Biology subjects

Swanson, K.

Publications and source records attributed to Swanson, K..

At least 19 recordsLinked to original sources

bioRxiv2025

IBD risk locus rs1077773 is a pharmacogenomic eQTL for aryl hydrocarbon receptor activity and modulates immune cell function.

IntroductionThe inflammatory bowel diseases (IBD) Crohns disease (CD) and ulcerative colitis (UC) are disorders that cause chronic inflammation of the gastrointestinal tract. Both genetic and environmental factors contribute to the pathogenesis of IBD. There are currently >200 known genetic susceptibility loci for the development of IBD. The physiological impact of the majority of these loci remain a gap in our knowledge. One such locus is the single nucleotide polymorphism rs1077773, located [~]56kbp downstream from the aryl hydrocarbon receptor (AHR) gene. AHR is a ligand-activated transcription factor that is crucial to maintaining intestinal homeostasis. We hypothesized that rs1077773 enhances AHR activity to regulate mucosal immune response and maintain intestinal homeostasis. MethodsAll study procedures and reagents were approved by the Washington University Institutional Review Board (#202011003). Patient biopsies were collected at Barnes Jewish Hospital and genotyped using the IBD Genetics Consortium custom GSA SNP chip (Broad Institute) followed by imputation using TopMed Imputation Server at University of Michigan. Patient derived organoids (PDOs; N=3 G/G, N=4 G/A, N=5 A/A) were derived and maintained in 3D culture and supplemented with 50% L-WRN conditioned medium with passage every 3-4 days as previously described. PDOs were treated with AHR agonist 6-Formylindolo[3,2-b]carbazole (FICZ) or vehicle for 48h. Expression of AHR and its transcriptional targets Cytochrome P450 1A1 (CYP1A1) and CYP1B1 was assessed by RT-qPCR. Blood was collected from pediatric patients undergoing intestinal resection at St. Louis Childrens Hospital and was genotyped with custom TaqMan SNP assay (N=3 G/G, N=5 G/A). Peripheral blood monocyte-derived macrophages (MDM{Phi}s) were treated with lipopolysaccharide in the presence or absence of AHR ligands FICZ or indole-3-carboxaldehyde for 24h. Cytokine levels in culture supernatant were measured via using the ProcartaPlex human cytokine, chemokine, and growth factor 45-plex (ThermoFisher) on a Luminex FLEXMAP3D instrument. ResultsAHR expression was similar across genotypes and treatments. PDOs homozygous for rs1077773 demonstrate enhanced CYP1A1 expression in response to AHR activation. In PBM{Phi}s, cytokine secretion was stimulated by LPS treatment and was abrogated by FICZ treatment. PBM{Phi}s with rs1077773 alternate allele demonstrated significant reduction in secretion of 17 cytokines and chemokines. ConclusionsThis work demonstrates that rs1077773 is an expression quantitative trait locus (eQTL) for AHR activity and modulates epithelial and immune cell function in vitro. Further mechanistic understanding of this locus and its correlates could improve our understanding of the molecular mechanisms of IBD susceptibility and may lead to novel personalized therapeutic approaches in IBD. SummaryOur work demonstrates rs1077773 alternate allele is associated with enhanced aryl hydrocarbon transcriptional activity in human primary epithelial organoids and reduced lipopolysaccharide-induced inflammatory cytokine production in human peripheral blood monocyte-derived macrophages. Key MessageWe have identified rs1077773 as a pharmacogenomic regulator of inflammatory response in human primary intestinal organoids and peripheral blood monocyte-derived macrophages via AHR activity. Individual genetic variation affecting this pathway may account for differences in response to environmental stimuli and the development and progression of IBD.

cell biology↗

bioRxiv2025

RL-Finetuning of OpenAI o1-mini to Enhance Biomedical Reasoning

Recent breakthroughs in advanced reasoning large language models (LLMs), such as OpenAIs o1, have achieved impressive results in domains like math and coding. However, its not clear how much this type of reasoning helps in solving biomedical problems that involve more domain specialized knowledge and open-ended reasoning. Across two biomedical domains--gene characterization and small molecule property prediction--we find that the commercially available o1-mini model does not consistently outperform non-reasoning LLMs like GPT-4o. This motivated us to explore how much we can improve o1-minis biomedical reasoning through reinforcement learning (RL) finetuning. We show that RL finetuning of o1-mini results in large improvements in performance on gene classification, where it surprisingly outperformed domain-specific state-of-the-art models on some tasks. The results are mixed for small molecule prediction, suggesting that chemical reasoning could be more challenging for LLMs. We conclude with a discussion of the challenges and takeaways from this initial exploration of RL finetuning reasoning models for biomedical tasks.

bioinformatics↗

bioRxiv2025

SyntheMol-RL: a flexible reinforcement learning framework for designing novel and synthesizable antibiotics

The rise of antibiotic-resistant pathogens such as Staphylococcus aureus has created an urgent need for new antibiotics. Generative artificial intelligence (AI) has shown promise in drug discovery, but existing models often fail to propose compounds that are both effective and synthetically tractable. To address these challenges, we introduce SyntheMol-RL, a reinforcement learning-based generative model that can rapidly design synthetically accessible small molecule drug candidates from a massive chemical space of 46 billion compounds. SyntheMol-RL improves upon our prior Monte Carlo Tree Search (MCTS)-based SyntheMol model by generalizing across chemically similar building blocks and enabling multi-parameter optimization. We applied SyntheMol-RL to generate candidate antibiotics against S. aureus by optimizing for both antibacterial activity and aqueous solubility, and we found that SyntheMol-RL generated molecules with improved predicted properties compared to both the previous MCTS version of SyntheMol as well as an AI-based virtual screening baseline. We synthesized 79 SyntheMol-RL compounds that were unique relative to the training dataset and found that 13 showed potent in vitro activity, of which seven were structurally novel after detailed literature searches. Furthermore, one hit compound, synthecin, demonstrated efficacy in a murine wound infection model of methicillin-resistant S. aureus (MRSA). These results validate SyntheMol-RLs ability to generate novel and synthetically accessible candidate antibiotics and position SyntheMol-RL as a powerful tool for drug design across therapeutic domains.

biochemistry↗

bioRxiv2024

The Virtual Lab: AI Agents Design New SARS-CoV-2 Nanobodies with Experimental Validation

Science frequently benefits from teams of interdisciplinary researchers. However, most scientists dont have access to experts from multiple fields. Fortunately, large language models (LLMs) have recently shown an impressive ability to aid researchers across diverse domains by answering scientific questions. Here, we expand the capabilities of LLMs for science by introducing the Virtual Lab, an AI-human research collaboration to perform sophisticated, interdisciplinary science research. The Virtual Lab consists of an LLM principal investigator agent guiding a team of LLM agents with different scientific backgrounds (e.g., a chemist agent, a computer scientist agent, a critic agent), with a human researcher providing high-level feedback. We design the Virtual Lab to conduct scientific research through a series of team meetings, where all the agents discuss a scientific agenda, and individual meetings, where an agent accomplishes a specific task. We demonstrate the power of the Virtual Lab by applying it to design nanobody binders to recent variants of SARS-CoV-2, which is a challenging, open-ended research problem that requires reasoning across diverse fields from biology to computer science. The Virtual Lab creates a novel computational nanobody design pipeline that incorporates ESM, AlphaFold-Multimer, and Rosetta and designs 92 new nanobodies. Experimental validation of those designs reveals a range of functional nanobodies with promising binding profiles across SARS-CoV-2 variants. In particular, two new nanobodies exhibit improved binding to the recent JN.1 or KP.3 variants of SARS-CoV-2 while maintaining strong binding to the ancestral viral spike protein, suggesting exciting candidates for further investigation. This demonstrates the ability of the Virtual Lab to rapidly make impactful, real-world scientific discovery.

bioinformatics↗

bioRxiv2024

Learning to Choose: Behavioral Dynamics Underlying the Initial Acquisition of Decision Making

Current theories of decision making propose that decisions arise through competition between choice options. Computational models of the decision process estimate how quickly information about choice options is integrated and how much information is needed to trigger a choice. Experiments using this approach typically report data from well-trained participants. As such, we do not know how the decision process evolves as a decision-making task is learned for the first time. To address this gap, we used a behavioral design separating learning the value of choice options from learning to make choices. We trained male rats to respond to single visual stimuli with different reward values. Then, we trained them to make choices between pairs of stimuli. Initially, the rats responded more slowly when presented with choices. However, as they gained experience in making choices, this slowing reduced. Response slowing on choice trials persisted throughout the testing period. We found that it was specifically associated with increased exponential variability when the rats chose the higher value stimulus. Additionally, our analysis using drift diffusion modeling revealed that the rats required less information to make choices over time. Surprisingly, we observed reductions in the decision threshold after just a single session of choice learning. These findings provide new insights into the learning process of decision-making tasks. They suggest that the value of choice options and the ability to make choices are learned separately, and that experience plays a crucial role in improving decision-making performance.

neuroscience↗

bioRxiv2023

ADMET-AI: A machine learning ADMET platform for evaluation of large-scale chemical libraries

SummaryThe emergence of large chemical repositories and combinatorial chemical spaces, coupled with high-throughput docking and generative AI, have greatly expanded the chemical diversity of small molecules for drug discovery. Selecting compounds for experimental validation requires filtering these molecules based on favourable druglike properties, such as Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET). We developed ADMET-AI, a machine learning platform that provides fast and accurate ADMET predictions both as a website and as a Python package. ADMET-AI has the highest average rank on the TDC ADMET Benchmark Group leaderboard, and it is currently the fastest web-based ADMET predictor, with a 45% reduction in time compared to the next fastest ADMET web server. ADMET-AI can also be run locally with predictions for one million molecules taking just 3.1 hours. Availability and ImplementationThe ADMET-AI platform is freely available both as a web server at admet.ai.greenstonebio.com and as an open-source Python package for local batch prediction at github.com/swansonk14/admet_ai (also archived on Zenodo at doi.org/10.5281/zenodo.10372930). All data and models are archived on Zenodo at doi.org/10.5281/zenodo.10372418.

bioinformatics↗

bioRxiv2023

Preclinical Characterization of the Omicron XBB.1.5-Adapted BNT162b2 COVID-19 Vaccine

As SARS-CoV-2 continues to evolve, increasing in its potential for greater transmissibility and immune escape, updated vaccines are needed to boost adaptive immunity to protect against COVID-19 caused by circulating strains. Here, we report features of the monovalent Omicron XBB.1.5-adapted BNT162b2 vaccine, which contains the same mRNA backbone as the original BNT162b2 vaccine, modified by the incorporation of XBB.1.5-specific sequence changes in the encoded prefusion-stabilized SARS-CoV-2 spike protein (S(P2)). Biophysical characterization of Omicron XBB.1.5 S(P2) demonstrated that it maintains a prefusion conformation that adopts a flexible and predominantly open one-RBD-up state, with high affinity binding to the human ACE-2 receptor. When administered as a 4th dose in BNT162b2-experienced mice, the monovalent Omicron XBB.1.5 vaccine elicited substantially higher serum neutralizing titers against pseudotyped viruses of Omicron XBB.1.5, XBB.1.16, XBB.1.16.1, XBB.2.3, EG.5.1 and HV.1 sublineages and the phylogenetically distant BA.2.86 lineage than the bivalent Wild Type + Omicron BA.4/5 vaccine. Similar trends were observed against Omicron XBB sublineage pseudoviruses when the vaccine was administered as a 2-dose primary series in naive mice. Strong S-specific Th1 CD4+ and IFN{gamma}+ CD8+ T cell responses were also observed. These findings, together with prior experience with variant-adapted vaccine responses in preclinical and clinical studies, suggest that the monovalent Omicron XBB.1.5-adapted BNT162b2 vaccine is anticipated to confer protective immunity against dominant SARS-CoV-2 strains. ONE-SENTENCE SUMMARYThe monovalent Omicron XBB.1.5-adapted BNT162b2 mRNA vaccine encodes a prefusion-stabilized spike immunogen that elicits more potent neutralizing antibody responses against homologous XBB.1.5 and other circulating sublineage pseudoviruses compared to the bivalent Wild Type + Omicron BA.4/5 BNT162b2 vaccine, thus demonstrating the importance of annual strain changes to the COVID-19 vaccine.

immunology↗

bioRxiv2023

Towards Universal Cell Embeddings: Integrating Single-cell RNA-seq Datasets across Species with SATURN

Analysis of single-cell datasets generated from diverse organisms offers unprecedented opportunities to unravel fundamental evolutionary processes of conservation and diversification of cell types. However, inter-species genomic differences limit the joint analysis of cross-species datasets to homologous genes. Here, we present SATURN, a deep learning method for learning universal cell embeddings that encodes genes biological properties using protein language models. By coupling protein embeddings from language models with RNA expression, SATURN integrates datasets profiled from different species regardless of their genomic similarity. SATURN has a unique ability to detect functionally related genes co-expressed across species, redefining differential expression for cross-species analysis. We apply SATURN to three species whole-organism atlases and frog and zebrafish embryogenesis datasets. We show that cell embeddings learnt in SATURN can be effectively used to transfer annotations across species and identify both homologous and species-specific cell types, even across evolutionarily remote species. Finally, we use SATURN to reannotate the five species Cell Atlas of Human Trabecular Meshwork and Aqueous Outflow Structures and find evidence of potentially divergent functions between glaucoma associated genes in humans and other species.

bioinformatics↗

bioRxiv2022

Biologically-informed deep neural networks provide quantitative assessment of intratumoral heterogeneity in post-treatment glioblastoma

Intratumoral heterogeneity poses a significant challenge to the diagnosis and treatment of glioblastoma (GBM). This heterogeneity is further exacerbated during GBM recurrence, as treatment-induced reactive changes produce additional intratumoral heterogeneity that is ambiguous to differentiate on clinical imaging. There is an urgent need to develop non-invasive approaches to map the heterogeneous landscape of histopathological alterations throughout the entire lesion for each patient. We propose to predictively fuse Magnetic Resonance Imaging (MRI) with the underlying intratumoral heterogeneity in recurrent GBM using machine learning (ML) by leveraging image-localized biopsies with their associated locoregional MRI features. To this end, we develop BioNet, a biologically-informed neural network model, to predict regional distributions of three tissue-specific gene modules: proliferating tumor, reactive/inflammatory cells, and infiltrated brain tissue. BioNet offers valuable insights into the integration of multiple implicit and qualitative biological domain knowledge, which are challenging to describe in mathematical formulations. BioNet performs significantly better than a range of existing methods on cross-validation and blind test datasets. Voxel-level prediction maps of the gene modules by BioNet help reveal intratumoral heterogeneity, which can improve surgical targeting of confirmatory biopsies and evaluation of neuro-oncological treatment effectiveness. The non-invasive nature of the approach can potentially facilitate regular monitoring of the gene modules over time, and making timely therapeutic adjustment. These results also highlight the emerging role of ML in precision medicine.

cancer biology↗

bioRxiv2022

Predicting Immune Escape with Pretrained Protein Language Model Embeddings

Assessing the severity of new pathogenic variants requires an understanding of which mutations enable escape of the human immune response. Even single point mutations to an antigen can cause immune escape and infection by disrupting antibody binding. Recent work has modeled the effect of single point mutations on proteins by leveraging the information contained in large-scale, pretrained protein language models (PLMs). PLMs are often applied in a zero-shot setting, where the effect of each mutation is predicted based on the output of the language model with no additional training. However, this approach cannot appropriately model immune escape, which involves the interaction of two proteins--antibody and antigen-- instead of one protein and requires making different predictions for the same antigenic mutation in response to different antibodies. Here, we explore several methods for predicting immune escape by building models on top of embeddings from PLMs. We evaluate our methods on a SARS-CoV-2 deep mutational scanning dataset and show that our embedding-based methods significantly outperform zero-shot methods, which have almost no predictive power. We also highlight insights gained into how best to use embeddings from PLMs to predict escape. Despite these promising results, simple statistical and machine learning baseline models that do not use pretraining perform comparably, showing that computationally expensive pretraining approaches may not be beneficial for escape prediction. Furthermore, all models perform relatively poorly, indicating that future work is necessary to improve escape prediction with or without pretrained embeddings1.

bioinformatics↗

bioRxiv2022

7-UP: generating in silico CODEX from a small set of immunofluorescence markers

Multiplex immunofluorescence (mIF) assays multiple protein biomarkers on a single tissue section. Recently, high-plex CODEX (co-detection by indexing) systems enable simultaneous imaging of 40+ protein biomarkers, unlocking more detailed molecular phenotyping, leading to richer insights into cellular interactions and disease. However, high-plex imaging can be slower and more costly to collect, limiting its applications, especially in clinical settings. We propose a machine learning framework, 7-UP, that can computationally generate in silico 40-plex CODEX at single-cell resolution from a standard 7-plex mIF panel by leveraging cellular morphology. We demonstrate the usefulness of the imputed biomarkers in accurately classifying cell types and predicting patient survival outcomes. Furthermore, 7-UPs imputations generalize well across samples from different clinical sites and cancer types. 7-UP opens the possibility of in silico CODEX, making insights from high-plex mIF more widely available.

bioinformatics↗

bioRxiv2022

SPACE-GM: geometric deep learning of disease-associated microenvironments from multiplex spatial protein profiles.

Multiplexed immunofluorescence imaging enables high-dimensional molecular profiling at subcellular resolution. However, learning disease-relevant cellular environments from these rich imaging data is an open challenge. We developed SPAtial CEllular Graphical Modeling (SPACE-GM), a geometric deep learning framework that flexibly models tumor microenvironments (TMEs) as cellular graphs. We applied SPACE-GM to 658 head-and-neck and colorectal human cancer samples assayed with 40-plex immunofluorescence imaging to identify spatial motifs associated with cancer recurrence and patient survival after immunotherapy. SPACE-GM is substantially more accurate in predicting patient outcomes than previous approaches for modeling spatial data using neighborhood cell-type compositions. Computational interpretation of the disease-relevant microenvironments identified by SPACE-GM generates insights into the effect of spatial dispersion of tumor cells and granulocytes on patient prognosis.

bioinformatics↗

bioRxiv2022

Human anelloviruses produced by recombinant expression of synthetic genomes.

Human anelloviruses are acquired universally in infancy, highly prevalent, abundant in blood, and extremely diverse. Their apparent lack of pathogenicity indicates that they are a major component of the commensal human virome. Despite their being extensively intertwined with human biology, these viruses are poorly understood. A major impediment in studying anelloviruses is the lack of an in vitro system for their production and/ or propagation. Here we show that the T cell-derived human cell line MOLT-4 can be transfected with plasmids comprising tandem anellovirus genomes to produce viral particles visualized by electron microscopy. We found that a previously described human anellovirus of the Betatorquevirus genus (LY2), as well as a second Betatorquevirus detected by sequencing DNA extracted from a human retinal pigmental epithelium (nrVL4619), can be synthesized and produced by these means, enabling further molecular virology studies. Southern blot was used to demonstrate replication, and site-directed mutagenesis of the viral genome was performed to show that the production of anellovirus in this cell line is dependent on the expression of certain viral proteins. Finally, experiments performed in mice using purified nrVL4619 particles produced in MOLT-4 cells demonstrated infectivity in vivo in the tissue of origin. These results indicate that anelloviruses can be produced in vitro and manipulated to improve our understanding of this viral family which is ubiquitous in humans and many other mammals. Applications of this work to gene therapy and other therapeutic modalities are currently under investigation. IMPORTANCEAnelloviruses are a major component of the human virome. However, their biology is not well understood mainly due to the lack of an in vitro system for anellovirus production and/or propagation. In this study, we used multiple orthogonal measures to show that two different anelloviruses belonging to the Betatorquevirus genus can be produced in a T-cell-derived human cell line, MOLT-4, via recombinant expression of synthetic genomes. Additionally, we show that anellovirus particles generated in this in vitro system demonstrate infectivity in vivo. Our findings enable new molecular virology studies of this highly prevalent, non-pathogenic, and weakly immunogenic family of viruses, potentially leading to therapeutic applications.

microbiology↗

bioRxiv2022

Neutralization of Omicron BA.1, BA.2, and BA.3 SARS-CoV-2 by 3 doses of BNT162b2 vaccine

The newly emerged Omicron SARS-CoV-2 has 3 distinct sublineages: BA.1, BA.2, and BA.3. BA.1 accounts for the initial surge and is being replaced by BA.2, whereas BA.3 is at a low prevalence at this time. Here we report the neutralization of BNT162b2-vaccinated sera (collected at 1 month after dose 3) against the three Omicron sublineages. To facilitate the neutralization testing, we engineered the complete BA.1, BA.2, or BA.3 spike into an mNeonGreen USA-WA1/2020 SRAS-CoV-2. All BNT162b2-vaccinated sera neutralized USA-WA1/2020, BA.1-, BA.2-, and BA.3-spike SARS-CoV-2s with titers of >20; the neutralization geometric mean titers (GMTs) against the four viruses were 1211, 336, 300, and 190, respectively. Thus, the BA.1-, BA.2-, and BA.3-spike SARS-CoV-2s were 3.6-, 4.0-, and 6.4-fold less efficiently neutralized than the USA-WA1/2020, respectively. Our data have implications in vaccine strategy and understanding the biology of Omicron sublineages.

microbiology↗

bioRxiv2020

An open source platform for presenting dynamic visual stimuli

Operant behavior procedures often rely on visual stimuli to cue the initiation or secession of a response, and to provide a means for discriminating between two or more simultaneously available responses. While primate and human studies typically use LCD or OLED monitors and touch screens, rodent studies use a variety of methods to present visual cues ranging from traditional incandescent light bulbs, single LEDs, and, more recently, touch screen monitors. Commercially available systems for visual stimulus presentation are costly, challenging to customize, and are typically closed source. We developed an open-source, highly-modifiable visual stimulus presentation platform that can be combined with a 3D-printed operant response device. The device uses an eight by eight matrix of LEDs, and can be expanded to control much larger LED matrices. Implementing the platform is low-cost (<$70 USD per device in the year 2020). Using the platform, we trained rats to make nosepoke responses and discriminate between two distinct visual cues in a location-independent manner. This visual stimulus presentation platform is a cost-effective way to implement complex visually-guided operant behavior, including the use of moving or dynamically changing visual stimuli. Significance StatementThe design of an open source and low cost device for presenting visual stimuli is described. It is capable of presenting complex visual patterns and dynamically changing stimuli. A practical demonstration of the device is also reported, from an experiment in which rats performed a luminance based visual discrimination. The device has utility for studying visual processing, psychophysics, and decision making in a variety of species.

neuroscience↗

bioRxiv2020

Noradrenergic regulation of Win-Stay/Lose-Shift policy and choice determinism in a two-armed bandit task

Reversal learning depends on cognitive flexibility. Many reversal learning studies assess cognitive flexibility based on the number of reversals that occur over a test session. Reversals occur when an option is repeatedly chosen, e.g. eight times in a row. This design feature encourages win-stay behavior and thus makes it difficult to understand how win-stay decisions influence reversal performance. We used an alternative design, reversals over blocks of trials independent of performance, to study how perturbations of the medial orbital cortex and the noradrenergic system influence reversal learning. We found that choice accuracy varies independently of win-stay behavior and the noradrenergic system controls sensitivity to positive feedback during reversal learning.

neuroscience↗

bioRxiv2020

A prefusion SARS-CoV-2 spike RNA vaccine is highly immunogenic and prevents lung infection in non-human primates

To contain the coronavirus disease 2019 (COVID-19) pandemic, a safe and effective vaccine against the new severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) is urgently needed in quantities sufficient to immunise large populations. In this study, we report the design, preclinical development, immunogenicity and anti-viral protective effect in rhesus macaques of the BNT162b2 vaccine candidate. BNT162b2 contains an LNP-formulated nucleoside-modified mRNA that encodes the spike glycoprotein captured in its prefusion conformation. After expression of the BNT162b2 coding sequence in cells, approximately 20% of the spike molecules are in the one-RBD up, two-RBD down state. Immunisation of mice with a single dose of BNT162b2 induced dose level-dependent increases in pseudovirus neutralisation titers. Prime-boost vaccination of rhesus macaques elicited authentic SARS-CoV-2 neutralising geometric mean titers 10.2 to 18.0 times that of a SARS-CoV-2 convalescent human serum panel. BNT162b2 generated strong TH1 type CD4+ and IFN{gamma}+ CD8+ T-cell responses in mice and rhesus macaques. The BNT162b2 vaccine candidate fully protected the lungs of immunised rhesus macaques from infectious SARS-CoV-2 challenge. BNT162b2 is currently being evaluated in a global, pivotal Phase 2/3 trial (NCT04368728).

immunology↗

bioRxiv2018

Glioblastoma recurrence and the role of MGMT promoter methylation

Tumor recurrence in glioblastoma multiforme (GBM) is often attributed to acquired resistance to the standard chemotherapeutic agent temozolomide (TMZ). Promoter methylation of the DNA repair gene MGMT has been associated with sensitivity to TMZ, while increased expression of MGMT has been associated with TMZ resistance. Clinical studies have observed a downward shift in MGMT methylation percentage from primary to recurrent stage tumors. However, the evolutionary processes driving this shift, and more generally the emergence and growth of TMZ-resistant tumor subpopulations, are still poorly understood. Here we develop a mathematical model, parameterized using clinical and experimental data, to investigate the role of MGMT methylation in TMZ resistance during the standard treatment regimen for GBM (surgery, chemotherapy and radiation). We first find that the observed downward shift in MGMT promoter methylation status between detection and recurrence cannot be explained solely by evolutionary selection. Next, our model suggests that TMZ has an inhibitory effect on maintenance methylation of MGMT after cell division. Finally, incorporating this inhibitory effect, we study the optimal number of TMZ doses per adjuvant cycle for GBM patients with high and low levels of MGMT methylation at diagnosis.

cancer biology↗