Biology subjects

Schulze, T. K.

Publications and source records attributed to Schulze, T. K..

5 recordsLinked to original sources

bioRxiv2025

Supervised learning of protein variant effects across large-scale mutagenesis datasets

The increasing availability of data from multiplexed assays of variant effects (MAVEs) enables supervised model training against large quantities of experimental data to learn sequence-function relationships. Variant effect scores from MAVEs can, however, be influenced by the experimental method to create experiment-to-experiment differences in the mapping from molecular level-variant effects to MAVE readout, which presents a challenge for supervised learning across datasets. We here propose a framework for performing supervised learning with MAVE data that takes the influence of the experimental protocol into account, thus enabling variant effects to be learned across datasets produced in independent experiments. We apply the framework to train a model against variant effect scores collected with VAMP-seq, a MAVE technique that quantifies the steady-state cellular abundance of protein variants. We show that mapping variant abundance to VAMP-seq readout in a dataset-specific manner during model training improves the learned abundance model and moreover allows the learned model to predict variant effects on an interpretable scale. Our work highlights the importance of validating MAVE results with low-throughput methods to facilitate MAVE score interpretation and supervised model training.

biophysics↗

bioRxiv2024

Effects of residue substitutions on the cellular abundance of proteins

Multiplexed assays of variant effects (MAVEs) make it possible to measure the functional impact of all possible single amino acid residue substitutions in a protein in a single experiment. Combination of variant effect data from several such experiments provides the opportunity to conduct large-scale analyses of variant effect scores measured across proteins, but can be complicated by variations in the phenotypes that are probed across experiments. Thus, using variant effect datasets obtained with similar MAVE techniques can help reveal general rules governing the effects of amino acid variation for a single molecular phenotype. In this work, we accordingly combined data from six individual variant abundance by massively parallel sequencing (VAMP-seq) experiments and analysed a total of 31,614 variant effect scores reporting solely on the impact of single amino acid residue substitutions on the cellular abundance of proteins. Using our combined variant effect dataset, we derived and analysed a collection of amino acid substitution matrices describing the average impact on cellular abundance of all residue substitution types in different structural environments. We found that the substitution matrices predict the cellular abundance of protein variants with surprisingly high accuracy when given structural information only in the form of whether a residue is buried or exposed. We thus propose our substitution matrix-based predictions as strong baselines for future abundance model development.

biophysics↗

bioRxiv2023

Deep mutational scanning reveals a tight correlation between protein degradation and toxicity of thousands of non-native aspartoacylase protein variants

When the structural stability of a protein is compromised, the protein may form non-native interactions with other cell proteins and thus becomes a hazard to the cell. To mitigate this danger, destabilized proteins are targeted by the cellular protein quality control (PQC) network, which either corrects the folding defect or targets the protein for degradation. However, the details of how the protein folding and degradation systems collaborate to combat potentially toxic non-native proteins are unknown. To address this issue, we performed systematic studies on destabilized variants of the cytosolic aspartoacylase, ASPA, where loss-of-function variants are linked to Canavans disease, an autosomal recessive and lethal neurological disorder, characterized by the spongy degeneration of the white matter in the brain. Using Variant Abundance by Massively Parallel sequencing (VAMP-seq), we determined the abundance of 6152 out of the 6260 ([~]98%) possible single-site missense and nonsense ASPA variants in cultured human cells. The majority of the low abundance ASPA variants are degraded through the ubiquitin-proteasome system (UPS) and become toxic upon prolonged expression. Variant cellular abundance data correlates with predicted thermodynamic stability, evolutionary conservation, and separates most known disease-linked variants from benign variants. Systematic mapping of degradation signals (degrons) shows that inherent primary degrons in ASPA are located in buried regions, and reveals that the wild-type ASPA C-terminal region functions as a degron. Collectively, our data can be used to interpret Canavans disease variants and also offer mechanistic insight into how ASPA missense variants are targeted by the PQC system. These are essential steps towards future implementation of precision medicine for Canavans disease.

molecular biology↗

bioRxiv2023

Characterizing glucokinase variant mechanisms using a multiplexed abundance assay

Amino acid substitutions can perturb protein activity in multiple ways. Understanding their mechanistic basis may pinpoint how residues contribute to protein function. Here, we characterize the mechanisms of human glucokinase (GCK) variants, building on our previous comprehensive study on GCK variant activity. We assayed the abundance of 95% of GCK missense and nonsense variants, and found that 43% of hypoactive variants have a decreased cellular abundance. By combining our abundance scores with predictions of protein thermodynamic stability, we identify residues important for GCK metabolic stability and conformational dynamics. These residues could be targeted to modulate GCK activity, and thereby affect glucose homeostasis.

biochemistry↗

bioRxiv2021

Accurate model of liquid-liquid phase behaviour of intrinsically-disordered proteins from data-driven optimization of single-chain properties

Many intrinsically disordered proteins (IDPs) may undergo liquidliquid phase separation (LLPS) and participate in the formation of membraneless organelles in the cell, thereby contributing to the regulation and compartmentalisation of intracellular biochemical reactions. The phase behaviour of IDPs is sequence-dependent, and its investigation through molecular simulations requires protein models that combine computational efficiency with an accurate description of intra- and intermolecular interactions. We developed a general coarse-grained model of IDPs, with residue-level detail, based on an extensive set of experimental data on single-chain properties. Ensemble-averaged experimental observables are predicted from molecular simulations, and a data-driven parameter-learning procedure is used to identify the residue-specific model parameters that minimize the discrepancy between predictions and experiments. The model accurately reproduces the experimentally observed conformational propensities of a set of IDPs. Through two-body as well as large-scale molecular simulations, we show that the optimization of the intramolecular interactions results in improved predictions of protein self-association and LLPS.

biophysics↗