Biology subjects

Hahn, W.

Publications and source records attributed to Hahn, W..

11 recordsLinked to original sources

bioRxiv2026

A Multiclass Mechanistic Model of Vaccine-Induced Humoral Response

Vaccine-induced humoral immunity emerges from complex interactions among competing B-cell populations and antibody production in response to antigen exposure, yet the mechanisms governing these interactions remain only partially understood. We developed a mechanistic mathematical model that integrates vaccine regimen features, immunogen kinetics, multiclass B-cell dynamics, and germinal center processes. The model was calibrated to data from the G001 HIV vaccine trial evaluating eOD-GT8, a germline-targeting immunogen designed to activate VRC01-class precursor B cells capable of developing into broadly neutralizing antibody-producing cells. The model reproduced key features of the observed B-cell and antibody responses, enabled estimation of class-specific activation and competition parameters, and supported investigation of the identifiability of competition parameters. Activation rates varied across B-cell classes, with memory B cells exhibiting substantially higher activation rates than naive B cells. Importantly, competition was stronger within than between classes, with the strongest competition occurring among VRC01-class B cells. These findings suggest that class-specific differences in activation and competition shape the magnitude and composition of vaccine-induced humoral response and may influence the expansion of broadly neutralizing antibody precursors. More broadly, the model provides a quantitative framework for dissecting multiclass B-cell competition and antibody responses, and for evaluating how vaccination strategies shape humoral immunity.

immunology↗

bioRxiv2025

O-mannosylation and protein maturation check-points represent therapeutic opportunities in BRAF fusion protein oncogenesis

Fusions between protein-coding genes are common oncogenic drivers across cancers, typically pairing a proto-oncogene with partner that does not independently drive cancer. In all therapeutically actionable fusions, the proto-oncogene is the drug target, the contributions to oncogenicity of the fusion partner have largely been ignored. We studied the role of BRAF fusion partners and found that they are necessary for transformation. In the setting of KIAA1549::BRAF, the most common fusion protein across brain tumors, we found that KIAA1549 is necessary for the oncogenicity of KIAA1549::BRAF and engenders a striking and specific dependency on the protein O-mannosyltransferase complex (POMT1/2). Specifically, we show that genetic silencing or pharmacologic inhibition of the protein O-mannosyltransferase complex (POMT1/2) reverses fusion-induced transformation, thereby representing a novel and MAPK independent therapeutic target. Furthermore, POMT1/2 is required to glycosylate and enable maturation of the K::B fusion protein. These findings represent a proof-of-concept for targeting the partners in oncogenic fusions as a potential cancer therapeutic strategy.

cancer biology↗

bioRxiv2022

Developmental reprogramming mediates aberrant cell state plasticity in colorectal cancer initiation

Cell state (phenotypic) plasticity is a carefully regulated feature of adult epithelial cells that enables adaptive responses to injury, inflammation, and other forms of stress. Aberrant expansion of the normally restricted capability for cell state plasticity to escape terminal differentiation is a critical aspect of neoplasia. The nongenetic factors and specific programs that mediate aberrant cell state plasticity and impaired differentiation require deeper characterization to understand this elusive aspect of cancer pathogenesis. Using genetically engineered and carcinogen-induced murine models of intestinal neoplasia, we demonstrate that impaired differentiation is a conserved event preceding cancer development. Single cell RNA-sequencing (scRNA-seq) of neoplastic intestinal lesions from both mouse models and a patient with familial adenomatous polyposis revealed that cancer initiates by adopting an aberrant transcriptional state characterized by nonoverlapping expression of a regenerative pathway, marked by Ly6a (Sca-1), and a fetal intestinal program, positive for Tacstd2 (Trop2). Genetic inactivation of Sox9 prevented adenoma formation in ApcKO mice, obstructed emergence of aberrant regenerative and fetal intestinal programs, and restored multi-lineage differentiation by scRNA-seq. Expanded chromatin accessibility at regeneration and fetal genes upon Apc inactivation was reduced by concomitant Sox9 suppression. These studies indicate that aberrant cell state plasticity mediated by unabated regenerative activity and developmental reprogramming precedes cancer development.

cancer biology↗

bioRxiv2022

Partial gene suppression improves identification of cancer vulnerabilities when CRISPR-Cas9 knockout is pan-lethal

Hundreds of genome-wide loss-of-function screens have been performed, as part of efforts such as The Cancer Dependency Map, to create a catalog of genetic dependencies in a diverse set of cancer contexts. In recent years, large-scale screening efforts have shifted perturbation technology from RNAi to CRISPR-Cas9, due to the superior efficacy and specificity of CRISPR-Cas9-mediated approaches. However, questions remain about the extent to which partial suppression of gene targets could result in selective dependency across cell lines, potentially revealing a larger set of targetable cancer vulnerabilities than can be identified using CRISPR knockout alone. Here, we use CRISPR-Cas9 and RNAi screening data for more than 400 shared cell lines to represent knockout and partial suppression genetic perturbation modalities and evaluate the utility of each for therapeutic target discovery and the inference of gene function. We find that CRISPR screens identify more dependencies, and yield more accurate predictive models and co-dependency relationships overall. However, RNAi outperforms CRISPR in identifying associations (omics, drug, co-dependencies) with genes that are common dependencies for most cell lines (pan-dependencies). As pan-dependencies occur frequently in the CRISPR dataset (~2,000 genes), using results from both RNAi and CRISPR analyses facilitates the discovery of predictive models and associated co-dependencies for a wider range of gene targets than could be detected using either dataset alone. These findings can aid in the interpretation of contrasting results obtained from CRISPR and RNAi screens and reinforce the importance of partial gene suppression methods in building a cancer dependency map.

cancer biology↗

bioRxiv2021

Pan-cancer organoid validation of tumor outlier chromosomal amplification events

Somatic copy number gains are pervasive in many cancer types, yet their roles in oncogenesis are often poorly explored. This lack of understanding is in part due to broad extensions of copy gains across cancer genomes spanning large chromosomal regions, obscuring causal driver loci. Here we employed a multi-tissue pan-organoid modeling approach to validate candidate oncogenic loci identified within pan-cancer TCGA data by the overlap of extreme copy number amplifications with extreme expression dysregulation for each gene. The candidate outlier loci nominated by this integrative computational analysis were functionally validated by infecting cancer type-specific barcoded full length cDNA lentiviral libraries into cognate minimally transformed human and mouse organoids bearing initial oncogenic mutations from esophagus, oral cavity, colon, stomach, pancreas and lung. Presumptive amplification oncogenes were identified by barcode enrichment as a proxy for increased proliferation. Iterative analysis validated DYRK2 at 12q15, encoding a serine-threonine kinase, as an amplified head and neck squamous carcinoma oncogene in p53-/- oral mucosal organoids. Similarly, FGF3, amplified at 11q13 in 41% of esophageal squamous carcinomas, was validated in p53-/- esophageal organoids in vitro and in vivo with pharmacologic inhibition by small molecule and soluble receptor FGFR antagonists. Our studies establish the feasibility of pan-organoid contextual modeling of pan-cancer candidate genomic drivers, enabling oncogene discovery and preclinical therapeutic modeling.

cancer biology↗

bioRxiv2021

CREB5 reprograms nuclear interactions to promote resistance to androgen receptor targeting therapies

Metastatic castration resistant prostate cancers (mCRPC) are treated with therapies that antagonize the androgen receptor (AR). Nearly all patients develop resistance to AR-targeted therapies (ART). Our previous work identified CREB5 as an upregulated target gene in human mCRPC that promoted resistance to all clinically-approved ART. The mechanisms by which CREB5 promotes progression of mCRPC or other cancers remains elusive. Integrating ChIP-seq and rapid immunoprecipitation and mass spectroscopy of endogenous proteins (RIME), we report that cells overexpressing CREB5 demonstrate extensive reprogramming of nuclear protein-protein interactions in response to the ART agent enzalutamide. Specifically, CREB5 physically interacts with AR, the pioneering actor FOXA1, and other known co-factors of AR and FOXA1 at transcription regulatory elements recently found to be active in mCRPC patients. We identified a subset of CREB5/FOXA1 co-interacting nuclear factors that have critical functions for AR transcription (GRHL2, HOXB13) while others (TBX3, NFIC) regulated cell viability and ART resistance and were amplified or overexpressed in mCRPC. Upon examining the nuclear protein interactions and the impact of CREB5 expression on the mCRPC patient transcriptome, we found CREB5 was associated with TGF{beta} and Wnt signaling and epithelial to mesenchymal transitions, implicating these pathways in ART resistance. Overall, these observations define the molecular interactions among CREB5, FOXA1, and pathways that promote ART resistance.

cancer biology↗

bioRxiv2020

Massively parallel phenotyping of variant impact in cancer with Perturb-seq reveals a shift in the spectrum of cell states induced by somatic mutations

Genome sequencing studies have identified millions of somatic variants in cancer, but their phenotypic impact remains challenging to predict. Current experimental approaches to distinguish between functionally impactful and neutral variants require customized phenotypic assays that often report on average effects, and are not easily scaled. Here, we develop a generalizable, high-dimensional, and scalable approach to functionally assess variant impact in single cells by pooled Perturb-seq. Specifically, we assessed the impact of 200 TP53 and KRAS variants in >300,000 single lung cancer cells, and used the profiles to categorize variants into phenotypic subsets to distinguish gain-of-function, loss-of-function and dominant negative variants, which we validated by comparison to orthogonal assays. Surprisingly, KRAS variants did not merely fit into discrete functional categories, but rather spanned a continuum of gain-of-function phenotypes driven by quantitative shifts in cell composition at the single cell level. We further discovered novel gain-of-function KRAS variants whose impact could not have been predicted solely by their occurrence in patient samples. Our work provides a scalable, gene-agnostic method for coding variant impact phenotyping, which can be applied in cancer and other diseases driven by somatic or germline coding mutations.

genomics↗

bioRxiv2020

Predicting Residues Involved in Anti-DNA Autoantibodies with Limited Neural Networks

Computer-aided rational vaccine design (RVD) and synthetic pharmacology are rapidly developing fields that leverage existing datasets for developing compounds of interest. Computational proteomics utilizes algorithms and models to probe proteins for functional prediction. A potentially strong target for such a computational approach is autoimmune antibodies which are the result of broken tolerance in the immune system where it cannot distinguish "self" from "non-self" resulting in attack of its own structures (proteins and DNA, mainly). The information on structure, function and pathogenicity of autoantibodies may assist in engineering RVD against autoimmune diseases. Current computational approaches exploit large datasets curated with extensive domain knowledge, most of which include the need for many computational resources and have been applied indirectly to problems of interest for DNA, RNA, and monomer protein binding. Here, we present a novel method for discovering potential binding sites. We employed long short-term memory (LSTM) models trained on FASTA primary sequences directly to predict protein binding in DNA-binding hydrolytic antibodies (abzymes). We also employed CNN models applied to the same dataset. While the CNN model outperformed the LSTM on the primary task of binding prediction, analysis of internal model representations of both models showed that the LSTM models highlighted sub-sequences that were more strongly correlated with sites known to be involved in binding. These results demonstrate that analysis of internal processes of recurrent neural network models may serve as a powerful tool for primary sequence analysis.

bioinformatics↗

bioRxiv2019

Self-Organizing Map Methodology for Sorting Differential Expression Data of MMP-9 Inhibition

An unsupervised machine-learning model, based on a self-organizing map (SOM), was employed to extract suggested target genes from DESeq2 differential expression analysis data. Such methodology was tested on matrixmetalloproteinase 9 (MMP-9) inhibitors. The model generated information on several novel gene hits that may be regulated by MMP-9, suggesting the self-organizing map method may serve as a useful analytic tool in degradomics research for further differential expression data analysis. Original data was generated from a previous study, which consisted of quantitative measures in changes of levels of gene expression from 32,000 genes in four different conditions of stimulated T-cells treated with an MMP-9 inhibitor. Since intracellular target of MMP-9 are not yet well characterized, the functional enrichment analysis program, WebGestalt, was used for validation of the SOM identified regulated genes. The proposed data analysis method indicated MMP-9s prominent role in biological regulatory and metabolic processes as major categories of regulation of the predicted genes. Both fields suggest extensive intracellular targets for MMP-9-triggered regulation, which are new interests in MMP-9 research. The methodology presented here is useful for similar knowledge and discovery from quantitative datasets and a proposed extension of DESeq2 or similar data analysis.

bioinformatics↗

bioRxiv2018

Target Binding and Sequence Prediction With LSTMs

Deep recurrent neural networks (DRNNs) have recently demonstrated strong performance in sequential data analysis, such as natural language processing. These capabilities make them a promising tool for inferential analysis of sequentially structured bioinformatics data as well. Here, we assessed the ability of Long Short-Term Memory (LSTM) networks, a class of DRNNs, to predict properties of proteins based on their primary structures. The proposed architecture is trained and tested on two different datasets to predict whether a given sequence falls into a certain class or not. The first dataset, directly imported from Uniprot, was used to train the network on whether a given protein contained or did not contain a conserved sequence (homeodomain), and the second dataset, derived by literature mining, was used to train a network on whether a given protein binds or doesnt bind to Artemisinin, a drug typically used to treat malaria. In each case, the model was able to differentiate between the two different classes of sequences it was given with high accuracy, illustrating successful learning and generalization. Upon completion of training, an ROC curve was created using the homeodomain and artemisinin validation datasets. The AUC of these datasets was 0.80 and 0.87 respectively, further indicating the models effectiveness. Furthermore, using these trained models, it was possible to derive a protocol for sequence detection of homeodomain and binding motif, which are well-documented in literature, and a known Artemisinin binding site, respectively [1-3]. Along with these contributions, we developed a python API to directly connect to Uniprot data sourcing, train deep neural networks on this primary sequence data using TensorFlow, and uniquely visualize the results of this analysis. Such an approach has the potential to drastically increase accuracy and reduce computational time and, current major limitations in informatics, from inquiry to discovery in protein function research.

bioinformatics↗

bioRxiv2018

Renal medullary carcinomas depend upon SMARCB1 loss and are sensitive to proteasome inhibition

Renal medullary carcinoma (RMC) is a rare and deadly kidney cancer in patients of African descent with sickle cell trait. Through direct-to-patient outreach, we developed genomically faithful patient-derived models of RMC. Using whole genome sequencing, we identified intronic fusion events in one SMARCB1 allele with concurrent loss of the other allele, confirming that SMARCB1 loss occurs in RMC. Biochemical and functional characterization of these RMC models revealed that RMC depends on the loss of SMARCB1 for survival and functionally resemble other cancers that harbor loss of SMARCB1, such as malignant rhabdoid tumors or atypical teratoid rhabdoid tumors. We performed RNAi and CRISPR-Cas9 loss of function genetic screens and a small-molecule screen and identified UBE2C as an essential gene in SMARCB1 deficient cancers. We found that the ubiquitin-proteasome pathway was essential for the survival of SMARCB1 deficient cancers in vitro and in vivo. Genetic or pharmacologic inhibition of this pathway leads to G2/M arrest due to constitutive accumulation of cyclin B1. These observations identify a synthetic lethal relationship that may serve as a therapeutic approach for patients with SMARCB1 deficient cancers.

cancer biology↗