fold change in peptide abundance between 4 min and 8 h trypsin digestions was assigned to each residue covered by the corresponding peptide. Both deuterium uptake and log2 fold change values were standardized across the protein using z-score transformation to enable direct comparison of relative accessibility. Z-score transformation was applied by subtracting the mean and dividing by the standard deviation of each data set, centering the data, and expressing each value in terms of standard deviations from the mean. To reduce local variability and capture broader structural trends, residues were grouped into sequential bins of 8 residues. Within each bin, the average HDX-MS and HOLSER z-scores were computed along with the corresponding residue range. These binned values were then used to perform a Pearson correlation analysis, quantifying the relationship between HDX-MS and HOLSER-derived accessibility along the protein sequence. The data were first smoothed using a 20-point moving window to reduce noise, and a smoothing line was generated using a GAM with a spline on the x-axis, which produces a flexible, nonlinear curve that follows the trend of the data without assuming a specific parametric form. For visualization, per-residue z-scores were mapped onto the three-dimensional structure of the protein using PyMOL. For KPNA2, the AlphaFold-predicted model (e.g., AF–P52292-F1-model_v4.pdb) was used. For FKBP3, the crystal structure (PDB ID: 1pbk) was employed for mapping. A custom blue–white–red color ramp was applied, representing low (blue), intermediate (white), and high (red) accessibility. Residues with no data in either data set (i.e., not common to both HDX-MS and HOLSER) were shown in light yellow. Results HOLSER Enhances Proteome Analysis Depth and Sequence Coverage The 15 HeLa cell samples combined in a single TMT set included partial digests treated with vehicle DMSO (control), methotrexate (MTX), staurosporine, and rapamycin, as well as a full cell digest (Figure B). The multiplexed TMT set was separated into 48 fractions by reversed-phase chromatography and analyzed by LC-MS/MS using an Orbitrap mass spectrometer (Thermo). The obtained results were compared with data obtained in PELSA using HeLa cells, extracted in the form of final lists of proteins and peptides with relative quantification. Protein sequence coverage of the proteins reported in HT-PELSA was calculated using the peptide-level results for identifying dasatinib targets in crude K562 cell lysates in that study. For each protein, the start and end positions of all identified peptides were mapped onto the corresponding full-length protein sequence (retrieved from UniProt). Sequence coverage was defined as the percentage of amino acid residues of the protein that were covered by at least one peptide. Overlapping peptides were merged, and each protein was reported once. In HOLSER, the abundances in the TMT channels corresponding to partially digested samples were, on average, half of those in the fully digested “carrier” (Figure C, Supplementary Table 1A), reflecting the chosen balance between the depth of analysis and its specificity. The “empty” channels exhibited low intensity signals, likely due to TMT leakage from the carrier proteome channels. This leakage did not affect other channels that contained partially digested samples. At the peptide level, HOLSER quantified a total of 143,683 (Supplementary Table 1A) peptides, compared to 71,708 peptides in PELSAan increase by almost a factor of 2. Surprisingly, only 63% of the PELSA peptides were found in our much deeper analysis. This observation is frequently reported in proteomics when comparing data sets generated using different acquisition strategies and is consistent with the combined effects of stochastic precursor selection in DDA and previously reported challenges associated with peptide identification at low sequence coverage in DIA. , This result also warns against the reliance on a single peptide for target identification. The protein-level comparison showed a larger (79%) overlap between the two techniques, but HOLSER could still not confirm the presence of 1403 PELSA proteins while finding an additional 3445 proteins. The average protein sequence coverage, 39% in HOLSER (Supporting Information Table 2), was more than twice that of PELSA (17%) and HT-PELSA (17%), which is likely attributable to the combined effects of the HOLSER workflow, including the incorporation of full-digest carrier channels, extensive peptide fractionation, and deeper LC-MS analysis. The a priori probability of finding the correct target protein in the data set scales up, respectively. While shotgun proteomics can very rarely achieve 100% sequence coverage even for individual proteins, such as bovine serum albumin (BSA), frequently used for calibration and quality control, obtaining ≥50% sequence coverage may be considered a desirable target. In that respect, PELSAyielded 619 such proteins, while HOLSERyielded 3149 proteins (Supplementary Table 2), an increase by over 400%. The improved proteome depth and sequence coverage observed with HOLSER likely reflect the combined effects of the signal enhancement via the carrier proteome, reduction of peptide crowding via sample fractionation, missing value elimination due to TMT multiplexing, and other methodological optimizations. HOLSER Reveals Peptide-Resolved Stabilization of DHFR upon MTX Treatment MTX is a favorite test molecule for novel drug target identification techniques, as its treatment significantly changes (increases) both abundance and solubility of its cognate target, dihydrofolate reductase (DHFR). We have also successfully used MTX as a test molecule in our above-filter partial digestion technique, AFDIP, and it has also shown the expected behavior in PELSA. In HOLSER, multiple DHFR peptides exhibited strong ligand-induced protection (Figure A). When peptides were ranked by a score combining the absolute fold change with statistical significance, 13 top candidates were DHFR-derived peptides (Figure B, Supplementary Table 1B). Peptide and protein abundance changes upon MTX treatment. (A) Volcano plot displaying peptide abundance changes in MTX-treated samples compared with DMSO controls (n = 3 biological replicates). Peptides derived from DHFR are highlighted in purple. Statistical significance was assessed by t test. (B) Ranked list of peptides according to their combined score (absolute log2 fold change × – log10 p p) calculated from replicate data. (C) Volcano plot displaying protein-level abundance changes in MTX-treated samples compared with DMSO controls (n = 3 biological replicates). (D) Ranked list of proteins according to their combined score (absolute log2 fold change × – log10 p-value) calculated from replicate data. Note that not all DHFR peptides have shifted strongly, and in AFDIP, some peptides of the target protein shifted in the opposite direction. This is a distinct feature of the limited proteolysis techniques that provide structural resolution. Therefore, traditional shotgun proteomics aggregation of the peptide data into protein results poorly suits these techniques. Here we combined the ranks of the top three peptides for each protein to derive the protein rank (see Methods section), with DHFR expectedly appearing in the first position (Figure C,D and Supplementary Table 3). HOLSER Reveals Targets of Kinase-Targeting Staurosporine Staurosporine is the broad-spectrum kinase inhibitor. , To identify its targets in a PELSA-like way, all statistically significant peptides (p p < 0.05) were matched to the UniProt kinase database. HOLSER peptides identified 226 kinases as potential targets, as opposed to 191 in PELSA (Figure A), with a better-than-typical for proteomics 81% overlap between the two approaches (Supplementary Table 1C). In the kinases detected by HOLSER, 185 (81.9%) had at least one significant peptide located within the annotated kinase domain compared to 155 kinases (81.1%) in PELSA. Due to its deeper analysis, HOLSER uniquely identified several low-abundance transmembrane RGC kinases (Supplementary Figure 1A). When peptide data were aggregated into proteins using the top three peptides based on their combined score and with a Fisher-combined p-value <0.05, HOLSER identified 203 kinases as potential targets, and PELSA identified 151 kinases, with an 85.4% overlap between the two studies (Supplementary Figure 1B, Supplementary Table 4). Kinase-targeting analysis using HOLSER and PELSA. (A) Venn diagram showing overlap of kinase proteins identified from significantly regulated peptides in HOLSER and PELSA data sets. (B) Mapping of peptides identified in HOLSER and PELSA onto the CSK sequence. Dashed lines indicate the known binding sites from PDB. (C) Structural visualization of the CSK–staurosporine complex based on the crystal structure (PDB ID: 1BYG), rendered in PyMOL. As most, if not all, staurosporine targets are kinases, we explored the specificity of target identification in HOLSER and PELSA by two complementary approaches. In the first approach, the top peptides ranked by their combined score were mapped to UniProt proteins, and specificity was calculated as the proportion of assigned kinases. The top 150 peptides yielded 99% specificity for HOLSER and 97% for PELSA, the top 100 peptides yielded 99% and 98%, respectively, while for the top 50 peptides, specificity reached 100% for HOLSER and 98% for PELSA (Supplementary Figure 1C). In the second approach, the statistical significance of the peptides was determined using the Benjamini–Hochberg (BH) correction for multiple hypotheses. Among all peptides with adjusted p-values <0.05, 96% were from kinases for HOLSER versus 90% for PELSA. Together, these results demonstrated consistently high kinase specificity for both HOLSER and PELSA overall, with more targets identified by HOLSER due to its deeper analysis. An exploration of the ability to identify the site of ligand binding was performed on selected targets with high sequence coverage. In CSK4, a well-characterized kinase target of staurosporine, HOLSER identified 27 peptides, while PELSA identified 14 peptides. Of the HOLSER peptides, 16 showed significant changes, with 15 (94%) peptides mapping to the kinase domain, compared to 9 (64%) in PELSA. We used the known X-ray structure of the CSK–staurosporine complex (PDB ID: 1BYG) to correlate the peptide scores with proximity to the binding site. Both in PELSA and in HOLSER, all higher-scoring peptides come from the kinase domain (Figure B), but only in HOLSER were the highest-scoring peptides close to the known staurosporine binding sites (Figure B). One of the topmost peptides, GGLYIVTEYMAK, includes the T266, E267, Y268, and M269 residues located 3.4 Å, 2.9 Å, 3.5 Å, and 2.7 Å away from the ligand, respectively (Figure C, Supplementary Table 4). Notably, M269 is the closest residue to the ligand among all annotated binding sites (Supplementary Table 5). One example where HOLSER’s deeper analysis provided more information on a staurosporine target is CDK2. , HOLSER identified 23 peptides, compared to 13 peptides by PELSA. Of the HOLSER peptides, 18 showed significant changes, with all of them mapping to the kinase domain, whereas PELSA identified 12 peptides in the domain (Supplementary Figure 2A). Furthermore, significant peptides from HOLSER covered 14 of 20 residues known to belong to the binding site, while PELSA-only covered 7 such residues (Supplementary Table 6). Notably, the HOLSER peptide LYLVFEFLHQDLKK includes residues F80–Q85 and D86 positioned 2.6–3.9 Å from the ligand (Supplementary Figure 2B, Supplementary Table 5), L83 being the closest, while PELSA did not detect a peptide with these residues. Similarly, the most significantly protected CHEK1 peptides in HOLSER correctly localized the protein kinase domain as the preferred staurosporine binding site. HOLSER identified 10 peptides with significant changes, compared to 4 peptides detected by PELSA. Of the HOLSER peptides, 8 (80%) mapped to the kinase domain, whereas 3 peptides (75%) identified by PELSA were located within this domain (Supplementary Figure 2C). Notably, in HOLSER, the topmost peptide DIKPENLLLDERDNLK includes residues L134, I135, and T137, which are positioned 2.9, 3.7, and 3.3 Å from the ligand, respectively (Supplementary Figure 2D, Supplementary Table 7). HOLSER Profiling of FKBP–mTOR Complex Response to Rapamycin Rapamycin targets FKBP proteins and mTOR, forming a ternary complex critical for mTOR pathway regulation. , At the peptide level, 77 detected peptides belonging to the FKBP family showed statistically significant shifts in HOLSER, compared to 52 peptides in PELSA (Supporting Information Table 1D). Among the proteins, FKBP family members were clearly identified as the most likely target candidates (Figure A and Supplementary Table 8), with FKBP3 being the top candidate. FKBP response to rapamycin treatment. (A) Volcano plot displaying protein-level abundance changes in rapamycin-treated samples compared with DMSO controls (n = 3 biological replicates). (B) Mapping of FKBP3-derived peptides identified in HOLSER and PELSA onto the protein sequence. Dashed lines indicate the known binding sites from PDB. (C) Structural visualization of the FKBP3–substrate complex based on the crystal structure (PDB ID: 1PBK), rendered in PyMOL. Sequence mapping of FKBP3-derived HOLSER peptides provided broader coverage across the protein sequence (42 peptides and 81.7% sequence coverage versus 20 and 64.3%, respectively, in PELSA) and identified multiple additional peptides from the binding region (Figure B). Notably, the most-protected in HOLSER peptide LEIEPEWAYGKK includes the residue Y198 located just 1.76 Å away from rapamycin in the FKBP3–rapamycin complex structure (PDB ID: 1PBK) (Figure C, Supplementary Table 9). In PELSA, the peptide containing this residue has only a modest significance, with the most significant peptide from the binding region surprisingly showing a reversed protection sign. HOLSER Global Proteome Profiling As already mentioned, the ratio between the peptide abundance in the partial digest of the DMSO-treated channel and full digest provides the accessibility factor related to protein tertiary structure. As an example, Figure A shows the HOLSER accessibility factors of the KPNA2 protein and compares them with the protection factors from HDX-MS (Supplementary Table 10). A significant positive correlation (R = 0.43, p = 0.003) was observed between these two parameters. The FKBP3 protein exhibited an even better agreement (Figure B, Supplementary Table 11), with a comparably high positive correlation (R = 0.49, p = 0.011). The correlation between HOLSER and HDX-MS could likely be even better if not for the significant differences in their experimental frameworks, with HOLSER probing accessibility to proteases of peptide bonds in complex cell lysates and HDX-MSsolvent accessibility of backbone amides in purified recombinant proteins. Comparisons of the protein structures with HDX-MS protection factors against the HOLSER accessibility values (Figure C,D) confirm the similarities, while the differences can at least partially be explained by the absence of heterogeneous protein–protein interactions and post-translational modifications in HDX-MS that is performed on isolated, recombinant proteins as opposed to native lysate in HOLSER. Note that a single HOLSER analysis potentially provides 105 independent pieces of structural information on several thousands of proteins, making it one of the most informative structural approaches. Analysis of accessibility. Residue-level accessibility comparison between HDX and HOLSER conditions for KPNA2 (A) and FKBP3 (B). Per-residue values were smoothed using a rolling window of 20 residues; each dot represents the average within that window. Smooth curves were fitted using a generalized additive model (GAM) with thin plate regression splines to highlight overall trends across the protein sequence. Protein structure (AF–P52292-F1-model_v4) for KPNA2 (C) and 1PBK for FKBP3 (D), colored by HDX-MS (left) and HOLSER (right) z-scores in PyMOL. Blue means low accessibility, red means high accessibility, and light-yellow marks residues without data. Conclusions We developed HOLSER, a partial proteolysis-based proteomics workflow that enables enhanced depth of proteome analysis, sequence coverage, improved quantification in detecting ligand-induced protein accessibility changes, and provides global profiling of protein quaternary structures. The technique combines short digestion under native conditions with a high enzyme-to-substrate ratio, TMT-based multiplexing, deep fractionation, and the inclusion of full-digest carrier channels. Compared to PELSA and its high-throughput version (HT-PELSA), HOLSER detects a greater number of peptides and proteins, yields higher sequence coverage, and offers higher certainty in determining binding sites. Thus, HOLSER provides a robust and sensitive platform for profiling protein–ligand interactions under native conditions, with broad potential applications in target engagement studies, drug mechanism-of-action analysis, and proteome-wide conformational profiling. Supplementary Material