Biology subjects

Reddel, R.

Publications and source records attributed to Reddel, R..

4 recordsLinked to original sources

bioRxiv2025

Machine Learning-Driven Discovery of Synergistic Protein Interactions Identifies ATL3 as a Putative Biomarker for Cancer Drug Response

BackgroundResistance to targeted molecular therapies--both primary and acquired--remains a major obstacle to effective cancer treatment. Despite extensive research, the molecular determinants of treatment resistance are still incompletely understood, underscoring the need to identify robust resistance drivers to improve therapeutic outcomes. Recently, the ProCan and Wellcome Sanger Institute released the worlds largest pan-cancer proteomic dataset to date, comprising 949 cancer cell lines treated with 625 anti-cancer agents. Despite progress in identifying single protein markers, scalable methods for detecting synergistic protein interactions driving drug susceptibility remain limited. Machine learning models such as random forests may offer a robust framework for capturing non-linear protein interactions across diverse cancer types. ResultsOur study presents synerOmics, a scalable framework for identifying putative synergistic protein interactions by leveraging parent-child co-occurrences in random forest regression trees to reduce the interaction search space. We validated our approach using simulated data and two independent cancer proteomic datasets, identifying both pan-cancer and breast cancer-specific markers associated with drug susceptibility. Synergistic interactions that consistently replicated across datasets were enriched for endoplasmic reticulum stress pathways. Notably, shared targets included established sensitivity markers for tyrosine kinase inhibitors (TKIs) and revealed a resistance-associated network centred on ATL3, which demonstrated both prognostic and predictive relevance. ConclusionsApplying synerOmics to the largest cancer cell line proteomic dataset to date, we identified ATL3 as a strong candidate biomarker for lapatinib resistance in breast cancer, with superior predictive performance compared to the current gold standard sensitivity biomarker, ERBB2. HighlightsO_LIWe present a machine learning framework for drug response biomarker discovery, leveraging the largest proteomic cancer cell line dataset, ProCan-DepMapSanger C_LIO_LIATL3 is proposed as a candidate predictive biomarker for lapatinib resistance in breast cancer C_LIO_LIElevated expression of a novel resistance-conferring subcluster centered on ATL3 is associated with poor prognosis across 11 TCGA cancer types C_LIO_LIBaseline drug resistance may be broadly driven by stress pathways, with ER stress and protein ATL3 implicated in ER-phagy in response to TKIs. C_LI

cancer biology↗

bioRxiv2024

Blocking MIF secretion enhances CAR T-cell efficacy against neuroblastoma

While chimeric antigen receptor (CAR) T-cell therapies are showing highly promising first results in neuroblastoma, immunosuppressive tumor microenvironments (TME) limit T cell persistence and durable clinical efficacy. To improve CAR T-cell efficacy further, we applied a multi-omics approach including single-cell RNA sequencing and proteomics, which identified 13 targetable immunosuppressive factors in neuroblastoma. Of these, macrophage migration inhibitory factor (MIF) and midkine (MDK) were validated across multiple published RNA datasets. Moreover, they were secreted in high abundance by neuroblastoma tumoroids. Functional validation experiments revealed MIF as a potent inhibitor of CAR T-cells, in vitro and in vivo. Degradation of MIF by PROTAC technology significantly enhanced CAR T-cell activation targeting GPC2 and B7-H3, providing a potential intervention against MIF. By defining the immunosuppressive effects of neuroblastomas TME on CAR T-cell efficacy, particularly the pivotal role of MIF, we provide a therapeutic strategy for improving adoptive cell therapies for this pediatric malignancy.

cancer biology↗

bioRxiv2023

Heat n Beat: A universal high-throughput end-to-end proteomics sample processing platform in under an hour

Proteomic analysis by mass spectrometry (MS) of small ([≤]2 mg) solid tissue samples from diverse formats requires high throughput and comprehensive proteome coverage. We developed a near universal, rapid and robust protocol for sample preparation, suitable for high-throughput projects that encompass most cell or tissue types. This end-to-end workflow extends from original sample to loading the mass spectrometer and is centred on a one tube homogenisation and digestion method called Heat n Beat (HnB). It is applicable to most tissues, regardless of how they were fixed or embedded. Sample preparation was divided to separate challenges. The initial sample washing, and final peptide clean-up steps were adapted to three tissue sources: fresh frozen (FF), optimal cutting temperature (OCT) compound embedded (FF-OCT), and formalin-fixed paraffin-embedded (FFPE). Thirdly, for core processing, tissue disruption and lysis were decreased to a 7 min heat and homogenisation treatment, and reduction, alkylation and proteolysis were optimised into a single step. The refinements produced near doubled peptide yield, delivered consistently high digestion efficiency of 85-90%, and required only 38 minutes for core processing in a single tube, with total processing time being 53-63 minutes. The robustness of HnB was demonstrated on six organ types, a cell line and a cancer biopsy. Its suitability for high throughput applications was demonstrated on a set of 1,171 FF-OCT human cancer biopsies, which were processed for end-to-end completion in 92 hours, producing highly consistent peptide yield and quality for over 3,513 MS runs. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=152 SRC="FIGDIR/small/559846v2_ufig1.gif" ALT="Figure 1"> View larger version (34K): org.highwire.dtl.DTLVardef@4b8399org.highwire.dtl.DTLVardef@1acc573org.highwire.dtl.DTLVardef@1d7155eorg.highwire.dtl.DTLVardef@1bbef10_HPS_FORMAT_FIGEXP M_FIG C_FIG

biochemistry↗

bioRxiv2021

CRISPR screens identify gene targets and drug repositioning opportunities at breast cancer risk loci

Genome-wide association studies (GWAS) have identified >200 loci associated with breast cancer (BC) risk. The majority of candidate causal variants (CCVs) are in non-coding regions and are likely to modulate cancer risk by regulating gene expression. We recently developed a scoring system, INQUISIT, to predict candidate risk genes at BC-risk loci. Here, we used pooled CRISPR activation and suppression screens to validate INQUISIT predictions, and to define the cancer phenotypes they mediate. We measured proliferation in 2D, 3D, and in immune-deficient mice, as well as the effect on the DNA damage response. We performed 60 CRISPR screens and identified 21 high-confidence INQUISIT predictions that mediate a cancer phenotype. We validated the direct regulation of a subset of genes by BC-risk variants using HiCHIP and CRISPRqtl. Furthermore, we show the utility of expression profiling for drug repurposing against these targets. We provide a platform for identifying gene targets of risk variants, and lay a blueprint of interventions for BC risk reduction and treatment.

cancer biology↗