Biology subjects

Crook, O. M.

Publications and source records attributed to Crook, O. M..

11 recordsLinked to original sources

bioRxiv2026

Population-scale subcellular proteomics reveals intracellular remodelling across the Alzheimer's disease-resilience spectrum

Proteome-wide analyses of human tissue have transformed our understanding of disease, but provide limited insight into protein localisation, a functionally informative dimension of the proteome. In Alzheimer's disease, amyloid-{beta} and tau exhibit aberrant localisation, yet whether spatial reorganisation extends proteome-wide has remained inaccessible to abundance-based proteomics. Here, we develop comparative subcellular proteomics applied to dorsolateral prefrontal cortex from 75 individuals spanning the Alzheimer's disease-resilience spectrum, modelling protein localisation across disease. We identify 217 disease-associated localisation shifts enriched for endolysosomal function, intracellular trafficking, and RNA processing, and resolve tau proteoforms within insoluble aggregates. Our strongest localisation candidates show only modest differences in whole-tissue abundance, highlighting disease biology inaccessible to conventional proteomics. We validate co-localisation of CSNK1A1 with pathological tau and identify an unexpected neuronal localisation pattern for SCAI, a cancer-associated protein not previously characterised in human brain, highlighting the discovery potential of subcellular proteomics in tissue.

neuroscience↗

bioRxiv2025

Decode-gLM: Tools to Interpret, Audit, and Steer GenomicLanguage Models

While genomic language models are enabling the de novo design of entire genomes, they remain challenging to interpret, limiting their trustworthiness. Here, we show that sparse autoencoders (SAEs) trained on Nucleotide Transformer activations decompose hidden representations into interpretable biological features without supervision. Across layers and model sizes, SAEs identified over 60 diverse functional annotations encoded in the models activations. This included viral regulatory elements such as the CMV enhancer, despite viral genomes being excluded from training data. Tracing this signal revealed contamination in reference databases, demonstrating that interpretability methods can audit training data and identify hidden data leakage. We then show that Meta-SAEs, trained on the decoder weights of another SAE, can identify conceptual hierarchies encoded in the model, including a more abstract feature related to multiple HIV annotations. We confirmed that the features identified by our SAEs were learned during pretraining through probing a randomly initialised model. Finally, we demonstrate that our SAEs allow us to steer model predictions in biologically meaningful ways, showing that we can use an antibiotic-resistance SAE-feature to steer the model toward the A1408G aminoglycoside-resistance mutation in the ribosomal gene 16S rRNA. Together, these results establish SAEs as a method for both discovery and auditing, providing a toolkit for interpretable and trustworthy genomic foundation models. Readers can explore our findings at https://interpretglm.netlify.app/.

genomics↗

bioRxiv2024

Global proteomics indicates subcellular-specific anti-ferroptotic responses to ionizing radiation

Cells have many protective mechanisms against background levels of ionizing radiation (IR) orchestrated by molecular changes in expression, post-translation modifications and subcellular localization. Radiotherapeutic treatment in oncology attempts to overwhelm such mechanisms, but radio-resistance is an ongoing challenge. Here, global subcellular proteomics combined with Bayesian modelling identified 544 differentially localized proteins in A549 cells upon 6 Gy x-ray exposure, revealing subcellular-specific changes of proteins involved in ferroptosis, an iron-dependent cell death, suggestive of potential radio-resistance mechanisms. These observations were independent of expression changes, emphasizing the utility of global subcellular proteomics and the promising prospect of ferroptosis-inducing therapies for combatting radioresistance.

cell biology↗

bioRxiv2024

Semi-supervised Bayesian integration of multiple spatial proteomics datasets

The subcellular localisation of proteins is a key determinant of their function. High-throughput analyses of these localisations can be performed using mass spectrometry-based spatial proteomics, which enables us to examine the localisation and relocalisation of proteins. Furthermore, complementary data sources can provide additional sources of functional or localisation information. Examples include protein annotations and other high-throughput omic assays. Integrating these modalities can provide new insights as well as additional confidence in results, but existing approaches for integrative analyses of spatial proteomics datasets are limited in the types of data they can integrate and do not quantify uncertainty. Here we propose a semi-supervised Bayesian approach to integrate spatial proteomics datasets with other data sources, to improve the inference of protein sub-cellular localisation. We demonstrate our approach outperforms other transfer-learning methods and has greater flexibility in the data it can model. To demonstrate the flexibility of our approach, we apply our method to integrate spatial proteomics data generated for the parasite Toxoplasma gondii with time-course gene expression data generated over its cell cycle. Our findings suggest that proteins linked to invasion organelles are associated with expression programs that peak at the end of the first cell-cycle. Furthermore, this integrative analysis divides the dense granule proteins into heterogeneous populations suggestive of potentially different functions. Our method is disseminated via the mdir R package available on the lead authors Github. Author summaryProteins are located in subcellular environments to ensure that they are near their interaction partners and occur in the correct biochemical environment to function. Where a protein is located can be determined from a number of data sources. To integrate diverse datasets together we develop an integrative Bayesian model to combine the information from several datasets in a principled manner. We learn how similar the dataset are as part of the modelling process and demonstrate the benefits of integrating mass-spectrometry based spatial proteomics data with timecourse gene-expression datasets.

systems biology↗

bioRxiv2022

A system-wide quantitative map of RNA and protein subcellular localisation dynamics

Existing methods to analyse RNA localisation are constrained to specific RNAs or subcellular niches, precluding the cell-wide mapping of RNA. We present Localisation of RNA (LoRNA), which maps, at once, RNAs to membranous (nucleus, ER and mitochondria) and membraneless compartments (cytosol, nucleolus and phase-separated granules). Simultaneous interrogation of all RNA locations allows the system-wide quantification of RNA proportional distribution and the comprehensive analysis of RNA subcellular dynamics. Moreover, we have re-engineered the LOPIT (Localisation Of Proteins by Isotope Tagging) method, enabling integration with LoRNA, to jointly map RNA and protein subcellular localisation. Applying this framework, we obtain a global re-localisation map for 31839 transcripts and 5314 proteins during the unfolded protein response, uncovering that ER-localised transcripts are more efficiently recruited to stress granules than cytosolic RNAs, and revealing eIF3d is key to sustain cytoskeletal function. Overall, we provide the most exhaustive map to date of RNA and protein subcellular dynamics.

cell biology↗

bioRxiv2021

Empirical Bayes functional models for hydrogen deuterium exchange mass spectrometry

A proteins structure is state-specific and a key determinant of its function. Proteins can undergo subtle structural changes when binding to another protein, small molecule or as a result of environmental perturbations. Hydrogen deuterium exchange mass spectrometry (HDX-MS) is a technique to explore differential protein structure by examining the rate of deuterium incorporation for specific peptides. This rate will be altered upon structural perturbation and detecting significant changes to this rate requires a statistical test. To determine rates of incorporation, HDX-MS measurements are frequently made over a time course. However, current statistical testing procedures ignore the correlations in the temporal dimension of the data. Using tools from functional data analysis, we develop a testing procedure that explicitly incorporates a physical model of hydrogen deuterium exchange. To further improve statistical power, we develop an empirical Bayes version of our method, allowing us to borrow information across peptides and stabilise variance estimates for low sample sizes. Our approach has increased power, reduces false positives and improves interpretation over linear model-based approaches. Due to the improved flexibility of our method, we can apply it to a multi-antibody epitope-mapping experiment where current approaches are inapplicable due insufficient flexibility. Hence, our approach allows HDX-MS to be applied in more experimental scenarios and reduces the burden on experimentalists to produce excessive replicates. Our approach is implemented in the R-package "hdxstats": https://github.com/ococrook/hdxstats.

bioinformatics↗

bioRxiv2021

Prior Signal Acquisition Software Versions for Orbitrap Underestimate Low Isobaric Mass Tag Intensities, Without Detriment to Differential Abundance Experiments

Tandem mass tags (TMT) enable simple and accurate quantitative proteomics for multiplexed samples by relative quantification of tag reporter ions. Orbitrap quantification of reporter ions has been associated with a characteristic notch region in intensity distribution, within which few reporter intensities are recorded. This has been resolved in version 3 of the instrument acquisition software, Tune. However, 53 % of Orbitrap Fusion, Lumos or Eclipse submissions to PRIDE were generated using prior software versions. To quantify the impact of the notch on existing quantitative proteomics data, we generated a mixed species benchmark and acquired quantitative data using Tune versions 2 and 3. Sub-notch intensities are systemically underestimated with Tune version 2, leading to over-estimation of the true differences in intensities between samples. However, when summarising reporter ion intensities to higher level features, such as peptides and proteins, few features are significantly affected. Targeted removal of spectra with reporter ion intensities below the notch is not beneficial for differential peptide or protein testing. Overall, we find the systematic quantification bias associated with the notch is not detrimental for a typical proteomics experiment.

bioinformatics↗

bioRxiv2020

A subcellular atlas of Toxoplasma reveals the functional context of the proteome

Apicomplexan parasites cause major human disease and food insecurity. They owe their considerable success to novel, highly specialized cell compartments and structures. These adaptations drive their recognition and non-destructive penetration of hosts cells and the elaborate reengineering of these cells to promote growth, dissemination, and the countering of host defenses. The evolution of unique apicomplexan cellular compartments is concomitant with vast proteomic novelty that defines these new cell organizations and their functions. Consequently, half of apicomplexan proteins are unique and uncharacterized, and these cells are, therefore, very poorly understood. Here, we determine the steady-state subcellular location of thousands of proteins simultaneously within the globally prevalent apicomplexan parasite Toxoplasma gondii. This provides unprecedented comprehensive molecular definition to these cells and their novel compartments, and these data reveal the spatial organizations of protein expression and function, adaptation to hosts, and the underlying evolutionary trajectories of these pathogens.

cell biology↗

bioRxiv2019

Determining the content of vesicles captured by golgin tethers using LOPIT-DC

The internal organisation of the cell depends on tethers at destination organelles to selectively capture incoming transport vesicles to facilitate SNARE-mediated fusion. The golgin long coiled-coil proteins function as tethers that contributes to this specificity at the Golgi (1). Golgin-97, golgin-245 and GCC88 golgins of the trans-Golgi capture vesicles derived from endosomes, which serve to recycle the critical Golgi machinery required to deliver lysosomal hydrolases and to maintain exocytosis. Retrograde trafficking from endosomes to the trans-Golgi network (TGN) is a complex process that involves the sorting of transmembrane cargo proteins into distinct transport vesicles by adaptors from multiple pathways. The content of these distinct vesicles, which golgin they target and the factors that mediate this targeting are not well understood. The major challenges that have limited advances in these areas is the transient nature of vesicle tethering, and the redundancies in their mechanisms that confound experimental dissection. To gain better insight into these problems, we performed organelle proteomics using the Localisation of Organelle Proteins by Isotope Tagging after Differential ultraCentrifugation (LOPIT-DC) method on a system in which an ectopic golgin causes vesicles to accumulate in a tethered state (2). By incorporating Bayesian statistical modelling into our analysis (3), we determined that over 45 transmembrane proteins and 51 peripheral membrane proteins of the endosomal network are on vesicles captured by golgin-97, including known cargo and components of the clathrin/AP-1, retromer-dependent and -independent transport pathways. We also determined a distinct class of vesicles shared by golgin-97, golgin-245 and GCC88 that is enriched in TMEM87A, a multi-pass transmembrane protein of unknown function that has previously been implicated in endosome-to-Golgi retrograde transport (4). Finally, we categorically demonstrate that the vesicles that these golgins capture are retrograde transport vesicles based on the lack of enrichment of lysosomal hydrolases in our LOPIT-DC data, and from correlative light electron tomography images of spherical vesicles captured by golgin-97. Together, our study demonstrates the power of combining LOPIT-DC with Bayesian statistical analysis in interrogating the dynamic spatial movement of proteins in transport vesicles.

cell biology↗

bioRxiv2018

LOPIT-DC: A simpler approach to high-resolution spatial proteomics

Hyperplexed Localisation of Organelle Proteins by Isotope Tagging (hyperLOPIT) is a well-established method for studying protein subcellular localisation in complex biological samples. As a simpler alternative we developed a second workflow named Localisation of Organelle Proteins by Isotope Tagging after Differential ultraCentrifugation (LOPIT-DC) which is faster and less resource-intensive. We present the most comprehensive high-resolution mass spectrometry-based human dataset to date and deliver a flexible set of subcellular proteomics protocols for sample preparation and data analysis. For the first time, we methodically compare these two different mass spectrometry-based spatial proteomics methods within the same study and also apply QSep, the first tool that objectively and robustly quantifies subcellular resolution in spatial proteomics data. Using both approaches we highlight suborganellar resolution and isoform-specific subcellular niches as well as the locations of large protein complexes and proteins involved in signalling pathways which play important roles in cancer and metabolism. Finally, we showcase an extensive analysis of the multilocalising proteome identified via both methods.

biochemistry↗

bioRxiv2018

A Bayesian Mixture Modelling Approach For Spatial Proteomics

AbstractAnalysis of the spatial sub-cellular distribution of proteins is of vital importance to fully understand context specific protein function. Some proteins can be found with a single location within a cell, but up to half of proteins may reside in multiple locations, can dynamically re-localise, or reside within an unknown functional compartment. These considerations lead to uncertainty in associating a protein to a single location. Currently, mass spectrometry (MS) based spatial proteomics relies on supervised machine learning algorithms to assign proteins to sub-cellular locations based on common gradient profiles. However, such methods fail to quantify uncertainty associated with sub-cellular class assignment. Here we reformulate the framework on which we perform statistical analysis. We propose a Bayesian generative classifier based on Gaussian mixture models to assign proteins probabilistically to sub-cellular niches, thus proteins have a probability distribution over sub-cellular locations, with Bayesian computation performed using the expectation-maximisation (EM) algorithm, as well as Markov-chain Monte-Carlo (MCMC). Our methodology allows proteome-wide uncertainty quantification, thus adding a further layer to the analysis of spatial proteomics. Our framework is flexible, allowing many different systems to be analysed and reveals new modelling opportunities for spatial proteomics. We find our methods perform competitively with current state-of-the art machine learning methods, whilst simultaneously providing more information. We highlight several examples where classification based on the support vector machine is unable to make any conclusions, while uncertainty quantification using our approach provides biologically intriguing results. To our knowledge this is the first Bayesian model of MS-based spatial proteomics data.\n\nAuthor summarySub-cellular localisation of proteins provides insights into sub-cellular biological processes. For a protein to carry out its intended function it must be localised to the correct sub-cellular environment, whether that be organelles, vesicles or any sub-cellular niche. Correct sub-cellular localisation ensures the biochemical conditions for the protein to carry out its molecular function are met, as well as being near its intended interaction partners. Therefore, mis-localisation of proteins alters cell biochemistry and can disrupt, for example, signalling pathways or inhibit the trafficking of material around the cell. The sub-cellular distribution of proteins is complicated by proteins that can reside in multiple micro-environments, or those that move dynamically within the cell. Methods that predict protein sub-cellular localisation often fail to quantify the uncertainty that arises from the complex and dynamic nature of the sub-cellular environment. Here we present a Bayesian methodology to analyse protein sub-cellular localisation. We explicitly model our data and use Bayesian inference to quantify uncertainty in our predictions. We find our method is competitive with state-of-the-art machine learning methods and additionally provides uncertainty quantification. We show that, with this additional information, we can make deeper insights into the fundamental biochemistry of the cell.

systems biology↗