Developing new chemoproteomic platforms to decode and perturb the degradome

Abstract

Proteins are the functional units of the cell, orchestrating virtually every biological process from cellular metabolism and signal transduction to structural support and immune defense. The precise and temporal regulation of protein abundance is critical to maintain cellular homeostasis and, by extension, human health. Dysregulation of protein homeostasis can lead to many diseases including cancer, neurodegeneration, and metabolic disorders, which highlights the importance of understanding how exactly a protein is degraded. Doing so opens new opportunities to therapeutically harness and reprogram protein homeostasis. A subset of the human proteome is remarkably short-lived, with some proteins persisting for just minutes. Despite their brief existence, these proteins play fundamental regulatory roles, controlling key cellular processes and enabling rapid responses to external stimuli. Previous proteomic studies from our lab using cycloheximide pulse-chase experiments have revealed that approximately 5–10% of the human proteome consists of such short-lived proteins (SLPs). These proteins are enriched in functionally important protein classes, including transcription factors, ubiquitin ligases, and membrane-bound signaling receptors. Their transient nature reflects the cell’s need to tightly regulate their abundance, often through active degradation pathways. In a way, dissecting the mechanisms of SLP degradation provides a direct line of sight into the mind of the proteostasis network itself. The ubiquitin-proteasome system (UPS) and autophagy-lysosomal pathway (ALP) constitute a highly coordinated network of proteins responsible for maintaining protein homeostasis through the regulated degradation and stabilization of nearly all expressed proteins. Given their central role in cellular protein quality control and overall cellular function, it is striking how little is known about the turnover mechanisms of a substantial portion of the proteome. Despite these knowledge gaps, therapeutic strategies targeting the UPS have already yielded FDA-approved drugs effective in treating various cancers. These successes validate the UPS as a therapeutically tractable target and underscore the importance and opportunity of understanding natural protein regulatory mechanisms to better treat human disease. Key substrate-ligase pairs—such as NRF2-KEAP1, HIF1A-VHL, and TP53-MDM2—serve as paradigms for how the successful decoding of how an SLP protein is degraded can be exploited for drug development. Aside from these select few examples, however, the precise mechanisms underlying the degradation of the overwhelming majority of SLPs remain poorly understood. This raises fundamental questions that I sought to answer during my PhD: through which cellular pathways are these proteins degraded? What intrinsic features of a protein determine their specific degradation routes? Can we exploit our SLP degradome map to identify therapeutic pinch points to target historically ‘undruggable’ proteins like transcription factors? And lastly, can we build new discovery platforms to find molecules that precisely modulate the degradome? Mass spectrometry–based proteomics provides a unique opportunity to capture a comprehensive snapshot of the entire proteome in a single experiment. With chemical multiplexing using tandem mass tag (TMT) reagents, researchers can now label and analyze up to 35 samples simultaneously in one combined run. By leveraging multiplexed mass spectrometry–based proteomics, I recognized a unique opportunity to uncover the mechanisms underlying the degradation of hundreds of short-lived proteins (SLPs), to elucidate how the oncogenic transcription factor ASCL1 is regulated by its endogenous ligase, and to build a complementary drug discovery platform for identifying molecules that target the proteostasis network. In my first project, I developed a multiplexed chemoproteomics platform to systematically profile the degradome of a cell line of interest. This was achieved through a simple assay in which cells were treated with potent inhibitors targeting the proteasome (Bortezomib), lysosome (Bafilomycin A1), ubiquitin system (TAK243), neddylation system (MLN4924), and translation machinery (cycloheximide). By quantifying protein accumulation in response to inhibitor treatment relative to DMSO, I could determine a protein’s pathway dependencies and whether it is short-lived in one unified proteomics-based assay. The specific patterns of accumulation in response to these inhibitors allowed me to then categorize the short-lived proteome into mechanistically distinct pathways. Of the SLPs that responded to chemical inhibition of the proteostasis network, I found that proteins are either degraded by the proteasome (~70%), lysosome (~25%), or by both pathways (~5%). Within these main degradation categories, proteins can further be subcategorized based on their ubiquitin-dependency (ubiquitin-dependent/CRL-independent, ubiquitin-dependent/CRL-dependent, and ubiquitin-independent). In total, I binned proteins into 9 mechanistically distinct degradation pathways. Interestingly, there were a subset of proteins that could not be binned into any one of the previously mentioned categories. This was either because a protein did not meet statistical cutoffs, or more interestingly, due to potentially being regulated by a protease or even an alternative pathway that may cause the acute depletion of these proteins. For SLPs degraded by the proteasome, I found that virtually all these proteins are degraded in a ubiquitin-dependent manner (~95%) with a slight bias toward non-CRL mediated degradation, which was the predominant mechanism of degradation of most transcription factors. A small subset of proteasomal substrates (~5%) are degraded through ubiquitin-independent mechanisms. These proteins defy the canonical model in which substrates are first ubiquitinated and then delivered to the proteasome via shuttling factors or through direct recognition by one of the proteasome’s many ubiquitin receptors. Notable examples included ODC1—a pioneering case that established the concept of ubiquitin-independent degradation by the 26S proteasome—and MIDN, a recently characterized substrate that has reignited major interest in this degradation pathway. These ubiquitin-independent proteasome substrates exhibit the fastest rate of degradation and are significantly smaller than SLPs that are eliminated by other pathways. Lysosomal proteins are degraded equally in a ubiquitin-dependent (mostly membrane-bound signaling receptors) and ubiquitin-independent manner (mostly structural proteins). Virtually all ubiquitin-dependent lysosomal degradation occurs in a CRL-independent manner. For proteins degraded by both pathways, the vast majority are targeted through a ubiquitin-dependent yet CRL-independent mechanism. These substrates are enriched at the endoplasmic reticulum, reflecting the unique mechanisms of degradation that occur at this organelle. Overall, the findings from this atlas can inform therapeutic strategies to modulate SLP abundance through the proteostasis network. This multiplexed degradomics assay can be extended to include additional selective inhibitors against other key nodes of the proteostasis network, such as heat shock proteins, p97/VCP, and specific proteasome subspecies. Expanding the scope in this way will add granularity and resolution to our degradome map, providing deeper insight into how these short-lived yet important proteins are regulated by the proteostasis network. From my short-lived protein degradation atlas, I found that ASCL1 is degraded by the proteasome in a ubiquitin-dependent, but CRL-independent manner. ASCL1 is normally a lineage-defining master regulator of neuronal cell identity, and it drives most small cell lung cancers (SCLCs). It is one of the deadliest cancers with limited treatment options. In close collaboration with Dr. Keita Masuzawa from Dr. Matthew Oser’s lab, we found that ASCL1 exhibits the “Goldilocks” effect in small cell lung cancers (SCLCs) whereby levels must be precisely maintained—too much or too little of this transcription factor are deleterious to ASCL1-dependent SCLC cells. This suggested that there may be a therapeutic opportunity to modulate ASCL1 levels through enhancing or inhibiting its natural turnover by its endogenous ligase. To mechanistically dissect this regulatory axis, we employed a combination of proteomic, functional genomics, in silico structural analysis, genetic deletion and chemical studies. From this work, we identified that the HECT E3 ligase HUWE1 recognizes the phosphorylated C-terminus of ASCL1 to mediate its destruction. We demonstrate that the C-terminus is necessary for recognition and degradation by HUWE1 and conversely, we show that this C-terminal sequence is sufficient to induce the rapid HUWE1-dependent degradation of GFP upon transplantation. Strikingly, C-terminal mutants demonstrate increased interactions with histones and the epigenetic machinery. This observation was corroborated via ChIP-seq experiments. Interestingly, despite significantly prolonging ASCL1 half-life and accumulating at canonical E-box binding motifs, C-terminal deletion mutants could not rescue survival defects when expressed in an ASCL1 knockout background. These results suggest that the C-terminus may serve a dual role as both a potent degron and a region essential for normal function. We performed a saturating base-editing screen to identify HUWE1 mutations that modulate ASCL1 degradation. Charge-flipping mutations within a helix enhanced degradation, nominating a region we term the bouncer helix. Although unresolved in prior cryo-EM structures and predicted to be disordered, AlphaFold 3 modeling revealed that the bouncer helix packs tightly against a positively charged pocket, termed the donut, where phosphorylated DDIT4 was previously shown to bind. In the apo structure, the bouncer blocks the donut, but structural modeling suggests it is displaced to permit substrate engagement. In an orthogonal base-editing screen, we show that these bouncer helix mutations also lead to enhanced degradation of MYCL, which we recently found to be another HUWE1 substrate based on complementary expression proteomic and functional genomic studies. Cells expressing these bouncer helix mutations exhibited shorter ASCL1 half-lives and enhanced ASCL1 ubiquitination compared to WT cells. Collectively, these results reveal new druggable interaction between the oncogenic transcription factor ASCL1 and its ligase HUWE1. We also propose a generalizable mechanism by which HUWE1 recognizes its substrates. Lastly, we propose that targeting molecules to the donut region of HUWE1 could enable selective gluing and enhanced degradation of ASCL1 as a therapeutic strategy for SCLC. For my third project, I developed a live-cell, targeted covalent drug discovery platform to identify functional binders. Ubiquitin-mediated degradation depends on a dedicated network of proteins that transfer ubiquitin onto substrates for proteasomal turnover. Both catalytic and allosteric cysteines within these proteins are critical for activity, and modulating these sites can fine-tune substrate levels. HUWE1, for example, contains a catalytic cysteine in its active site that is amenable to covalent inhibition via cysteine-targeting electrophiles. Activity-based protein profiling (ABPP) is a powerful, unbiased approach to assess proteome-wide electrophile reactivity. However, ABPP datasets often suffer from missing values, limiting their utility when the goal is to find inhibitors for a specific target. To address this, I developed CysDig, a targeted chemoproteomics platform designed to overcome the missing data problem by enabling drug hunters to easily generate targeted proteomic assays for covalent drug discovery. Through integration with GoDig, a TMT-based targeted proteomics strategy developed in our lab, CysDig allows users to select up to 300 targets of interest. Unlike conventional approaches, CysDig bypasses cysteine enrichment. In doing so, the entire proteome remains available for quantification. Collectively, these innovations enable hybrid targeted assays in which covalent target engagement can be measured in parallel with changes in substrate abundance. This makes CysDig particularly well-suited for discovering small molecule binders against the proteostasis network, where simultaneous assessment of compound engagement with UPS proteins and corresponding substrate abundance changes is especially informative at the screening level. Using CysDig, I identified 31 covalent inhibitors that engage the HUWE1 active site. I prioritized CL129 for follow-up due to its relative specificity for HUWE1 over other HECT E3 ligase family members, UBE3A and UBE3C. Direct detection of the covalent adduct on HUWE1 and covalent docking studies confirmed target engagement. In cells, CL129 treatment caused a dose-dependent accumulation of known HUWE1 substrates—DDIT4, MCL1, SCNM1, and C16orf72/HAPSTR1. Importantly, in SCLC lines, CL129 increased ASCL1 levels at baseline and extended its half-life in a HUWE1-dependent manner. Taken together, this work provides the research community with three key contributions. First, my degradome atlas maps the mechanisms of short-lived protein degradation for hundreds of proteins, offering new opportunities to further dissect the exact mechanisms of degradation and to use that understanding to inform therapeutic discovery. Second, in collaboration with Dr. Keita Masuzawa from Dr. Matt Oser’s lab, we decoded how ASCL1 is degraded by its endogenous ligase, HUWE1. Zooming out, we developed a generalizable framework for translating the initial mechanistic insights provided by my degradome atlas into a detailed understanding of how exactly a protein is regulated. We believe that the experiments outlined in this chapter can be generalized to any protein found to be regulated by the proteostasis network. Third, the CysDig platform enables targeted screening of cysteine-reactive compounds against the UPS. Its utility will scale with our increasing knowledge of how proteins are degraded by the proteostasis machinery. Collectively, my thesis presents a blueprint to decode endogenous degradation pathways and contributes a new drug discovery platform that is specifically tailored for proteostasis-centric drug discovery.