Molecular mechanisms resulting in tumour recurrence and resistance
Patient diversity and tumour heterogeneity are biological phenomena that hinder the identification of efficient therapies against cancer. In this research effort, we sought to identify the molecular signatures that have an impact on disease progression, therapy resistance and ultimately, patient survival.
We developed new computational pipelines and applied them to multiple myeloma (MM) and glioblastoma (GBM), as paramount examples of therapy resistance and tumour recurrence. Starting from a publicly available large cohort of patient tumour molecular profiles in MM (n ~ 1,000), we learned the transcriptional regulation units (regulons) that facilitate a mechanistic understanding of risk, relapse and therapy responsiveness. Zooming in at single-cell resolution on GBM, we modelled tumour cell-state transitions upon standard-of-care therapy in primary, corresponding mouse xenografts and recurrent tumours derived from a patient. As an immediate prospect, our results uncover molecular vulnerabilities that can be targeted with available drugs.
Identification of gene regulatory mechanisms at single-cell resolution from a tumor biopsy. Starting from the single-cell transcriptome profiles a patient tumor biopsy, computational pipelines (left) infer which transcriptional programs are differentially active in particular cell populations (right).
Gene regulatory network topology governs resistance and treatment escape in glioma stem-like cells Park et al. Sci. Adv. (2024)
A single-cell based precision medicine approach using glioblastoma patient-specific models Park et al. NPJ Precision Oncology (2022)
Genetic program activity delineates risk, relapse, and therapy responsiveness in multiple myeloma Wall et al. NPJ Precision Oncology (2021)
A systems approach to brain tumor treatment Park et al. Cancers (2021)
Metabolic mechanisms leading to endothelial dysfunction
Acute critical illness (ACI) is a disparate group of conditions, including trauma, sepsis and myocardial infarction—with a roughly estimated incidence of 60 million patients worldwide each year—which are united by endothelial dysfunction associated with elevated sympathoadrenal signalling. The regulation of endothelial cell metabolism is distinct and profoundly important to endothelium homeostasis. However, a precise catalogue of the metabolic alterations caused by sustained high catecholamine levels that results in endothelial dysfunction is still underexplored.
Here, we uncovered a set of up to 46 metabolites that directly respond to high levels of catecholamines in human umbilical vein en- dothelial cells (HUVECs). The identified metabolites align with the glutathione-ascorbate cycle and the nitric oxide biosynthesis pathway. Furthermore, we quantified an increase in glucose consumption and aerobic respiration upon catecholamine stimulation. A precise understanding of the metabolic response in endothelial cells to pathological levels of catecholamines will facilitate the identification of more efficient clinical interventions in trauma patients.
Metabolic response of ECs in an in vitro model of ACI. HUVECs became more glycolytic and increased respiration (top) upon catecholamine exposure. Treatment-responding identified metabolites (right) align with the glutathione-ascorbate cycle and the nitric oxide biosynthesis pathway.
Metabolic response in endothelial cells to catecholamine stimulation associated with increased vascular permeability López García de Lomana et al. Int J Mol Sci (2022)
Microbial cell state transitions
Cell state transitions are arguably one of the most fascinating phenomena in biology. Mechanistic understanding of such complex processes is a paramount scientific challenge. Through a systems approach, we can better comprehend how cells explore and transit the genotype-to-phenotype map to secure homeostasis. A pertinent question is whether cellular state transitions are a succession of potentially divergent micro-states that connect initial to final states. This framework posits the basis to reveal the molecular mechanisms of such transitions, ultimately informing on molecular drivers whose rational perturbation would grant us control over desired phenotypes.
We applied this framework to diatom resilience under global warming environmental conditions and a microbial syntrophic community under resource fluctuating conditions.
Synergistic epistasis enhances the co-operativity of mutualistic interspecies interactions Turkarslan et al. The ISME Journal (2021)
Ocean acidification conditions increase resilience of marine diatoms Valenzuela et al. Nature Communications (2018)
Mechanism for microbial population collapse in a fluctuating resource environment Turkarslan et al. Molecular Systems Biology (2017)
Robustness of a model microbial community emerges from population structure among single cells of a clonal population Thompson et al. Environmental Microbiology (2017)
Systems biology approaches towards predictive microbial ecology Otwell et al. Environmental Microbiology (2018)
Cell state transitions mapped into a new space learned from expression distributions of cell state gene descriptors.
Integration of metabolic and gene regulation to rationally predict phenotypic consequences of genetic perturbations
Computational methods that enable the prediction of phenotypic consequences of genetic perturbations are valuable to a diverse range of applications from novel drug targets discovery to bioengineering strain improvement. We developed new methods to predict metabolic vulnerabilities in Mycobacterium tuberculosis and revealed the transcriptional waves that bring Chlamydomonas reinhardtii from nitrogen starvation to lipid accumulation states.
Transcriptional program for nitrogen starvation-induced lipid accumulation in Chlamydomonas reinhardtii López García de Lomana et al. Biotechnology for Biofuels (2015)
A refined genome-scale reconstruction of Chlamydomonas metabolism provides a platform for systems-level analyses Imam et al. The Plant Journal (2015)
Dynamics of transcriptional waves responding to nitrogen starvation. Each line represents a set of co-regulated gene modules.
New method to predict phenotypic consequences of genetic perturbations in M. tuberculosis through the integration of metabolic and gene regulatory network models.
Quantitative prediction of conditional vulnerabilities in regulatory and metabolic networks using PRIME Immanuel et al. npj Systems Biology and Applications (2021)
Translational regulation in Archaea
The ribosome is a defining structure of all cellular organisms and represents the uniquely biological process of translation. Although ribosomes are typically represented as molecular assembly lines that constitutively perform coded protein synthesis, translation is a system property that emerges from interactions among diverse types of cellular components, including different classes of RNAs, proteins, and amino acids. The prevailing view that all ribosomes within a cell are structurally identical and functionally equivalent was challenged long ago, and is becoming increasingly less tenable with modern experimental tools. Molecular interrogation of Halobacterium salinarum in total mRNA abundance (RNA-seq), ribosome-associated transcripts (ribosome profiling), and protein abundance (SWATH-MS proteomics) reveals growth-associated changes in ribosome composition and regulation. These data provide evidence that variability in translational systems confers further intrinsic regulation of protein translation in Archaea. Environment-specific ribosome-based regulation is a mechanism for steering physiological state transitions, therefore acting as a fundamental constraint on biological evolution.
A comprehensive spectral assay library to quantify the Halobacterium salinarum NRC-1 proteome by DIA/SWATH-MS Kusebauch et al. Scientific Data (2023)
A Genome-Scale Atlas Reveals Complex Interplay of Transcription and Translation in an Archaeon Lorenzetti et al. mSystems (2023)
Selective Translation of Low Abundance and Upregulated Transcripts in Halobacterium salinarum López García de Lomana et al. mSystems (2020)
Interplay of transcription and translational regulation. A substantial proportion of transcripts are translationally regulated only (orange and green) or exhibit compensatory mechanisms (red and blue). Intriguing, some transcriptionally down-regulated transcripts are translationally up-regulated (magenta).
Experimental evidence for the emergence of adaptive prediction
Adaptive prediction is the capability of diverse organisms to sense a cue and prepare in advance to deal with a future environmental challenge. We investigated the dynamics and molecular mechanisms of adaptive prediction emergence when an organism encounters an environment with a novel structure. We subjected yeast to laboratory evolution in a structured environment with repetitive, coupled exposures to a neutral chemical cue (caffeine), followed by a sublethal dose of a toxin (5-FOA). We demonstrate the remarkable ability of yeast to internalize a novel environmental pattern within 50-150 generations by adaptively predicting 5-FOA stress upon sensing caffeine. We also demonstrate how novel environmental structures can be internalized by coupling two unrelated response networks, such as the response to caffeine and signalling-mediated conditional peroxisomal localization of proteins.
Adaptive Prediction Emerges Over Short Evolutionary Time Scales López García de Lomana et al. Genome Biology and Evolution (2017)
Experimental design for adaptive prediction emergence.
Noise injection-based model selection framework
Selecting the most likely ODE model for a given experimental data is not a trivial task. Common problems are (1) broad parameter regions compatible with the data instead of a single parameter value, (2) selection of different complexity alternative explicative models and (3) refinement of promising models. With the help of an efficient tool to explore the parameter space, a Bayesian framework, and a mathematical filter, we build a model selection framework that tackled all those commonly presented problems. We studied an interesting process of metabolite transport in yeast, providing with a novel understanding of the dynamics of such process.
Inferred model for yeast glutamine transport dynamics. A. Experimental data for external (blue dots) and intracellular (red dots) glutamine concentrations and model simulations as mean (solid lines) and standard error (shaded regions) of the predicted trajectories. B. Viable space for three parameters. Each dot represents a model instance in parameter space, colour maps to model likelihood (E).
Topological Augmentation to Infer Hidden Processes in Biological Systems Sunnåker et al. Bioinformatics (2014)
Graph properties of biochemical networks
Biochemical entities such as genes, proteins or metabolites interact in the cell to deploy their functions. As with any other ensemble of interacting elements, a such network can be represented as a graph. Several graph properties are relevant to characterize a biological network, probability degree distribution being one of the most studied. The probability degree distribution of biochemical graphs is substantially different from the Poisson distribution and multiple fat-tailed distributions show an unequivocally good match.
Statistical Analysis of Global Connectivity and Activity Distributions in Cellular Networks López García de Lomana et al. Journal of Computational Biology (2010)
Visualization of yeast protein-protein interaction network. Each dot represents a protein, dot size represents the number of interactions and colour maps for node centrality. Figure generated with LaNet-vi.
Optimal experimental design for dynamical systems in biochemistry
Coupled biochemical reactions can be modelled dynamically as a system of ordinary differential equations. While typically kinetic interactions among reactants are fairly described, parameter values entail large uncertainties. Acquisition of such data for biochemical systems is at best difficult and expensive. Therefore, the determination of the most useful data set, i.e., the most informative system output for the model calibration is of critical importance. Optimal experimental techniques deal with this problem from very different approaches. Local strategies are fast but impractical due to many local minima. Global methods, although computationally expensive, appear to be more adequate for biochemical systems which are commonly large and poorly constrained.
Optimal Experimental Design in the Modelling Of Pattern Formation López García de Lomana et al. Lecture Notes in Computer Science (2008).
Map of most informative variables across time. For four key molecules, each colour represents the goodness for model calibration (lower values are more informative), guiding which variables and which time points represent optimal information for model calibration.