Authors
Jonathan Palmer1,2; Victor Adebomi1,2; Jimmy Eng3 Patrick Salveson2; Ta-Yi Yu2; Stephen Rettie1,2; Gaurav Bhardwaj1,2; Miklos Guttman1

Institutions
1 University of Washington, Department of Medicinal Chemistry, 2University of Washington, Institute of Protein Design, 3University of Washington, Proteomics Resource Center

Abstract
High throughput screening (HTS) is an incredibly powerful tool for drug discovery and development. HTS has enabled researchers to test millions of linear peptides, but applications for cyclized peptides lag behind. Cyclized peptides can offer greater chemical control, metabolic stability, and passive permeability than linear peptides, but have added complexity when sequencing via mass spectrometry. Notably, this increased complexity prevents regular peptide sequencing tools from accurately identifying cyclic peptides. To address this gap, we utilized a novel combination of LC-MSn with an expanded Comet sequencing program to demonstrate that we can use HTS to identify macrocyclic peptides with binding affinity for protein targets.

Authors
James M. Fulcher,1,* Lye Meng Markillie,1 Hugh D. Mitchell, 1 Sarah M. Williams,1 Kristin M. Engbrecht,3 Ronald J. Moore,2 William B. Chrisler,2 Joshua Cantlon-Bruce,4,5 Johannes W. Bagnoli,5 Anjali Seth,5 Ljiljana Paša-Tolić,1 Ying Zhu1,6 *

Institutions
1Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, Washington 99354, United States 2Biological Sciences Division, Pacific Northwest National Laboratory, Richland, Washington 99354, United States 3Nuclear, Chemistry, and Biology Division, Pacific Northwest National Laboratory, Richland, WA 99354, United States 4Scienion AG, Volmerstraße 7, 12489 Berlin, Germany 5Cellenion SASU, 60 Avenue Rockefeller, Bâtiment BioSerra2, 69008 Lyon, France 6Present address: Department of Microchemistry, Lipidomics and Next Generation Sequencing, Genentech, 1 DNA Way, South San Francisco, 94080, United States *Correspondence: Dr. Ying Zhu, zhu.ying@gene.com and Dr. James M. Fulcher james.fulcher@pnnl.gov

Abstract
Single-cell multiomics provides comprehensive insights into gene regulatory networks, cellular diversity, and temporal dynamics. While tools for co-profiling single-cell genomes, transcriptomes, or epigenomes are available, accessing proteomes in parallel is more challenging. We developed nanoSPLITS (nanodroplet SPlitting for Linked-multimodal Investigations of Trace Samples), an integrated platform that enables global profiling of the transcriptome and proteome from same single-cells using RNA sequencing and mass spectrometry-based proteomics, respectively. nanoSPLITS can precisely quantify over 5000 genes, 2000 proteins, and 140 phosphopeptides per single cell and identify cell-type-specific markers from these modalities. In the context of Cdk1-mediated cell cycle arrest, we demonstrate how nanoSPLITS-based single-cell multiomics can provide comprehensive molecular characterization with insights including covarying protein/gene clusters, unique phosphorylation events, and mitotic pathways.

Authors
Alex Zelter, Michael Riffle, Michael R. Hoopmann, Ellen Riddle, David Shteynberg, Daniel Jaschob, Robert L. Moritz, Michael J. MacCoss, Nina Isoherranen

Institutions
Alex Zelter, Michael Riffle, Daniel Jaschob, Michael J. MacCoss = Department of Genome Sciences, University of Washington; Ellen Riddle and Nina Isoherranen = Department of Pharmaceutics, University of Washington; Michael R. Hoopmann, David Shteynberg, Robert L. Moritz = Institute for Systems Biology, Seattle, WA.

Abstract
Drugs are often metabolized by cytochrome P450 (CYP) enzymes to reactive intermediates that form protein adducts. Adducts can inhibit protein activity, elicit immune responses, and cause life-threatening adverse drug reactions. The modification of cytochrome P450 3A4 (CYP3A4) active site cysteine Cys239 by raloxifene was previously proposed to be responsible for time-dependent inactivation (TDI) of CYP3A4. Previous work detected raloxifene adducts only at Cys239 and thus there were no other candidate sites for inhibition of CYP3A4. In the current work we demonstrate that raloxifene modifies multiple sites on CYP3A4 along with cytochrome P450 reductase and cytochrome b5, proteins also important for CYP3A4 activity. We show that raloxifene modifies multiple sites on cytochrome P450 2C9 (CYP2C9). In contrast to CYP3A4, raloxifene does not inactivate CYP2C9. We therefore hypothesize that determination of both the location and abundance of raloxifene adducts is important in understanding the effect of drug-protein adducts on enzyme activity. Here we present methods to discover and quantify drug-protein adducts by bottom-up liquid chromatography/mass spectrometry and discuss the challenges associated with these methods, which are necessarily applied to peptides with rare modifications. Low abundance modifications present challenges at all stages of the methodological pipeline that are not as impactful in the study of higher abundance targets. Such challenges are present in sample workup; discovery of modifications by database searching and assignment of error rates by post processing tools; and the quantification of drug modified peptides. To study the effect of drug-protein adducts on enzyme activity these challenges will have to be overcome.

Authors
Chris Hsu, Lilian R Heil, Philip Remes, Jesse Canterbury, Ping Yip, Christine Wu, Mariya T. Sweetwyne, Michael J. MacCoss

Institutions
University of Washington, Thermo Fisher Scientific (San Jose)

Abstract
The single-cell proteomics field aims to provide insights into complex biological changes, such as spatial or cellular heterogeneity, that would otherwise be lost in aggregated, bulk sample experiments. Quantification at the single-cell level or very low amounts of input material remains a challenge because protein loss from surface adsorption needs to be minimized during sample preparation and instrument sensitivity needs to be increased for peptide quantification. We aim to improve sample recovery by including a carrier proteome prior to adding in the low input sample, and we aim to increase instrument sensitivity by acquiring MS/MS data using a massively parallel targeted method coupled with a linear ion trap as a detector. With our method, we can quantify 2,200 peptides (or 800 proteins) from individually prepped 10ng samples with great precision and reproducibility. Our data highlights the quantitative performance of this workflow by demonstrating both technical reproducibility and linearity for low input samples.

Authors
Josh Eckels1; Sweta Jewargikar2; Wendy Innis2; Ankur Juneja1; Nicholas Shulman3; Vagisha Sharma3; Michael J. MacCoss3; Brendan X. MacLean3; and the Panorama Partners Program

Institutions
1LabKey, San Diego, CA; 2LabKey, Seattle, WA; 3University of Washington, Seattle, WA;

Abstract
Panorama is a web-based data management system for targeted mass spectrometry. It integrates closely with Skyline, a free, open source targeted mass spec analysis tool. Panorama’s system suitability tools automatically report on instrument health and identify problems. Features for multi-attribute method (MAM) analysis make characterizing antibodies across samples easy. For researchers doing proteomics, it also visualizes peptide and protein sequence data. Free cloud-based projects are available to everyone on PanoramaWeb.org. Panorama uses a relational database to store targeted mass spectrometry data, importing data from Skyline’s XML file. Tight integration between Panorama and Skyline ensures complementary and compatible feature sets. Panorama’s QC folders offer user-friendly, web-based visualizations and reporting of system suitability data, storing data from each mass spectrometer in a separate folder. AutoQC, a utility that monitors for newly acquired raw data files, automates the otherwise tedious process of adding new data. Panorama also offers a growing assortment of reporting and visualization tools, both available out-of-the-box, and for customization. Since Panorama first implemented system suitability monitoring in 2014, users have created more than 1,500 QC folders on PanoramaWeb.org. Of those, more than 300 folders have accumulated 100 or more samples worth of data and many have thousands of acquired files. Because statistical process controls are not necessarily intuitive to mass spectrometrists, Panorama has recently added new visualizations that show trailing mean and coefficient of variation values for its full set of metrics which may feel more intuitive to everyday users. They join the Levey-Jennings, moving range, and cumulative sum plots. Other improvements to all of Panorama’s system suitability plots help reduce visual clutter from a great deal of historical data. Users doing Multi-Attribute Method (MAM) analyses can utilize additional built-in reporting to monitor post-translational modifications and assess risk based on criteria such as complementarity-determining regions (CDRs) and sample metadata. Additionally, Panorama now supports visualization of cross-linked peptides to easily monitor peptides with disulfide bonds and other more complex PTMs. Finally, Panorama benefits from Skyline’s recent addition of tracking protein groups, capturing the full list of proteins to which a set of peptides may belong. Panorama’s protein/peptide sequence view now also displays intensity and confidence score data, either summarized across multiple samples or one sample at a time. As of February 2023, more than 650 labs are using Panorama projects, free of charge, to manage targeted mass spectrometry assays on http://panoramaweb.org, a server hosted by the MacCoss lab at the University of Washington. Additionally, major pharmaceutical companies and other organizations have deployed their own in-house installations of Panorama.

Authors
Yuan Feng, Benjamin P. Zercher, Matthew F. Bush*

Institutions
University of Washington, Department of Chemistry, Box 351700, Seattle, WA, 98115

Abstract
Introduction Analogous to the advantages of tandem mass spectrometry (MS) relative to single-stage MS, multidimensional ion mobility (IM) can increase the information content of IM experiments by isolating subpopulations of ions for further analysis. Electrostatic ion mobility (IM) enables the direct determination of mobilities without calibration, but using traditional implementations, increasing the number of dimensions also requires corresponding increases in applied potentials. One solution to this challenge is traveling-wave IM, which has been used to enable long separation pathlengths and multidimensional separations with relatively low potentials relative to electrostatic IM. Here, we present an alternative strategy that reduces the magnitude of potentials required for electrostatic IM by dynamically resetting the potential of ions between dimensions of IM.
Methods Ubiquitin, concanavalin A, and cytochrome c were purchased and prepared in aqueous 200 mM ammonium acetate adjusted to pH 7.0. Denatured cytochrome c samples were prepared by solvent exchanging proteins using size-exclusion spin columns that had been equilibrated with three washes of 0.1% acetic acid. These experiments were performed on a Structures for Lossless Ion Manipulations (SLIM) architecture IM instrument that uses electrostatic drift fields to separate ions. Ion trapping is enabled by biasing adjacent modules to create a potential well that we call a “junction trap”. The potentials that comprise the junction trap are modulated by applying time-dependent voltage programs. Preliminary In tandem IM experiments, ion packets were generated on the first module (1M) and separated in the first dimension of IM (1D) before the time-dependent selection of a specific charge state at 7M. Selected ions were accumulated in a junction trap for a short time (~10 ms) prior to the 2D. In potential resetting experiments, the 2D region was held at ground while ions were separated and selected in the 1D region. The selected ions were accumulated in a junction trap between the 1D and 2D at near ground potential. After accumulation, the potentials of the junction trap and the input of the 2D were raised in 5 V increments every 125 µs. The ions were then released from the junction trap into the 2D. For comparison, ions were also analyzed in tandem IM experiments that used the same drift fields, but applied using higher, static drift potentials. Ion lifetimes, as monitored by the depletion of the precursor, were the same in both experiments. The well-overlaid 2D arrival time distributions and mass spectra from each experiment suggest that potential resetting does not perturb trapped ions. We then isolated three subpopulations of 7+ cytochrome c, reset their potentials, and demonstrated that all subpopulations remain resolved in the 2D, which suggests that potential resetting of ions does not induce conformational changes that significantly change the arrival time of ions.

Authors
Mukul K. Midha, Charu Kapil, Michal Maes, David H. Baxter, Seamus R. Morrone, Timm J. Prokop, Robert L. Moritz

Institutions
Institute for Systems Biology, Seattle, WA 98109

Abstract
Introduction Nano-flow chromatography-tandem mass spectrometer (nLC-MS/MS) is a popular choice in proteome research as it provided high-sensitivity analysis with minimal sample usage, but its quantitative proteomics applications become restricted by low analytical throughput, and low robustness. To circumvent this limitation, we describe the performance of the new Vacuum Insulated Probe Heated ElectroSpray Ionization source (VIP-HESI) coupled to the Bruker timsTOF mass spectrometer and show it enhances micro-spray flow rate chromatography signals in comparison to nanospray source conditions using the CaptiveSpray (CS) and ElectroSpray Ionization (ESI) sources. In addition, using a 0.5 x 200mm and 1 × 150 mm columns, achieved excellent reproducibility of chromatographic retention time, coefficient of variation, and precision quantification using data from un-depleted mouse plasma, HeLa and K562 digest peptide samples.
Methods PepCalMix peptides with seven different concentration points to cover a range of 12.5fmol to 800fmol were measured to assess linear response using VIP-HESI source and compared to CaptiveSpray and ElectroSpray Ionization sources. Using un-depleted mouse plasma sample, different injection amounts of 0.4μg at 1μL/min with CaptiveSpray were compared to 2μg, 4μg, 10μg, 20μg, and 40μg injection amounts at 40μL/min coupled with VIP-HESI source. Lastly, we also assess the performance of VIP-HESI microflow setup in a slice-PASEF mode for analysis of low sample amounts. For dia-PASEF runs and slice-pasef, Spectronaut 16.0 and DIA-NN 1.8.2 tools respectively were used to identifications filtered with a q-value of <0.01 and MS2 ion peak areas of quantified peptides were summed to estimate the protein quantification.
Preliminary Data We determined excellent linear gain in the total protein quantity as we increased the PepCalMix concentration with the R-Squared (R²) of more than 99% and linear response enabled excellent comparison between the CS, ESI, and VIP-HESI sources. VIP-HESI source performed better than than ESI by providing 50% more PepCalMix quantities and 30% more abundances than CS after injecting thirty-two times more injection amounts. Using mouse undepleted plasma samples, DIA-MS analysis using Spectronaut software resulted in the cumulative identification of 5,795 - 8,610 precursors and 493 - 802 unique proteins at <1% protein FDR across the different injection amounts. Interestingly, injecting fifty times (20μg) more sample amounts (with 40 times more sample dilution) using VIP-HESI, outperforms CS 0.4μg measurement by identifying 3% more precursors and 12% more proteins. Next, on evaluating the reproducibility of quantitative measurements, VIP-HESI 20μg measurement, a total of 7,074 precursors were quantified by both biological replicates with a high positive correlation (r2 = 0.94). Whereas with CS 0.4μg replicate measurements, 6,387 precursors were quantified with a positive correlation (r2 = 0.89), indicating more variation was observed with nanoflow rates, compared to VIP-HESI-microflow runs. Subsequently, we observed similar median number of data points (5) measured across the elution profiles and similar base peak widths (0.16 and 0.14 minutes) in both VIP-HESI 20μg and CS 0.4μg measurements, signifying that VIP-HESI provides an optimal quantification of an identified precursor using spectronaut analysis. For slice-pasef experiment, DIA-NN tool with different HeLa sample amounts resulted in the identification of 6,592 - 49,130 precursors at <1% precursor FDR and 1,769 - 6,343 proteins at <1% protein FDR. For 10ng sample injection amounts, slice-PASEF yielded 63% more precursors and 38% more protein groups than the VIP-HESI source setup, demonstrating significant increase in the number of precursor and protein groups, especially with low injection amount measurements.
Novel Aspect VIP-HESI source with improved analytical setup capable of highly reproducible chromatography that can accurately quantify with a wider proteome coverage.

Authors
Sophie Moggridge, Ricard Rodriguez-Mias, Kyle Hess, Judit Villén

Institutions
Center for the Multiplexed Assessment of Phenotype, Department of Genome Sciences, University of Washington, Seattle, USA

Abstract
Protein variants arising from single amino acid substitutions contribute to many human diseases. These substitutions can affect protein properties such as structure, stability, activity, subcellular localization, and interactions with other biomolecules. Previous work has focused on characterizing the effects of substitutions using classic biochemical assays on individual protein variants. Here we streamline this process by coupling biochemical assays to mass spectrometry detection, enabling the characterization of many protein variants in a single experiment. We selected 13 previously characterized missense variants of PGM1, conducted protein solubility and thermal stability assays in a pooled format and compared our results to previously published data using purified variants. Our pooled mass spectrometry-based assay successfully recapitulated the results from individually assayed variants and achieved improved resolution and variant library coverage. Scaling these assays to thousands of variants will greatly enhance our understanding of how single amino acid substitutions affect protein function and will contribute to better classifying and understanding variants.

Authors
Matthew D. Berg(1), Ricard A. Rodriguez-Mias(1) and Judit Villén(1)

Institutions
(1) Department of Genome Sciences, University of Washington, Seattle, WA, USA

Abstract
While DNA sequencing has identified millions of natural genetic variants that alter protein sequence, determining the functional impact of these variants remains challenging. Traditional mutagenic approaches are not scalable for millions of variants and high-throughput approaches such as deep mutational scanning are limited to investigating one protein per experiment. Recently, our lab established MiRo – a high-throughput proteomic technology to enable functional annotation of thousands of missense mutations across entire proteomes. In this approach, stochastic errors in protein synthesis are induced to create variants across all expressed proteins within a cell. Variants are subject to biochemical selections that probe general protein properties like solubility, thermal stability, ligand binding, protein-protein interactions and posttranslational modifications. After selection, variants are quantified by mass spectrometry to determine the functional impact of each mutation on the measured property. Previously, non-canonical amino acids that are mis-incorporated in place of one of the 20 canonical amino acids were used to create protein variants. In this work, we expand the MiRo approach to determine the impact of canonical amino acid substitutions by engineering alanine tRNAs that mis-incorporate alanine at non-alanine positions. By mutating the anticodon of tRNAAla and expressing these engineered tRNAs in yeast, we achieve alanine mis-incorporation at nearly all non-alanine codons. We identify hundreds to thousands of mis-incorporation events with each tRNA variant using mass spectrometry. We then apply thermal proteome profiling to the alanine-substituted proteomes to determine the impact of individual substitutions on protein thermal stability and identify functionally important residues and protein regions. This work represents the first proteome-wide alanine scan and provides insight into various aspects of protein biology including the structural and functional context underlying mutational sensitivity.

Authors
Ellen Riddle, Alex Zelter, and Nina Isoherranen

Institutions
University of Washington, Department of Pharmaceutics (Ellen Riddle and Dr. Nina Isoherranen); University of Washington, Department of Biochemistry (Dr. Alex Zelter)

Abstract
A major concern for drug-drug interactions (DDI) includes the irreversible inhibition of cytochrome P450 enzymes (CYPs). Time dependent inhibition (TDI) may occur via adduction of CYP enzymes by reactive drug metabolites. Traditional methods to detect and identify reactive metabolites during drug development include glutathione trapping, radiolabeled compounds, or metabolite identification; none of which yield information regarding the precise proteins and amino acids that are adducted and thus cannot be used to study the impact of adduct location on TDI. We hypothesize that TDI via adduct formation is a function of the specific placement of the adduct and the adduct formation kinetics. Therefore, the goal of this project is to establish a relationship between protein inactivation and adduct formation kinetics. We tested our hypothesis using a combination of in vitro incubations and proteomic methods with CYP2C9 as a model enzyme, and two model compounds: tienilic acid, a known CYP2C9 TDI, and raloxifene, which does not inhibit CYP2C9 in a time-dependent manner. Both raloxifene and tienilic acid are known substrates of CYP2C9 that form reactive metabolites. In vitro IC50 shift experiments confirmed that tienilic acid, but not raloxifene, inhibits CYP2C9 in a time-dependent manner. Tienilic acid and raloxifene treated CYP2C9 reconstituted with P450 reductase were then analyzed using high-resolution mass spectrometry, and subsequently searched using Magnum, an open mass searching algorithm, which identified the adduct masses and the specific adducted peptides of tienilic acid and raloxifene treated CYP2C9. Two treatment specific significant modifications were identified on cysteines in CYP2C9 following CYP2C9 incubation with tienilic acid; R.YIDLLPTSLPHAVTCDIK.F was found to be modified by +329.96 Da, and R.CLVEELR.K was found to be modified by +349.94 Da. The MS1 scan of the +329.96 Da modified R.YIDLLPTSLPHAVTCDIK.F shows a clear chlorine isotope pattern from the tienilic acid derived adduct, and this is reflected in several y-ions in the MS2. Additionally, +329.96 Da adduct on R.YIDLLPTSLPHAVTCDIK.F is present in samples digested with and without reduction and alkylation, whereas the +349.94 Da adduct on R.CLVEELR.K is only present in samples digested without reduction and alkylation. Furthermore, a shift in retention time is present for all adducted peptides compared to the non-adducted counterpart. These adducts were then quantified by DIA analysis in Skyline. For raloxifene treated CYP2C9 purified protein, +471.15 Da was identified as a treatment specific significant modification and the +471.15 Da modification was found on multiple cysteine and tyrosine residues in CYP2C9, despite the lack of TDI. These findings suggest that although adducts of the inhibitors tested are found on CYP2C9, inactivation is dependent on both the localization and abundance of adducts present. Overall, a definitive characterization of the specific adducts formed is necessary to understand the functional consequences of protein adduction.

Authors
Nicholas Day, Xiaolu Li, Matthew Gaffrey, James Sanford, Tyler Sagendorf, Kwame Attah, Sue Bodine, Josh Adkins, Wei-Jun Qian, and the MoTrPAC consortium.

Institutions
Pacific Northwest National Laboratory, University of Iowa

Abstract
The Molecular Transducers of Physical Activity consortium (MoTrPAC) aims to identify the underlying molecular changes that occur in response to physical activity to understand how exercise improves health. Among these molecular changes are redox-based post-translational modification of protein Cysteine thiol groups by molecules such as reactive oxygen species (ROS). Quantitative measurement of reversible protein thiol oxidation enables a better understanding of the redox environment and helps to identify what protein thiols are susceptible to oxidative modifications. Herein, we investigate the temporal responses of thiol redox proteome in rat muscle tissue derived from an endurance exercise training study. 60 lateral gastrocnemius muscle tissue samples were collected from male or female rats assigned to sedentary (Control) or exercise (1, 2, 4, or 8 weeks) groups. The samples were processed using an integrated bottom-up global and ‘total oxidation’ redox proteomics workflow that incorporated multiplexed isobaric labeling and enrichment using a thiol affinity resin. Due to the presence of highly abundant proteins in muscle tissue, we implemented our recently developed deep redox profiling workflow to improve coverage that surpasses other recent studies. The peptides were analyzed using a Thermo Scientific Q Exactive HFX mass spectrometer and identified using MSGF Plus. Statistical analysis was performed using the PlexedPiper pipeline R package. Thiol oxidation of 17,201 Cysteine (Cys) sites were quantified, corresponding to 5242 proteins. Compared to the control group, significant changes in thiol oxidation (redox) are observed across all exercise groups for both sexes. Female rats show more dynamic changes in Cys oxidation compared to male rats during the first 4 weeks of exercise. However, by the 8th week of exercise, male rats show a marked increase in the number of significantly oxidized sites compared to male rats at earlier timepoints and female rats at all timepoints. Significantly changing Cys oxidation in male rats is predominantly associated with bioenergetic pathways. On the other hand, significantly changing Cys sites in female rats are involved in processes that control muscle function and quality. This distinction suggests that exercise training may have sex-specific effects on the physiology of skeletal muscle. Ongoing work will integrate redox changes with those observed with phosphorylation or metabolites, where data are already available from MoTrPAC.

Authors
Alex Zelter, King CB Yabut, Benjamin Zercher, Alice Martynova Michael J MacCoss, Matthew F. Bush, Nina Isoherranen

Institutions
Departments of Genome Sciences, Pharmaceutics and Chemistry, University of Washington

Abstract
4-Hydroxynonenal (4HNE) is a well-known lipid peroxidation product that modifies proteins via Michael addition (cysteine, histidine and lysine) and a Schiff base (lysines) mechanism. These covalent modifications have been proposed to contribute to the initiation and progression of multiple diseases associated with oxidative stress. However, the modifications formed by 4HNE are typically low abundance and difficult to identify and quantify from complex samples. The stoichiometry and kinetics of specific modifications formed by 4HNE are poorly understood and the quantitative relationships between modification and decreased protein function have not been defined. We used a combination of mass spectrometry techniques to determine the relationship between 4HNE mediated modifications in cellular retinoic acid binding protein 1 (CRABP1) and its function.
CRABP1 was expressed and purified from E. coli and treated with increasing concentrations of 4HNE to generate 4HNE-modified protein. 4HNE-modified CRABP1 was analyzed via intact protein mass spectrometry to determine the number of 4HNE modifications formed per CRABP1 protein and to estimate the fraction of the total CRABP1 modified by 4HNE with increasing concentrations of 4HNE. Treated and untreated CRABP1 protein was digested with trypsin and the 4HNE modified residues and the specific modification masses determined via shotgun proteomics. The relative abundance of the modifications at specific residues was quantified using DIA. Intact protein MS showed that following a one-hour incubation of CRABP1 with 10- or 25-fold molar excess of 4HNE, only 15-20% of CRABP1 was modified by 4HNE.
The major CRABP1-4HNE product contained just a single modification of 156 Da. This mass corresponds to the Michael addition of 4HNE to cysteine, histidine or lysine. To determine the specific locations of 4HNE modifications within the protein we performed tryptic digestion and bottom-up proteomics. 4HNE modifications were found in six of the 12 modifiable amino acids in CRABP1. Both histidines (H40 and H94) and two of the three cysteines (C95 and C129) were found modified by 4HNE. In addition, the peptides containing C95 and C129 were completely alkylated in untreated samples, while the presence of peptides without alkylation of C95 and C129 increased considerably when samples were treated with 4HNE. Of the seven lysines, only K38 and K92 were modified, consistent with the predicted lowest reactivity of lysine with 4HNE. Both the intact protein and bottom-up MS analysis supported 156 Da being the predominant modification by 4HNE. The only Schiff base modification (138 Da) found was at K92 but even in this residue the 156 Da modification was more abundant. This modification also resulted in a missed cleavage for that tryptic site. No significant decrease in the abundance of any of the unmodified peptides was observed, which highlights the limitations of using the depletion of unmodified peptides as a proxy for quantifying protein modifications. Taken together these data show that the relatively low abundance of 4HNE modified peptides in shotgun proteomics is due to incomplete modifications of the specific residues in the target protein and the distribution of modifications amongst different modifiable residues.

Authors
Alexis Chang, Mario Leutert, Ricard A. Rodriguez-Mias, Judit Villen

Institutions
Department of Genome Sciences, University of Washington, Seattle, Washington 98195, United States

Abstract
Phosphotyrosine (pY) enrichment is critical for expanding the fundamental and clinical understanding of cellular signaling by mass spectrometry-based proteomics. However, current pY enrichment methods exhibit a high cost per sample and limited reproducibility due to expensive affinity reagents and manual processing. We present rapid-robotic phosphotyrosine proteomics (R2-pY), which uses a magnetic particle processor and pY superbinders or antibodies. R2-pY can handle up to 96 samples in parallel, requires 2 days to go from cell lysate to mass spectrometry injections, and results in global proteomic, phosphoproteomic, and tyrosine-specific phosphoproteomic samples. We benchmark the method on HeLa cells stimulated with pervanadate and serum and report over 4000 unique pY sites from 1 mg of peptide input, strong reproducibility between replicates, and phosphopeptide enrichment efficiencies above 99%. R2-pY extends our previously reported R2-P2 proteomic and global phosphoproteomic sample preparation framework, opening the door to large-scale studies of pY signaling in concert with global proteome and phosphoproteome profiling.

Authors
Aprajita Yadav1, Alex Zelter2, Nina Isoherranen1

Institutions
1. Department of Pharmaceutics 2. Department of Genome Sciences, University of Washington, Seattle, WA

Abstract
Retinol binding protein 4 (RBP4) is a 21 kDa protein that forms a tetramer with transthyretin (TTR), a 55 kDa homotetrameric protein, and binds retinol in blood with high affinity. Multiple studies in humans have shown an association between increased serum levels of retinol binding protein (RBP4) and obesity, diabetes, insulin resistance, triglyceride levels, cardiovascular disease, and chronic kidney disease1. However, RBP4 and TTR concentrations have typically been measured by western blots, or using enzyme linked immunoassays (ELISA) kits, which have limited dynamic range, can be variable between labs or kits, and response can be altered by disease states, matrix effects or anticoagulant2,3.
LC-MS/MS based protein quantification has been developed as an alternative strategy for ELISAs and western blotting, but plasma proteomic analysis can be confounded by the complexity of the human plasma proteome, abundance of albumin and IgG, and the presence of protease inhibitors which can affect tryptic digest.
We hypothesized that LC-MS/MS quantification of RBP4 and TTR in plasma would provide an accurate, rigorous, and reproducible method to determine the absolute concentrations of RBP4 and TTR in human plasma/serum in patients with different disease states and allow quantification of RBP4-retinol and RBP4-TTR concentration ratios. A novel, quantitative mass spectrometry based analytical method was developed and validated to allow for simultaneous targeted quantification of RBP4 and TTR in human plasma.
Surrogate peptides for proteomic analysis of RBP4 and TTR were selected from in silico analysis and monitored following tryptic digest of purified protein standards and plasma samples. To address the phenomena of digestion quality, missed cleavages, ragged ends and variable modifications present, shotgun proteomics was employed to characterize purified RBP4 standards expressed from multiple cell systems and RBP4 and TTR in human plasma. This analysis showed the presence of peptides incorporating missed cleavages in addition to the expected fully cleaved peptides. Division of signal between multiple peptide species is an important consideration when developing quantitative assays. Yeast enolase was added as a process control and the effectiveness of reducing signal variability by normalizing to this process control was assessed. Multiple internal standards, FSGTWYAMAK[13C615N2] and YWGVASF[13C915N]LQK for RBP4 and GSPAINVAVHVFR[13C615N4] for TTR, were used to determine quantitative variability across peptides from the same protein. A set of plasma samples was quantified multiple times to address technical digest variability, inter-day digest variability, freeze-thaw stability, and autosampler stability. Instrument response linearity and matrix effects for different peptides was also assessed. Instrument variability was observed to range from 5.4 - 7.3% for different enolase peptides, digestion variability added 8 – 32.5% variation which was peptide dependent.
Broadly, these results highlight the benefits of robust method development to account for incomplete digestion, efficiency of digestion with a process control and selecting appropriate normalization criteria for accurate quantification. The developed assay will provide an important new tool for understanding the validity of RBP4 and TTR as disease biomarkers.
1. Blaner, W. S. Vitamin A signaling and homeostasis in obesity, diabetes, and metabolic disorders. Pharmacology and Therapeutics vol. 197 153–178 (2019). 2. Hoofnagle, A. N. & Wener, M. H. The fundamental flaws of immunoassays and potential solutions using tandem mass spectrometry. Journal of Immunological Methods vol. 347 3–11 (2009). 3. Graham, T. E., Wason, C. J., Blüher, M. & Kahn, B. B. Shortcomings in methodology complicate measurements of serum retinol binding protein (RBP4) in insulin-resistant human subjects. Diabetologia 50, 814–823 (2007).

Authors
Theresa A. Gozzo, May A. Constabel, Matthew F. Bush

Institutions
Department of Chemistry, University of Washington, Box 351700, Seattle, WA, 98115

Abstract
Introduction Monoclonal antibodies form the basis of a swiftly growing class of therapeutics. In contrast to traditional small molecule drugs, antibodies are large, flexible proteins generated from a biological source. Their heterogeneity and complexity pose challenges to study via traditional structural biology techniques; therefore, gas-phase approaches have been leveraged to characterize and differentiate them. Here, we present two approaches that can help characterize the structures of antibodies as well as distinguish similar antibodies from each other. First, we use a combination of solution conditions, charge reduction via Cation-to-Anion Proton-Transfer Reactions (CAPTR), and energy-dependent ion mobility-mass spectrometry (IM-MS) experiments to distinguish between IgG1 and IgG4. Second, we introduce online, temperature-controlled disulfide reduction with potential for rapid characterization of IgG disulfide bond connectivity.
Methods IgG1κ and IgG4κ (human myeloma) were purchased from Sigma-Aldrich. For native-like conditions, samples were exchanged into 200 mM ammonium acetate with Micro Bio-Spin 6 columns (Bio-Rad). For denaturing conditions, samples were instead exchanged into 0.1% acetic acid. All experiments were conducted on a Synapt G2 HDMS modified with a glow-discharge ionization source and a radio-frequency confining drift cell. To perform CAPTR, perfluoro-1,3-dimethylcyclohexane (PDCH, Sigma-Aldrich) was introduced as a vapor. The monoanion [PDCH–F]– was reacted with cations of IgG1 and IgG4 in the trap cell. IM arrival-time distributions were measured in helium. For online disulfide reduction, experiments were performed using a home-built programmable, temperature-controlled nanoelectrospray (ptESI) source, and MS data were collected on a Waters Cyclic IM-MS instrument. DTT was added to native-like buffer exchanged IgG samples in various concentrations, and mass spectra were collected continuously during temperature programs consisting of a ramp from 30 °C to a higher temperature (varied) and back to 30 °C.
Preliminary data ΩHe values of native-like IgG1 and IgG4 ions and their charge-reduced CAPTR products depend weakly on charge state, a trend that is consistent with results from CAPTR of other native-like protein ions. Conversely, ions generated from denaturing solution conditions and their charge-reduced products exhibit ΩHe values that depend strongly on charge state. Taken together, these results contribute to the body of work suggesting that protein ions with lower charge densities have Ω values that depend less on charge state than protein ions with higher charge densities. Pre-CAPTR activation was applied to ions from native-like solutions by increasing the sampling cone voltage in the atmospheric-pressure interface. When the Ω distributions were quantitively compared, charge-reduced CAPTR products had more distinguishable apparent Ω distributions than those of the precursor charge states at most pre-CAPTR activation voltages. Relative to collision-induced unfolding (CIU) of the precursor ions, the Ω distributions of charge-reduced products were also more distinguishable.
ptESI enabled the online disulfide reduction of IgGs, as monitored by shifts in m/z. We hypothesize that the pattern in which heavy and light chains become freed from intact IgGs will depend strongly on both disulfide bond connectivity and other aspects of thermal stability in solution, which will facilitate the differentiation of similar antibodies. This hypothesis and the integration of IM data will be the subject of ongoing investigation.

Authors
Vorauer, Clint; Fries, Bettina; Guttman, Miklos

Institutions
University of Washington, Department of Medicinal Chemistry; Stony Brook University

Abstract
Staphylococcal enterotoxin B (SEB) is a small, secreted protein that causes food poisoning-like symptoms and can be lethal at microgram amounts. SEB elicits a profound immune response via the bridging of major histocompatibility complex class II (MHCII) molecules on antigen presenting cells with T-cell receptors (TCR). Several monoclonal antibodies (mAbs) have been identified as capable of neutralizing SEB. Using structural mass spectrometry, we successfully mapped the epitopes of several neutralizing antibodies and determined allosteric effects that contribute to the efficacy of select mAbs. Having established a robust approach for tracking mAb-SEB interactions, we examined the interactions within polyclonal sera for understanding antibody responses to SEB and their resulting protective efficacy.
Hydrogen Deuterium Exchange coupled to Mass Spectrometry (HDX-MS) was used to map the epitopes of known neutralizing antibodies against SEB. To study the response of polyclonal sera against SEB, we developed an HDX-MS method that assays protein interacting with a mixture of different species. Uniquely, the technique assays the same immobilized SEB that is used to bind to species in complex serum, thus interrogating the natural, unpurified, and most biologically relevant solution.
It was determined that that sera capable of neutralizing SEB focused on select epitopes on the structure of SEB. We observed differential uptake of deuterium across the structure of SEB after exposure to neutralizing antibodies relative to a control condition, which indicated regions of preferred binding from the serum condition. Across a panel of sera, we determined that immunofocusing occurred on different epitopes depending on species. This assay presents a new way of examining protein interactions in complex solutions and provides a novel way to test vaccine efficacy.

Authors
Christine C Wu (1,*), Kristine A Tsantilas (1, **), Jea Park (1), Deanna L Plubell (1), Previn Naicker (2), Ireshyn Govender (2), Sindisiwe Buthelezi (2), Stoyan Stoychev (3), Justin Jordaan (3), Gennifer E Merrihew (1), Eric Huang (1), Edward D Parker (4), Michael Riffle (5), Andrew N Hoofnagle (6), Michael J MacCoss (1,*)

Institutions
1 - Department of Genome Sciences, University of Washington, Seattle, WA, USA 2 - CSIR, Pretoria, South Africa 3 - ReSyn Biosciences, Gauteng, South Africa 4 - Vision Core, Department of Ophthalmology, University of Washington, Seattle, WA, USA 5 - Department of Biochemistry, University of Washington, Seattle, WA, USA 6 - Department of Lab Medicine and Pathology, University of Washington, Seattle, WA, USA * Corresponding Authors ** Presenting author

Abstract
Extracellular vesicles (EVs) are membrane-bound particles that carry critical biological cargo throughout the body – including in blood. The particles circulating in plasma including EVs like microvesicles and exosomes along with apoptotic bodies represent ~1-2% of the total protein composition. Analysis of this enriched membrane particle fraction by mass spectrometry is effectively a “liquid biopsy” that significantly improves the dynamic range of the proteins measurable in plasma. We have developed a one-step enrichment strategy (Mag-Net) using strong-anion exchange magnetic microparticles (MagReSyn® SAX, ReSyn Biosciences) to capture membrane-bound particles from plasma on a Kingfisher system (Thermo). After capture, the particles can either be eluted and their composition characterized by electron microscopy or particle counting, or they may be carried through SP3 protein digestion. Peptides are collected and analyzed with a quantitative liquid chromatography mass spectrometry strategy using data independent acquisition. The Mag-Net method is robust, reproducible, hemolysis compatible, inexpensive, requires less than 100 μL plasma input, and translates across species.
In human plasma using an Orbitrap EclipseTM TribridTM (Thermo) coupled to a VanquishTM Neo UHPLC (Thermo), we regularly identify more than 4,000 proteins and 30,000 peptides. We consistently and quantitatively capture known protein markers of exosomes and microvesicles including CD9, CD40, CD61, PDCD6IP, NCAM1, CD147, and flotillin are enriched in Mag-Net preparations relative to unfractionated plasma by >70 fold. Abundant plasma proteins such as albumin, transferrin, alpha-2-macroglobulin, alpha-1-antitrypsin, ApoA1, and ApoB are depleted relative to unfractionated plasma. The size distribution of the membrane particles was 50-500 nm as measured by Nanosight particle tracking, which is within literature expectations for a mixed population of EVs. Lipid bilayers and similar sizing of the membrane particles was observed by TEM.
Despite its many positive attributes, the substantial proteome dynamic range of plasma often necessitates depletion of abundant proteins or affinity reagents to enrich specific analytes of biological interest. However, depletion and affinity reagents are often unavailable for model organisms, which are commonly utilized in geroscience studies to characterize the basic biology of aging or improve our understanding of age-related disease. Plasma EVs were isolated using Mag-Net from three vertebrates of varied genetics, environment, and lifespan: humans, laboratory mice, and companion dogs. We found that Mag-Net captures membrane-bound particles from mouse and dog plasma of similar size distributions to humans (and as expected for exosomes and microvesicles) by Nanosight particle tracking. Samples were also run on an Orbitrap ExplorisTM 480 (Thermo) coupled to an EASY-nLCTM 1200 (Thermo). Using Mag-Net in humans, mice, and dogs, respectively, we found 3445, 2593, and 1845 proteins derived from 15613, 10605, and 5795 unique peptides. This was inclusive of similar exosome and microvesicle proteins (CD9, CD61, CD44, FLOT1, FLOT2 NCAM1, and CD147), neurological/platelet-derived proteins (amyloid beta precursor protein, clusterin), and common membrane proteins (calreticulin, cadherins, integrins, tetraspanins). Abundant plasma proteins that are commonly depleted in commercial columns using affinity reagents (including but not limited to albumin, transferrin, alpha-2-macroglobulin, alpha-1-antitrypsin, APOA1, APOB, C3) were reduced to similar magnitude in mouse and dog plasma.
Single step enrichment of membrane particles using Mag-Net has the potential to drive innovation in basic and translational science by facilitating in-depth quantification of the plasma proteome in humans and model organisms. Future work will use this simple and fast method to focus on interspecies characterization of disease etiology and aging biology.

Authors
Charlie Mundorff1, Michael Watson1, Lauren Carter2, Malika Hale3, Adian Valdez2, Marion Pepper3, Neil King2, Miklos Guttman1

Institutions
1Department of Medicinal Chemistry, University of Washington, Seattle WA 98195 2Department of Biochemistry, University of Washington, Seattle WA 98195 3Department of Immunology, University of Washington, Seattle WA 98195

Abstract
Introduction 118/120: IgM antibodies comprise a subclass of immunoglobulins that are potent activators of immune function. IgMs trigger this response through interactions with the C1 complex which are upstream complement proteins. C1q is a large collagen-like protein responsible for recognizing surface-bound IgM, but the nature of this interaction is not well understood. These experiments utilize HDX-MS to uncover key interactions and dynamics during complement cascade initiation. HDX-MS is a great tool to study this complex mixture of proteins in an atypical experimental environment. We used protein-based nanoparticles as a model surface to display IgM antigens and induce the surface-bound IgM conformation and this novel approach has the potential to show HDX-MS applied to conditions beyond typical in-solution 1:1 binding interactions.
Methods 115/120: Anti-SARS-CoV-2 spike receptor binding domain IgMs were used for all experiments including two different constructs, pentameric and hexameric IgM. These constructs were first incubated with nanoparticles that displayed SARS-CoV-2 spike receptor binding domain spaced at regular intervals. Samples were prepared using standard HDX conditions and labelled for time points between 1min and 4hrs in D2O buffer. The experiment was repeated but the immobilized IgM samples were incubated with C1q prior to labelling. Samples were quenched to pH 2.5 and frozen in liquid nitrogen before analysis. All samples were thawed and digested using an in-line Nepenthesin-2 protease column with subsequent LC/MS analysis. All experiments included a set of internal exchange standards to ensure comparability between conditions.
Preliminary Data 108/300: Previous experiments our lab has done have provided insight into changes that occur in IgM when bound to a surface vs. bound in solution. Changes in the Fc domain of the IgM upon surface binding showed that there is a unique conformation to surface-bound IgM that likely accounts for the increase in complement activity from these surface-bound molecules. Initial experiments have shown that the nanoparticles used in this experiment give rise to a conformation of IgM that can activate the complement cascade. All of the proteins tracked in this experiment have been digested using a NEP-2 column and have shown sufficient sequence coverage for meaningful HDX data interpretation.
Novel Aspect 18/20: HDX-MS highlights details of the C1-IgM interaction that initiates immune activation using a complex nanoparticle antigen display system.

Authors
Nicholas M. Riley1, Jingjing Huang2, David Bergen2, Amanda E. Lee2, Rafael D. Melani2, William D. Barshop2, John E.P. Syka2, Jesse D. Canterbury2, Vlad Zabrouskov2, Graeme C. McAlister2, Christopher Mullen2

Institutions
1Department of Chemistry, University of Washington, Seattle, Washington, USA 2Thermo Fisher Scientific, San Jose, California, USA

Abstract
Recent hardware changes introduced on the Orbitrap Ascend Tribrid MS include dual ion routing multipoles (IRMs) that can be used to parallelize accumulation, dissociation, and mass analysis of three separate ion populations. Here we explore how this architecture improves N- and O-glycopeptide characterization by increasing scan acquisition speeds without sacrificing spectral quality. The balance between scan speed and MS/MS product ion signal-to-noise is especially important in glycoproteomics. Complexities of glycopeptide fragmentation necessitate large precursor ion populations, and consequently, long ion accumulation times, for quality MS/MS spectra. To compound matters further, dissociation methods like electron transfer dissociation (ETD) that benefit glycopeptide characterization come with overhead times that also slow down scan acquisition. Conversely, heterogeneity inherent to glycosylation means that any given retention time during an LC-MS/MS analysis may contain numerous glycopeptide species to target through data-dependent acquisition. Often duty cycle is sacrificed to some degree, which results in higher quality spectra of abundant species but leaves other precursor ions under-sampled. We analyze mixtures of N- and O-glycopeptides to show that 20-30% more MS/MS scans can be acquired when parallelizing three ion populations using the dual IRMs of the Orbitrap Ascend. This translates to 10-20% gains in glycopeptide identifications depending on dissociation type(s), scan acquisition schemes, and method parameters (e.g., precursor ion accumulation and reaction times). Focusing on O-glycopeptide analysis with ETD-based methods, we also explore how acquisition rates and ion-ion reaction times affect identifications and product ions generation. We show what parameters need to be considered in O-glycopeptide characterization to generate c- and z-type ions that can be used for O-glycosite localization while also maximizing scan acquisition rates to improve total site-localized O-glycopeptide identification. In all, we show how architectural changes to the Tribrid MS platform benefit glycoproteomic experiments by parallelizing scan functions to minimize overhead time and improve sensitivity.

Authors
Abhigya Mookherjee, Alesi Escobedo, Mike Guttman

Institutions
Department of Medicinal Chemistry, University of Washington, Seattle

Abstract
Modern glycoproteomics platforms have enabled in-depth characterization of complex glycoprotein samples. LC-MS/MS can reveal much about glycopeptides including the peptide sequence, location of the glycan attachment, and the glycan composition. However, the information is often insufficient for extracting details of the glycan structure including linkage and anomeric configurations. We have taken various approaches to fill in this missing information about glycan structure using a combination of approaches including multi-stage tandem MS, high resolution ion mobility, and gas-phase hydrogen/deuterium exchange. These techniques have revealed that even seemingly simple carbohydrate structures can adopt a complicated ensemble of conformational states in the gas-phase that can confound direct structural interpretation. With the aim of generating a practical approach to improve glycan characterization, we have begun to benchmark the structures and fragmentation behaviors for various protonated mono and oligosaccharide structures that are relevant to glycoconjugates. We have now generated an LC-MS3/MS4 approach that is compatible with many proteomics workflows and can fill in much of the missing glycan structural information, particularly for O-linked glycosylation.

Authors
Kristian Swearingen 1, Jimmy Eng 2, Vladimir Vigdorovich 3, Priya Gupta 3, David Shteynberg 1, Luis Mendoza 1, Ashley Vaughan 3, D Noah Sather 3, Eric Deutsch 1, Robert Moritz 1

Institutions
1. Institute for Systems Biology 2. University of Washington 3. Seattle Children's Research Institute

Abstract
Glycosylation of proteins with covalently-linked sugar moieties is a ubiquitous and essential co- and post-translational modification. Until recently, however, it was debated whether protein glycosylation occurred in Plasmodium, the single-celled eukaryotic pathogen responsible for the disease malaria. Using mass liquid chromatography-mass spectrometry (LC-MS), we discovered that conserved thrombospondin type 1 repeat (TSR) domains in the parasite can be modified with two rare glycosylations: O-linked fucose at Ser and C-linked mannose at Trp. Here, we will present the specially tailored LC-MS methodologies we have developed in order to characterize TSR glycosylation in Plasmodium, techniques that enable us to enrich, target, and measure the stoichiometry of these protein modifications in the parasite.
When subjected to the gas-phase collision-induced dissociation typical of shotgun proteomics, glycopeptides exhibit species-specific fragmentation and neutral losses that render them invisible to standard proteomic database search workflows. To overcome this obstacle, we have modified an existing mass spectrometry data search engine, Comet, to accommodate these neutral losses, enabling us to detect these PTMs through the bioinformatics workflows contained in the Trans-Proteomic Pipeline, an open-source proteomics data processing platform. We also exploit a recently developed monoclonal antibody to enrich C-mannosylated peptides, thereby enabling detection of low-abundance glycopeptides. We will show how these methods have enabled us to produce the most comprehensive survey to date of protein glycosylation in Plasmodium, revealing that the parasite exhibits both highly conserved as well as apparently novel glycobiology.