2019 SUMS-RAS Posters
Assigned poster numbers are in square brackets [#]
[1]
Multiplexing in Charge Detection Mass Spectrometry: Rapid Measurement of Large Native Proteins and Macromolecular Complexes
Conner C. Harper, Andrew G. Elliott, Evan R. Williams
University of California, Berkeley
Measuring masses of large (MDa) biomolecules and macromolecular complexes using conventional mass spectrometry can be challenging with heterogeneous or high salt-containing samples owing to unresolved charge-state distributions. Charge detection mass spectrometry (CDMS) can eliminate the problem of sample heterogeneity by measuring the masses of individual ions via simultaneous measurements of m/z and charge. However, generating a mass histogram with high dynamic range can be time consuming, as many thousands of ions must be individually weighed. Although faster than other single ion techniques, CDMS is still limited by the time required to acquire enough individual ion measurements to accurately represent all the constituents of the sample. To decrease analysis time, we demonstrate multiple ion measurements in a single trapping period, which significantly increases the rate of mass measurements. To probe the potential effects of ion-ion interactions, ion trapping events with different numbers of trapped ions were binned into separate m/z and mass histograms. These results show that the m/z or mass measurements do not depend on the number of ions trapped within the limits of the analysis method. Thus, simultaneously trapping and measuring ions greatly decreases the time required to produce a CDMS mass histogram with virtually no disadvantages. Using this technique, mass histograms for biologically relevant samples, such as encapsulin nanocompartments (1.84 MDa), were obtained. For encapsulin, charge-state resolution could not be achieved using a conventional Q-TOF instrument, but CDMS analysis revealed two principal masses at ~1.84 MDa and ~2.00 MDa, corresponding to empty and cargo-containing species.
[2]
Elucidating alternative biological pathways with tailored enrichment strategies from clinical tissue samples
Kratika Singhal, Rowan Matney, Fang Liu, Ryan Leib, Allis Chien
Stanford University Mass Spectrometry, Stanford, CA
Human tissues obtained during autopsies and biopsies serve as invaluable clinical resources for diagnosis, treatment response, and disease progression. However, limitations in their availability, size, and preservation prove persistent challenges for proteomic assays. Preservation of biological specimens via formaldehyde fixation and paraffin embedding (FFPE) solves many challenges and keeps samples stable at room temperature for years, but introduces additional complexity in terms of sample preparation and protein modifications. Here, we compare frozen and FFPE tissues of various proteomic complexities, including heart, brain, and lung. We aim to compare biological variability introduced as an effect of formalin fixation, as well as differential selection of peptides enriched with hydrophilic interaction chromatography (HILIC) vs. reversed phase C18.
Peptides were identified with Byonic against identical protein databases allowing for common PTMs. In frozen samples, we observed 1749 proteins from C18 prep and 1678 proteins from HILIC prep, with 20k peptide spectrum matches in both cases. In FFPE samples, we observed 512 proteins from C18 prep and 1105 proteins from HILIC prep, with peptide spectrum matches of 2.5k and 8.5k respectively. Overlap in terms of proteins identified, by different enrichment protocols, was 55% for frozen tissues and 26% for FFPE samples. While enrichment efficiency of both techniques was comparable for frozen samples, HILIC enrichment was more efficient for FFPE samples.
FFPE and frozen samples are broadly similar in terms of the observed protein families and their biological functions. However, detailed analysis reveals that for specific proteins and pathways identified using tools such as GeneTrail2, the method of peptide enrichment results in alternative protein pathways being observed. Samples prepared using HILIC resin favor identification of proteins involved in glycolysis for both frozen and FFPE tissues, while C18 preparation of the same tissues preferentially identified proteins involved in GTP hydrolysis and ribosomal assembly. Additionally, some pathways are specific to both the sample preservation method and the method of sample preparation. For instance, we observe proteins from FFPE tissue prepared using HILIC sample preparation related to glucose transporter disorder which are not observed in frozen specimens in either preparation. Overall, we observe lower protein identifications in FFPE tissues, as has been reported previously, but some protein observations are favored in this preservation method. These observations could be attributed to formalin fixation step, perhaps due to different rates of degradation across protein families. Further analysis of additional FFPE and frozen tissues, including lung tissue with adenocarcinoma and brain tissue with glioblastoma, with both sample preparation methods will be presented.
[3]
Improved TOMAHAQ data normalization for large-scale protein quantification and characterization
Fang Liu (1), Swati Acharya (2), Eric Smith (2), Kratika Singhal (1), Rowan Matney (1), Nonhlanhla Lunjani (3), Dries Van Elst (3), Milena Sokolowska (3), Cezmi A. Akdis (3), Kari Nadeau (2), Ryan Leib (1), Allis Chien (1)
(1) Stanford University Mass Spectrometry, Stanford, CA; (2)Stanford University School of Medicine, Stanford, CA, 94305; (3) Swiss Institute for Allergy and Asthma Research, University of Zürich, Davos, Switzerland
Mass spectrometry-based targeted proteomics assays have enabled reproducible, sensitive, and selective protein identification and quantitation in biomedical research and clinical applications. However, the assay performance can be compromised when the number of peptide targets within the same run gets large. The present work utilizes a targeted proteomics assay (TOMAHAQ) that combines both peptide and sample multiplexing using synthetic peptide triggers for high-throughput protein characterization and quantification across a clinical population representing allergy-related disease states and controls. Improvements to assay accuracy, particularly in terms of data processing and normalization, will be discussed here.
In total, 185 peptides from 53 gastrointestinal proteins were selected to represent potential markers of allergic disease. Plasma samples from a clinical population representing multiple disease states (n=116) and healthy controls (n=90) were measured to define unique peptide signatures of allergic endotypes using the TOMAHAQ approach. The TOMAHAQ assays were optimized by allowing a 4-min elution window, 5 scans for trigger peptides and 6 SPS scans for target peptides. After method optimization, we observed 43 of the 53 targeted proteins with at least one marker peptide in one or more of the clinical samples. Quantitative comparison of these peptides depends on the presence or absence of given proteins in specific samples.
Interestingly, we observed systematically higher abundance in the 126, 127C and 128C channels of TMT10plex across all 23 series. We hypothesize that the observed isotopic envelope may be in part due to a superposition of the envelope originating from TMTzero label used for the synthetic peptide triggers. To test this hypothesis, the pooled synthetic peptides were analyzed alone using the same TOMAHAQ assays. Significant TMT reporter ion signals were observed in the 126, 127C and 128C channels, indicating inefficiency in the isolation of synthetic peptide reporter ions even under typical TOMAHAQ conditions where nominally they should not be observed.
Our preliminary results indicate that normalizations or corrections will be necessary to unlock the great potential in using a TOMAHAQ-based strategy for simultaneous and accurate quantitation of hundreds of peptide targets across tens of samples. Ion signals from TMTzero-labeled synthetic peptides were found to be greatly dependent on the factors affecting m/z offset between trigger and target peptides (e.g., charge state, whether or not TMT was incorporated into both N-terminus and lysine residues). Our efforts focused on potential data normalization strategies will be discussed.
[4]
Simultaneous Unbiased Structural Analysis of Cerebrospinal Fluid Proteins in N-Glycosylation and Aging Using Limited Proteolysis-Mass Spectrometry (LiP-MS)
Steven R. Shuken (1,2), Marie-Therese Mackmull (4), Oliver Hahn (2), Benoit Lehallier (2), Niclas Olsson (3), Joshua E. Elias (3), Paola Picotti (4), Tony Wyss-Coray (2)
(1) Dept. of Chemistry, Stanford University; (2) Dept. of Neurology & Neurological Sciences, Stanford SoM; (3) Dept. of Chemical & Systems Biology, Stanford SoM; (4) Institute of Molecular Systems Biology, ETH Zurich
Age is a major risk factor for neurodegenerative diseases, yet brain aging is a poorly understood process. An understanding of changes that occur with brain aging may allow us to discover needed biomarkers for pre-dementia disease states and/or elucidate effective therapeutic targets. Cerebrospinal fluid (CSF) is an existing source of abundance-based diagnostic biomarkers for brain aging and Alzheimer's disease (AD). Our project seeks to go beyond classical abundance-based biomarkers to find structure-based biomarkers in the CSF for aging and AD. Limited proteolysis-mass spectrometry (LiP-MS) is a method that induces non-residue-specific cleavages by the addition of proteinase K in the native state before trypsinization; the resulting semitryptic peptides and their tryptic counterparts can be quantitatively analyzed to reflect changes in protein structures. We validated LiP-MS in a well-understood structural context, N-glycosylation states, by performing LiP-MS after treating CSF with PNGase F. We then performed LiP-MS on whole mouse CSF proteomes from mice of different ages. Preliminary results suggest that disease-relevant structural changes occur in the CSF proteome with aging.
[5]
artMS: An R Package for the Analysis of Large-Scale Proteomics Datasets
David Jimenez-Morales (1), Alexandre Rosa Campos (2), John Von Dollen (3,5,6), Matthew T Wheeler (1), Euan A Ashley (1), Nevan J. Krogan (3,4,5,6), Danielle L. Swaney (3,4,5,6)
(1)Department of Medicine, Division of Cardiovascular Medicine, Stanford University, Stanford, California; (2) Sanford Burnham Prebys Medical Discovery Institute, La Jolla, California, USA; (3)UC San Francisco, Quantitative Biosciences Institute, QBI, San Francisco, CA, USA; (4)J. David Gladstone Institutes, San Francisco, CA, USA; (5)UC San Francisco, Department of Cellular and Molecular Pharmacology, San Francisco, CA, USA; (6)The Cancer Cell Map Initiative (CCMI), La Jolla and San Francisco, California, USA.
Mass Spectrometry-based proteomics is the state-of-the-art technique used to detect, identify, and quantify proteins, with a wide range of applications through the study of protein abundances, post-translational modifications, and protein-protein interactions. It has led to the discovery of molecular mechanisms in many biological processes and is emerging as a supporting technology for the development of promising diagnostics and therapies. Proteomics also plays an essential role in emerging multi-omics approaches to gain a system view of complex biological events. However, the basic processing of proteomics data is a laborious task requiring many different levels of analysis to ensure data quality, quantification, and frame results in a biological context. To automate and simplify these many steps and enable rapid and robust analysis and interpretation of proteomics data, we have developed the open-source R package artMS "Analytical R Tools for Mass Spectrometry". artMS provides a set of tools for the analysis and integration of large-scale MS proteomics datasets obtained using the popular proteomics software MaxQuant. Currently it supports both SILAC and label-free quantification experiments. The functions available in artMS can be grouped into the following categories: (1) quality control (QC), (2) relative quantification (3) downstream analysis and integration of quantifications (enrichment, clustering, PCA, etc), and (4) generation of input files for other tools, including Photon, and Phosfate. Importantly, artMS supports the analysis of protein abundance and posttranslational modification site quantification collected in both single-shot or fractionation-based experiments. Additionally, it supports protein-protein interaction analysis by enabling the generation of input files for the tools SAINTq and SAINTexpress. artMS is available in Bioconductor, the R open source project for bioinformatics tools (visit http://artms.org for details).
[6]
Proteomic Variation in Human Heart Tissues
Huaying Fang (1), Lihua Jiang (1), Ruiqi Jian (1), Joanne Chan (1), Felipe da Veiga Leprevost (2), Alexey Nesvizhskii (2), Michael Snyder (1), Hua Tang (1)
(1) Department of Genetics, Stanford University, (2) Computational Medicine and Bioinformatics, University of Michigan
Variation in gene expression between individuals, and the genetic determinisms that influence gene expression, have been extensively investigated at the RNA level across human tissues. In contrast, variation in protein abundance in normal tissues has remained largely uncharacterized. Here we report preliminary analysis on more than 100 human heart tissues, collected as part of the GTEx project. Substantial variation in protein abundance is observed, which is associated with donor's age, sex and health status, as well as with genetic variation. Although there is a consistent and positive correlation between the abundance of a protein and the corresponding RNA transcripts, the correlation is surprisingly low, supporting the important roles of post-transcriptional regulatory mechanisms.
[7]
Limit-of-Detection Proteomics Characterized More & Lower Abundance MS1 Peptides
David Chiang, Patrick Chu
Sage-N Research, Inc.
Clinical discovery proteomics requires analyzing low abundance modified peptides (LAMPs). We present a novel method down to limit-of-detection (LOD). Under ideal conditions, 12 raw ions (4 MS1, 8+ MS2) can identify/quantify a protein with localized PTMs.
Proteomics data typically contains more peptide ions (MS1) than fragment ions (MS2). One-third of MS1 peptides can be near-LOD with only 2 isotopic ions -- still enough to calculate accurate mass and gross quantitation (apex intensity). In other words, MS1 is a treasure trove of LAMPs with solid mass and quant info, but is largely untapped because of uncertain identity.
We show its's possible to infer anonymous identity by triangulating MS1 masses to MS2 IDs. For example, in a phospho-peptide dataset, we can infer the sequence of a LOD peptide if its mass is exactly one phosphorylation less than an identified phospho-peptide. For MS2 LOD analysis, a sequence of 8+ is often protein-unique. This means matching enough near-contiguous +1 y-ions can identify the protein plus in-range PTMs without statistics.
Many research projects involve PTMs of one known protein. Labs use trial-and-error DIA to select presumed modified peptides for MS2 identification. That is basically shooting in the dark and won't work for LAMPs, so it's possible to be empty-handed after months of experimentation.
A better way is to mine MS1 data to build a 2D map {(mass, time, quant)} of MS1 peptides of interest, for example those exactly one PTM mass different from MS2 peptide IDs (DIA or DDA). The map can either answer specific questions or optimize DIA experimentation from which another map is derived.
By letting researchers see the smallest details from the latest data, LOD analysis accelerates research while reducing risk and cost.
[8]
Tandem MS Strategies for Intact N- and O-Glycopeptide Characterization
Nicholas M. Riley (1), Stacy A. Malaker (1), Marc D. Driessen (1), Carolyn R. Bertozzi (1,2)
(1) Department of Chemistry, Stanford University, Stanford, California, USA; (2) Howard Hughes Medical Institute, Stanford, California, USA
Intact glycopeptide characterization is an imperative, yet challenging, component to modern glycoproteome analysis. Elucidation of both glycan and peptide modalities often requires multiple tandem mass spectrometry (MS/MS) approaches to identify a single glycopeptide species. Recent efforts have focused on developing fragmentation techniques to allow for rapid elucidation of both moieties to promote high-throughput glycoproteomic investigations. A few studies have highlighted the utility of stepped-collision energy higher energy collisional dissociation (SCE-HCD) and electron transfer dissociation with supplemental HCD activation (EThcD) for large-scale glycoproteomic applications, but these methods still require further investigation to understand their utility in the glycoproteomic toolbox. Both have been used with success in large-scale glycoproteomic experiments, but they each incur some degree of compromise. Here, we systematically explore the advantages and disadvantages of SCE-HCD and EThcD for intact glycopeptide analysis and comment on their suitability for both N- and O-glycoproteomic applications. We compare standard HCD, SCE-HCD, ETD, and EThcD for both N- and O-glycopeptides generated from a panel of glycoprotein standards and glycopeptides enriched from complex HEK 293 lysates. We rely mainly on trypsin to generate glycopeptides due to its ubiquitous use in glycoproteomic workflows, but we also investigate the performance of these MS/MS methods for O-glycopeptides generated using recently described O-glycoproteases. Throughout these comparisons, we evaluate both the number of identifications produced by each method along with six Figures of Merit that comment on spectral quality and the quality of information obtained from each approach. With this work we provide data to highlight the strengths and weaknesses of both SCE-HCD and EThcD for intact N- and O-glycopeptide analysis, commenting on which method is best suited for a given application.
[9]
Bone proteomics: enhancing homogenization of bone samples for increased proteomic depth
Rowan Matney, Kratika Singhal, Fang Liu, Ryan D. Leib, Allis S. Chien
Stanford University Mass Spectrometry, Stanford, CA
Vertebrates rely on bone integrity for both structure and mobility. Investigating the protein content of bone is difficult, due in part to the calcified extracellular structure and the presence of highly abundant structural proteins such as collagen. This calcified and crosslinked structure is a difficult medium from which to extract a robust, repeatable sample of the bone proteome. To improve the proteomic recovery for bone, we optimize the physical homogenization by pairing grinding, chopping, or crushing the tissue with repeated bead-beating. We compare these methods as a function of protein recovery and proteomic depth using an Orbitrap mass spectrometer. With our improvements to bone homogenization, we are able to achieve greater proteomic depth and consistency.
Using bone samples excised from rat shoulders, we compared solubilized proteins and peptides under various mechanical homogenization conditions. Visual inspection and initial protein quantification results indicated that bead-beating with 2.8 mm ceramic beads released more solubilized protein for analysis than the smaller metal or ceramic beads investigated. In all cases, the number of observed peptides and proteins was significantly lower in the second fraction. This drop-off indicates that the first round of bead-beating liberates the majority of proteins for analysis and is crucial for protein extraction. Additionally, initial homogenization method was not trivial – both hammer and mortar and pestle yielded slightly higher protein and peptide counts than the bead-beating only control or the rough chopping. We extracted 13.7 μg and 15.3 μg of protein from the first fractions of the hammered samples and the mortar and pestle ground samples, respectively. Note that these protein numbers represent only 1/5th of the total proteins extracted from each sample, due to aliquots taken before acetone precipitation. Raw data were searched for peptide and protein assignments using Byonic, allowing for common post translational modifications. We observed 307 proteins from hammered samples with 3324 peptide spectral matches. For the mortar and pestle samples, we observed 459 proteins with 4426 peptide spectral matches. These results indicate that while we can satisfactorily homogenize samples to produce similar amounts of total protein, there is a significant increase in proteomic depth by using a mortar and pestle to grind bone material prior to bead beating. In addition to these results, we will also explore options to dissolve powdered bone to further improve protein recovery, and investigate the robustness of each of these grinding methods across biological replicates.
[10]
A Recombinant Asp-Specific Protease for Bottom-up Mass Spectrometry Workflows
Chris Hosfield (1), Jim Hartnett (1), Alba Katiria González Rivera (1, 2), Ethan Strauss (1), Sergei Saveliev (1), Michael Rosenblatt (1) and Marjeta Urh (1)
(1) Promega Corporation, Madison, WI; (2) University of Wisconsin-Madison, Madison, WI
Bottom-up mass spectrometry workflows typically utilize trypsin to digest proteins into peptides suitable for LC-MS/MS analysis. While trypsin is an excellent protease, alternative proteases are useful for numerous applications including increasing protein sequence coverage and identifying post-translational modifications. LysC, another commonly used protease, is also robust and specific, but has specificity similar to trypsin. Site-specific proteases with orthogonal specificity such as GluC and AspN are useful but can suffer from relatively poor digestion performance. Here we report the expression, purification and characterization of a protease which displays both high cleavage efficiency and a strong preference for cleavage N-terminal to aspartic acid. This new recombinant protease should have broad utility for bottom-up LC-MS/MS workflows.
[11]
Integrated software platform for analyzing hydrogen-deuterium exchange and oxidative footprinting data for solvent accessibility
Wilfred Tang (1), Rose Lawler (1), Yong J. Kil (1), Eric Carlson (1), Marshall Bern (1), Saketh Chemuru (2), Nicole D. Wagner (2), Liuqing Shi (2), Henry Rohrs (2), Daisy W. Leung (2), Michael L. Gross (2)
(1) Protein Metrics, San Carlos, CA; (2) Washington University, St. Louis, MO
[12]
Integrated Omics Analysis Across 32 Human Tissues
Lihua Jiang, Meng Wang, Shin Lin, Ruiqi Jian, Joanne Chan, Xiao Li, Huaying Fang, Hua Tang, Michael Snyder
Stanford University, Stanford, CA
Using TMT/MS3 based quantitative proteomics strategy, we profiled the proteome of 200 samples from 32 types of human tissues. In total, proteins from more than 13k genes were accurately quantified. It is well known that RNA and protein has poor correlations due to post transcriptional/translational regulations. To investigate their correlation at tissue level, we integrated our proteomics data with the corresponding RNA expression data using a novel model based robust normalization method and robust z-score for each protein. Spearman correlation based on the robust z-score shown that more than 7000 protein and RNA were significant correlated across 32 tissues. The concordance and discordance analysis of protein and RNA in each tissue revealed specific post transcriptional/translational regulation at tissue level. In liver, we observed highest number of protein RNA discordance. A group of RNAs was highly enriched in liver but the corresponding proteins were instead enriched in arteries. GO analysis shown most of these proteins are complementary factors, albumin, and transporter proteins et.al. which are well known secreted proteins. Although high RNA level was translated to high protein level in liver, the proteins were constantly secreted to blood stream, which led to the discordance between proteins and RNAs. Among the entire proteome, more than 2500 proteins are predicted to be secreted proteins. The concordance analysis shown that not all of them are in discordance as one would expect. This analysis supports two secretion mechanisms: the continuous secretion (discordance) and regulated secretion (concordance).The integrated analysis can help us understand tissue level regulation and the communications through proteins. Different enrichment of protein and RNA, selection of isoforms and upstream regulators etc. from the integrative analysis shown that information from RNA to protein are fine-tuned through multiple level of regulation.
[13]
AMS and PAMMS: Unique measurement tools for microdosing experiments
Benjamin Stewart (1), Erin. Madeen (2), Ted Ognibene (1), Richard Corley (2), Tammie McQuistan (2), Marilyn Henderson (2), William Baird (2), Graham Bench (1), Kenneth Turteltaub (1), David Williams (2), Bruce Buchholz (1)
(1) Lawrence Livermore National Laboratory, (2) Oregon State University
Microdosing is a promising approach for rapid characterization of metabolic and pharmacokinetic (PK) profiles of sub-pharmacological doses of drug candidates. A microdose is defined as one one-hundredth of the anticipated therapeutic dose. Early human data can eliminate compounds with poor PK profiles early in the development process, accelerating investment in more promising candidates. Access to analytical methods with high sensitivity and specificity is a prerequisite for successful microdosing experiments. Accelerator mass spectrometry (AMS) and parallel accelerator and molecular mass spectrometry (PAMMS) provide these necessary analytical capabilities. AMS and PAMMS were used to identify DNA adducts generated by suspected carcinogens in human subjects and in vitro experiments at environmentally relevant exposure levels.
[14]
An Investigation into the use of cyclic ion mobility for the separation of biopharmaceutical peptide and protein modifications
James Langridge (1), Martin Palmer (1), Weibin Chen (2), Dale A. Cooper-Shepherd (1)
(1) Waters Corporation, Wilmslow, Cheshire, UK, (2) Waters Corporation, Milford, MA
The SELECT SERIES cyclic IMS enabled q-tof has a flexible geometry for research applications. Isomeric peptides containing Asp and isoAsp residues can be separated by ion mobility. This poster will demonstrate product ions containing isomeric residues differ in their gas phase structure, and that IgG1 and IgG2 have unique arrival times and can be separated by ion mobility.
[15]
Novel metabolic profiling using a single injection with reverse-phase/ion-exchange chromatography and mass spectrometry
Anthony Le (1), Tina Cowan (1), Justin Mak (2)
(1) Stanford University, Department of Pathology; (2) Stanford Health Care, Biochemical Genetics
Metabolomics is one of the newer -omics and supports numerous scientific fields, including drug development, fundamental biomedical research, translational clinical research, and clinical diagnostics. Conventionally, the underlying technologies used in metabolomics are reverse-phase (RP) and hydrophilic interaction chromatography (HILIC) coupled to mass spectrometry (MS). However, the parallel and independent use of RP and HILIC methods create redundant data. Therefore, a single chromatographic method with expanded coverage of the metabolome would be advantageous. An in-line C18-ion exchange column arrangement was developed to analyze polar and non-polar metabolites, such as amino acids, acylcarnitines, and organic acids, without derivatization or ion-pairing reagents. To evaluate metabolome coverage, 400 authentic standards were acquired. In addition to capturing amino acids, organic acids, acylcarnitines, bile acids, nucleosides, steroids, and vitamin cofactors, a 20-min method was able to resolve numerous isomers, such as isocitrate from citrate, methylmalonate from succinate, and alloisoleucine from isoleucine/leucine. Importantly, the method produced good data quality when 5 plasma samples were each injected 5 times and evaluated by unsupervised principal component analysis. Furthermore, the data showed good sensitivity at physiologic concentrations. Together, this method has the potential to supplant the existing chromatographic techniques for metabolic profiling.
[16]
Use of Metabolomics to Phenotype and Categorize Risk in Human Health Studies Accelerates Clinical Care and Drug Development
Gregory A. Michelotti, Michael Fanelli, Megan Showalter, Cynthia Taylor Lawley, Robert Mohney
Metabolon, Inc.