Agenda

Profiling human interactome is critical in understanding the molecular basis for nearly all processes of life. Over the years, we’ve advanced crosslinking mass spectrometry by developing experimental methods and software tools to identify tens of thousands of PPIs from whole cells. These data reveal numerous aspects of living systems - for example protein subcellular localizations, virus-host interactions, and architectures of suprabiomolecular machineries. Furthermore, these data offer unprecedented opportunities to profile interactome changes between tissues and disease states, providing invaluable training data for AI-based methods to identify PPI-mediating motifs, inform new protein/antibody designs and screen for drug targets.

Different stable isotope labeling strategies coupled to mass spectrometry have been used to measure protein turnover rates in vivo and in vitro. Our group developed Riana (Relative isotope abundance analyzer) which performs isotopomer quantification and kinetic curve fitting in a manner compatible with various SILAC and elemental labeling approaches. Using Riana, we have compared the turnover rates of proteins across four mouse tissues, obtained from the same inbred mouse strain maintained under identical husbandry conditions, measured using either heavy water or heavy lysine as the labeling precursor. We describe an isotopomer selection strategy that can improve the depth of protein turnover quantification in vivo and in vitro.

A wealth of proteomics data is being generated every day to study a wide range of biomedical challenges. These datasets are often multi-modal, paired with other omics and metadata. Further, there is a community push from funders and researchers to make data FAIR (Findable, Assessable, Interoperable, Reusable). This can be challenging with multi-modal datasets, but even when achieved, often the domain knowledge needed to engage AI/ML-experts is considerable. Thus, a core challenge to engage the AI/ML community is the packaging of the data in a form that can be easily understood by non-experts, or those not involved in the project from where the data originated. Best practices for generating AI/ML-datasets will be discussed and demonstrated though our current dissemination of AI/ML-ready datasets pipeline which is amenable to new methods to improve the quality of single- and multi-omics datasets (e.g., imputation, batch correction, etc.), as well as clean production level datasets for predictive modeling to elicit biomarkers and pathway-level molecular signatures from the data.

The declining capacity of cells to maintain a functional proteome is a major driver of cellular dysfunction and decreased fitness in aging. Here we assess the impact of aging on multiple proteome dimensions, which are reflective of function, across the replicative lifespan of Saccharomyces cerevisiae. We quantified protein abundance, protein turnover, protein thermal stability, and protein phosphorylation in mother yeast cells and their derived progeny at different ages. We find progressive and cumulative proteomic alterations that are reflective of dysregulation of complex assemblies, mitochondrial remodeling, post-translational activation of the AMPK/Snf1 energy sensor in mother cells, and an overall shift from biosynthetic to energy-metabolic processes. Our multidimensional proteomic study systematically corroborates previous findings of asymmetric segregation and daughter cell rejuvenation, and extends these concepts to protein complexes, protein phosphorylation, and activation of signaling pathways. Lastly, profiling age-dependent proteome changes in a caloric restriction model of yeast provided mechanistic insights into longevity, revealing minimal remodeling of energy-metabolic pathways, improved mitochondrial maintenance, ameliorated protein biogenesis, and decreased stress responses. Taken together, our study provides thousands of age-dependent molecular events that can be used to gain a holistic understanding of mechanisms of aging.

In recent years, protein footprinting coupled with mass spectrometry has been used to identify protein-protein interaction sites and regions of conformational change through modification of solvent accessible sites in proteins. The footprinting method, fast photochemical oxidation of proteins (FPOP), utilizes hydroxyl radicals to modify these solvent accessible sites. To date, FPOP has been used in vitro on relatively pure protein systems. We have further extended the FPOP method for in vivo analysis of proteins. This will allow for study of proteins in their native cellular environment and be especially useful for the study of membrane proteins which can be difficult to purify for in vitro studies. A major application of the in vivo method is for proteome-wide structural biology. To this end, we have further developed the method for patient samples, specifically peripheral blood mononuclear cells (PBMCs). We have optimized several parameters for labeling of proteins in these patient samples. This method have the potential to become a powerful tool in the structural biology toolbox.

The process of discovery requires stepping into the unknown and trying to learn something new.  In my experience good notes and thoughtful meditation on results have resulted in the most insightful discoveries.  This suggests a recipe to succeed in proteomics and any research-based career.  I will present a couple of specific examples from my career and try to outline principles that lead to success.  Interestingly, my recipe for success incorporates the essence of a couple frequently repeated idioms.  1. A well-ordered and thoughtful experiment will teach you something regardless of whether it supports your hypothesis.  2. Just because you failed doesn't mean you are a failure.  3. Luck favors the prepared.   In my experience the wealth of information and opportunity for confusion versus discovery in a proteomics experiment increases the value of these ideas.

Empirical evidence suggests that mass spectrometry data acquisition strategies commonly applied in bottom-up proteomics, such as data-dependent acquisition and collision-based activation of precursor ions, are not particularly effective for characterizing intact proteoforms >30 kDa. This should not be very surprising: substantial signal dilution occurs when large proteins are transferred into the gas-phase via electrospray ionization, and their fragmentation via traditional ion activation methods produces incomplete series of highly charged product ions that create congested mass spectra. Alternative approaches to the investigation of large proteoforms will be presented, demonstrating that the above-mentioned issues can be mitigated by the use of ion-ion and ion-photon reactions to simplify intact protein mass spectra and facilitate their interpretation.

Purpose: The low cancer specificity of Prostate-Specific Antigen (PSA), the principal blood biomarker for prostate cancer detection, leads to a high frequency of unwarranted prostate biopsies. To alleviate this deficit, we executed a proteomic discovery study seeking PSA reflex signatures using the Proteograph™ Product Suite, a multi-nanoparticle-based deep plasma/serum proteomics workflow (Seer, Inc.), with timsTOF Pro mass spectrometry (Bruker) to interrogate over 900 serum specimens from patients referred for biopsy based on elevated PSA and/or abnormal digital rectal exam.

Methods: Our study followed rigorous design principles including uniform collection, randomization and blinding of specimens, all of which were processed with the Proteograph Product Suite. Liquid chromatography-mass spectrometry (LC-MS) analyses leveraged the Bruker timsTOF Pro MS platform utilizing 30-minute reversed-phase chromatography and a label-free dia-PASEF (data independent acquisition - parallel accumulation serial fragmentation) data acquisition method. Peptides deriving from a chosen subset of specimens designed to maximize peptide diversity were pooled, fractionated and analyzed using DDA (data-dependent acquisition)-PASEF to build a spectral reference library.

Results: The DIA-NN algorithm was employed through the cloud-based Proteograph Analysis Suite (PAS) to search LC-MS data. Leveraging our study-specific spectral library proteomic depth achieved was approximately 3600 protein groups identified (median) per patient serum specimen and more than 4400 in at least 25% of samples across the study. Customized machine learning workflows and the SeerML pipeline were used to identify and assess diagnostic power of new proteomic signatures.

Conclusions: Our data demonstrate the remarkable proteomic depth achievable in a scaled patient serum specimen discovery study using a combination of Proteograph and timsTOF platforms. This significant effort revealed serum proteomic signatures with increased diagnostic performance over that provided by PSA blood measurement and traditional patient-associated metadata alone, providing a foundation for future clinical tests aimed to reduce the current high frequency of unnecessary prostate biopsy.

Advances in DNA sequencing and editing technologies have revolutionized our understanding of the molecular basis of many human diseases. However, many disease-relevant genes encode proteins that are poorly characterized and/or are considered “undruggable”, hindering our understanding of disease mechanisms and translating this knowledge into new therapies. Chemical probes offer a valuable way to directly interrogate the function and disease-relevance of proteins and can also serve as valuable leads for drug development, yet most proteins in the human proteome lack small-molecule ligands that can serve as probes. More generally, the boundaries, if any, on the ligandability, and therefore potential druggability, across native proteomes remains poorly understood. In this seminar, I will describe our lab’s efforts to develop powerful chemical proteomic strategies to broadly map ligandable proteins directly in cells, and how this information can be advanced into useful chemical probes to investigate protein function.

Deregulation of kinase function in cell signaling pathways is implicated in numerous cancers. In response, kinase inhibitors (KIs) have been developed to interact with these kinases for highly specific treatment. Though nearly 50 KIs have been FDA-approved, KI monotherapy is seldom curative, likely owing to tumor heterogeneity and acquired resistance. In response, effective combination therapies must be tailored to known resistance mechanisms to efficiently engage with their targets and exploit cellular vulnerabilities. However, standard drug screening tools (e.g., plasma analysis, western blot) are bulk in nature, and no established technology exists to quantify KI target engagement, concomitant with local protein expression, while assessing tumor response heterogeneity. To address these shortcomings, our group has (1) developed protocols to fluorescently label KIs (and other small molecule therapeutics) that mimic the native drug, (2) advanced a novel intracellular paired agent imaging (iPAI) platform to quantify drug target availability (DTA) with these fluorescent KIs, and (3) established and validated a highly multiplexed immunostaining strategy utilizing DNA barcoded antibodies, enabling in situ cyclic immunofluorescence (cyCIF) imaging. In this project, we combine these three complementary innovations into a fluorescence imaging platform we call TRIPODD (Therapeutic Response Imaging through Proteomics and Optical Drug Distribution and binding).Deregulation of kinase function in cell signaling pathways is implicated in numerous cancers. In response, kinase inhibitors (KIs) have been developed to interact with these kinases for highly specific treatment. Though nearly 50 KIs have been FDA-approved, KI monotherapy is seldom curative, likely owing to tumor heterogeneity and acquired resistance. In response, effective combination therapies must be tailored to known resistance mechanisms to efficiently engage with their targets and exploit cellular vulnerabilities. However, standard drug screening tools (e.g., plasma analysis, western blot) are bulk in nature, and no established technology exists to quantify KI target engagement, concomitant with local protein expression, while assessing tumor response heterogeneity. To address these shortcomings, our group has (1) developed protocols to fluorescently label KIs (and other small molecule therapeutics) that mimic the native drug, (2) advanced a novel intracellular paired agent imaging (iPAI) platform to quantify drug target availability (DTA) with these fluorescent KIs, and (3) established and validated a highly multiplexed immunostaining strategy utilizing DNA barcoded antibodies, enabling in situ cyclic immunofluorescence (cyCIF) imaging. In this project, we combine these three complementary innovations into a fluorescence imaging platform we call TRIPODD (Therapeutic Response Imaging through Proteomics and Optical Drug Distribution and binding).

Our recently established metabolome informed proteome imaging (MIPI) workflow was first implemented for a comprehensive molecular characterization of lignin degradation pathways at microscale resolution within a heterogeneous and multi-member microbial community.  Matrix-assisted laser desorption/ionization-mass spectrometry imaging (MALDI-MSI) was used to spatially resolve metabolite signatures and map lignin microhabitats that were further profiled by spatial proteomics utilizing microPOTS (microdroplet Processing in One pot for Trace Samples) technology. Integrated multi-omics approaches combined with statistical analyses correlated metabolic pathways with specific enzymes at spatial scale. Herein, we employ our multimodal MIPI workflow to map metabolomics hotspots and characterize functional pathways of arachidonic acid metabolism through human placental tissue.  These measurements provide a comprehensive and informative view of underlying inflammatory pathways in micron-scale regions of placenta villi.  High reproducibility of these MIPI studies show the ability to map distinct processes across diverse systems.

O-glycoproteases are an emerging class of enzymes that selectively digest glycoproteins at positions decorated with specific O-linked glycans. O-glycoprotease substrates range from any O-glycoprotein (albeit with specific O-glycan modifications) to only glycoproteins harboring specific O-glycosylated sequence motifs, such as those found in mucin domains. Their utility for multiple glycoproteomic applications is driving the search to both discover new O-glycoproteases and to understand how structural features of characterized O-glycoproteases influence their substrate specificities. Here, we show how O-Pair Search, a recently developed O-glycopeptide-centric identification platform that enables rapid searches and confident O-glycosite localization, can be used to determine substrate specificities of various O-glycoproteases de novo from LC-MS/MS data of O-glycopeptides.

All eukaryotes shared a common ancestor roughly 1.5 - 2 billion years ago, a single-celled, swimming microbe designated LECA, the Last Eukaryotic Common Ancestor. As a consequence, modern-day eukaryotes share a large fraction of their genes, nearly half of which were already present in a recognizable form in LECA, and strikingly, many modern genetic diseases and traits derive to some extent from these deeply conserved, ancient molecular systems. I'll talk about our work defining the proteome and interactome of these ancestral LECA proteins, based on integrating >26,000 mass spectrometry proteomics experiments from 31 current-day eukaryotes, and how we've applied this knowledge to learn contemporary human gene-disease relationships, identifying new causal genes for osteopetrosis, end-stage renal disease, and short-rib thoracic dysplasia.

Transcription factors (TFs) are the ultimate target of many signaling pathways, activating or deactivating gene expression to drive cellular responses. Dysfunction of TFs drives multiple diseases including cancer, autoimmune disease, fibrosis, and others; however discovering and developing small molecule therapeutics for TFs is challenging. One challenge is that TFs often have multiple functions and can regulate the expression of multiple genes, making it difficult to predict their exact effects on different processes in the body. In addition, TFs often have multiple binding sites on DNA, which can make it difficult to design drugs that specifically bind to and inhibit their activity. Another challenge is the lack of structural information for most TFs due to their high disorder, which means many of the most common drug development assays cannot be used for TFs. We have developed a mass spectrometry-based proteomics platform called TF-Scan to measure the effects of small molecules on the DNA binding function of TFs. The platform measures the entire TF landscape simultaneously, giving scientists working on TF drug development a method for assessing a compound’s in-cell activity against TF in the relevant cellular environment, and a selectivity assay that reports on the regulome-wide effects of the drug candidate. We have applied TF-Scan to several cancers, including neuroblastoma, rhabdomyosarcoma, and acute monocytic leukemia (AML). Here, we share the preliminary results from our latest screening campaign in AML where we profiled 1,000 covalent compounds against ~5,000 nuclear proteins in biological triplicate.