Workshop 3: Hybrid Multi-Scale Modelling and Validation
Coupled multiscale modeling and pathway analysis for prediction of drug efficacy in cystic kidney diseasesJames Glazier
Extensive research has uncovered many genetic changes associated with autosomal dominant polycystic kidney disease (ADPKD) and effects of ADPKD mutations on signaling pathways. However, we still do not know the precise sequence of events that lead to cyst initiation. One of the key changes during the initiation of cysts is abnormal expression of the juvenile cell adhesion molecule cadherin-8. We examined two hypothetical cell-level mechanisms by which abnormal expression of cadherin-8 could initiate cyst formation: i) reduction of cell-cell adhesion, which then leads to changes in cell proliferation or ii) direct reduction of contact inhibition of proliferation with no change in cell-cell adhesion. To test these mechanisms we built a 3D virtual-tissue (VT) computer model of the renal tubule using the CompuCell3D (CC3D) modeling environment (Swat et al., 2012). Our VT simulations showed that while both mechanisms could initiate cyst formation, only the loss of adhesion mechanism produced morphologies matching in vitro cadherin- 8 induced cysts (Belmonte et al., 2016).
Concurrently, we used the Transcriptogram method for whole-genome gene expression analysis to analyze microarray data from cell lines developed from cell isolates from normal kidney and from both non-cystic nephrons and cysts from the kidney of a patient with ADPKD. We identified novel pathways altered in ADPKD. Transcriptogram significance metrics identified increased expression of cGMP phosphodiesterases as the highest priority pathways for study (de Almeida et al., 2016). Our modeling and experimental efforts then focused on cGMP phosphodiesterase inhibitors, a class of drugs already FDA approved for other uses.
Using pathway analysis we linked the cell behaviors known to drive cyst formation with increased cGMP phosphodiesterase expression and constructed models of these pathways using Cell Designer. We are currently calibrating these pathway models using biological data. Preliminary in vitro and mouse model testing of phosphodiesterase inhibitors to reduce cyst formation have shown efficacy. We will next incorporate these pathway models into our CC3D VT cystogenesis model to predict drug effects on cyst formation.
Brain tumours represent a rare but serious medical condition. With an incidence of six cases per 100000, gliomas are the most frequent primary brain tumours in adults, accounting for 70% of cases. Gliomas are classified into four grades by increasing aggressiveness, based on their microscopic structure and cellular activity. Glioblastoma multiforme (GBM) is the most frequent and most malignant sub-type of glioma (grade IV), accounting for about 50% of diffuse gliomas. These tumours infiltrate surrounding healthy tissue, grow rapidly and form a necrotic core of high cell density, which is accompanied by compression and displacement of surrounding tissue. This so-called mass-effect leads to an increase in intra-cranial pressure and the progressive on-set of a multitude of pressure-related symptoms, such as headache and nausea. Compensation mechanisms for regulating intra-cranial pressure fail beyond a critical tumour volume, so that any additional volume increase will result in a decisive rise in intra- cranial pressure and related acute clinical worsening, including coma or death due to herniation
Given the importance of mechanical effects, we have started a systematic numerical analysis of the dependence of morphological and mechanical tumour characteristics on their growth location. This study is part of an ongoing effort to establish a model of the macroscopic mechanical aspects of tumour growth that can also be integrated into multi-scale cancer models, such as those proposed by the CHIC project.
The impact of collecting data at varying temporal resolution on parameter inference for biological transport modelsJonathan Harrison
When collecting time series data of biological transport processes, it is necessary to observe the system at discrete time points, for example via an imaging experiment. This can introduce errors when the motion is approximated with discrete steps. We study the impact of collecting data at different temporal resolutions on parameter inference for biological transport models. In this work, we have performed exact inference for velocity jump process models in a Bayesian framework. This allows us to obtain estimates of the turning rate and noise amplitude for noisy observations of this transport process. We show sensitivity of these estimates to changes in time discretisation and noise amplitude. For a fixed photon budget, our results suggest that better estimates of parameters can be obtained when imaging more frequently with more noise than imaging sparsely with low noise.
Stem cells are identical in many scales, they share the same molecular composition, DNA, genes and genetic networks, yet they should acquire di?erent properties to form a functional tissue. Therefore, they must interact and get some external information from their environment, either spatial (dynamical fields) or temporal (lineage). In this work we test to what extent coupled chemical and physical fields can underlie the cellâ€™s positional information during development. We choose the root apical meristem of Arabidopsis thaliana to model the emergence of cellular patterns. We built a model to study the dynamics and interactions between the cell divisions, the local auxin concentration and physical elastic fields.
Since clinical approval in 2004 of the vascular endothelial growth factor (VEGFA) blocking antibody, bevacizumab (Avastin), for treatment of colorectal cancer, a substantial number of additional anti-angiogenic compounds are now available for cancer treatment. Although these compounds are primarily targeted against the angiogenic process itself, they undoubtedly have additional effects on already established tumour blood vessels. In addition, a number of so-called tumour vascular disrupting agents (VDAs), which are specifically designed to target established tumour blood vessels, are in clinical trials. Despite this success, resistance to treatment is a major problem, with lack of predictive biomarkers to select those patients most likely to benefit from vascular targeted treatments a major limitation and biomarkers of response technically challenging.
VEGFA exists as multiple isoforms generated through alternative splicing and proteolysis. Recent retrospective analyses of data from several large phase III clinical trials have found an association between high concentrations of soluble VEGFA isoforms in plasma and poor prognosis, but also improved response to bevacizumab, making them potential predictive biomarkers. Using mouse fibrosarcoma cells genetically modified to express single isoforms of VEGFA, we have investigated the role of individual VEGFA isoforms in tumour vascularisation, patterning and function, metastasis and response to VEGFA pathway inhibitors and VDAs. Notably, soluble VEGFA-120 was associated with increased metastasis to the lung and a good response to the anti-VEGFA blocking antibody, B20-4.1.1. (the mouse equivalent of bevacizumab). Expression of VEGFA-120 was associated with highly permeable and dilated blood vessels in the primary tumour and a modified extracellular matrix, which could account for the increased metastasis. Analytical methods for measuring vascular and metabolic parameters in intravital microscopy and magnetic resonance imaging/spectroscopy (MRI/MRS) of tumours will be discussed.
Building on recent advances in computer vision and machine learning we are now in the position to monitor complex biological environments and events in the same way are analysing natural scenes. While challenges remain, algorithms for cell segmentation and tracking have matured significantly and can now be used in more routine high-throughput settings. Improved microscopy and imaging platforms not only allow us to image subcellular events at high spatial and temporal resolution, we can now image large tissue sections and capture how various different proteins modulate the cellular microenvironment. Enabled by advances in cell culturing technologies 3D cultures can restore specific biochemical and morphological features that are similar to their in vivo counterparts. This holds the potential for improving relevance of in vitro studies, improving our ability to predict what occurs in vivo.
We are now working towards establishing the spatial and temporal context for biological events and processes. Quantitative image analysis methods are necessary for monitoring the tissue formation process and enabling longer duration time-lapse imaging. The talk will highlight opportunities of interfacing imaging with mathematical modelling. One such example will be the modelling of cellular behaviour. Our current research focuses on analysing cellular viability, the interaction of epithelial cell populations and the evolution of organoid cell cultures.
The most prevalent, devastating, and complex diseases of our time, such as diabetes, cardiovascular disease, and infectious diseases, result from the interactions of heterogeneous cells with one another and with their environment. However, the emergence of disease from these interactions at the multi-cell level is still poorly understood, and drugs typically target single molecular pathways while disregarding how cellular heterogeneities might affect drug efficacy at the tissue-level. To address this void, we develop new computational tools in combination with experimental approaches in order to integrate and predict how individual cell behaviors dynamically give rise to physiological and pathological tissue-level adaptations. Leveraging the versatility and adaptability of agent-based modeling, we have simulated structural adaptations of large and small blood vessels, skeletal muscle regeneration following injury, and immune cell trafficking and differentiation during inflammation and infection. Our studies have suggested new mechanistic hypotheses and provided guidance for the design of novel therapies.
Problems in tissue engineering, developmental biology, cancer, and related areas require that we study 3-D multicellular systems, coupling dynamics at many scales such as protein signaling, cell phenotype "decisions," biotransport, and mechanics. Developing, calibrating, and validating models to study these systems requires not only sophisticated tools, but also a huge variety of data ranging from molecular to clinical scales. No one research team can develop all the necessary software and gather all the required data on their own, so we present our recent work to (1) contribute scalable open source software to simulate 3-D biotransport of many substrates (BioFVM), 3-D multicellular systems (PhysiCell), and extraction of cell phenotype from high-throughput experiments (CellPD), (2) help build a community for collaboration by creating a standard for multicellular data (MultiCellDS), and (3) work with this community to link biosimulation software and open data repositories through data standards.
The Hippo pathway, which is a central pathway in the control of cell proliferation and apoptosis in Drosophila and mammalian cells, contains a core kinase mechanism that affects control of the cell cycle and growth. Studies involving over- and under-expression of components in the morphogen and Hippo pathways in Drosophila reveal conditions that lead to over- or undergrowth. In this talk we discuss a mathematical model that incorporates the current understanding of the Hippo signal transduction network in Drosophila and which can explain qualitatively both the observations on whole-disc manipulations and the results arising from mutant clones. We find that a number of non-intuitive experimental results can be explained by subtle changes in the balances between inputs to the Hippo pathway. Since signal transduction and growth control pathways are highly conserved across species and directly involved in tumor growth, much of what is learned about Drosophila will have relevance to growth control in mammalian systems.