Projects and Grants

We are embarking on a mix of projects, most combining experiments with theory, but some purely theoretical. Here are descriptions of a few:

Predicting the Heterogeneity of Cell Fate Decisions (NIH 1R01GM104184)
Emilce Carrasco, SreeHarish Muppirisetty, and Suzanne Scarlata (collaborator at Stony Brook University)

In cancer, genetic heterogeneity is the focus of many investigations as it plays important roles in tumor progression and drug resistance by driving phenotypic diversity. Here, we consider another type of heterogeneity, one where natural cell-to-cell variability in protein levels in genetically-identical mammalian cells causes the same stimuli to yield different cell fates, such as life or death. We term this phenomenon "natural phenotypic divergence" (NPD). NPD can manifest as, for example, a persistent anticancer drug resistant subpopulation of cells, and understanding it is important for predicting cancer treatment efficacy. However, means to predict NPD from cell-based experiments have not been developed and are the subject of this project. We hypothesize that NPD can be predicted by characterizing how multivariate, endogenous protein expression noise is propagated non-linearly through signaling networks to regulate cell fate. It is the endogenous expression and degradation noise in the levels of multiple proteins within a signaling network that collectively manifest as NPD. We will test this hypothesis by combining experimental and computational approaches to examine NPD-based proliferation. First, experimentally, we will use live-cell imaging approaches with FRET probes to measure real-time signaling network dynamics and proliferation simultaneously. Although we can only measure one pathway at a time, our subsequent use of computational, dynamic modular response analysis theory allows us to reconstruct how these pathways dynamically interact in a stimulus-specific fashion to control stochastic proliferation fates. Second, we will build a chemical kinetics-based, stochastic computational model that simulates how the protein expression variability underlying NPD propagates into signaling dynamics heterogeneity. Analysis of this model will suggest sets of key proteins whose collective, multivariate fluctuations have a large influence on NPD-based proliferation. Finally, we will measure fluctuations in the levels of these key proteins in single live cells, use our computational models to predict whether these cells should proliferate or not in response to defined perturbations, and test the predictions by observing the actual proliferation decision in those same cells. If successful, this would be the first demonstration that the stochastic fates of individual live cells could be predicted based on biomarkers present prior to perturbation. This would be an important step towards identifying biomarker sets for individual patients and fashioning personalized therapeutic strategies.

Disease-centric Modeling of Glioblastoma Multiforme Signaling Pathways (2P50GM071558 -Systems Biology Center New York and T32GM062754)
Mehdi Bouhaddou, Emilce Carrasco

The past 15 years has led to a wide variety of differential equation-based chemical kinetics models, including ours on receptor tyrosine kinase and proliferation/growth signaling, that describe the chemical reaction mechanisms comprising signaling pathways. These models are built by representing the mechanisms of biochemistry as physical principle-based rate laws that describe how quickly reactions proceed based on concentrations of cellular entities. The rationale for building such models has been that because such pathways are often deregulated in cancer, then a detailed predictive understanding of how they respond to drugs would improve our ability to personalize cancer treatments. However, The Cancer Genome Atlas (TCGA) data has made it clear that cancer “does not care” about single pathways; rather, it uses a variety of pathways simultaneously for malignant progression. Thus, this pathway-centric modeling of the near past can now be used as a base on which to build disease-centric models, which capture a much larger breadth of relevant pathophysiology. We have are building such a model for glioblastoma multiforme (GBM), which includes, as guided by TCGA data, pathways for EGFR, cMET, PDGFR, PI-3K, PTEN, NF1, BRAF, CDKN1A/B, RB, CDK4, CDK6, and P53. As a first step, we are training this model based on publicly available data in the cancer cell line encyclopedia (CCLE), which contains data for 43 glioma lines for their mutations and response to 24 different chemotherapeutics, and the library of integrated network-based cellular signatures (LINCS) that contains molecular readouts (transcriptome and various pathway activities) in response to a variety of perturbations including FDA-approved drugs. We are also applying the model to a single cell stochastic setting and training it based on live-cell imaging data from U87 glioma cells. We aim to use this model as a vehicle for preclinical testing of new putative drugs based on a variety of genomic alteration contexts, as well as for proposing new potentially effective treatment strategies.

Derivative-based Parameter Estimation and Experimental Design
SreeHarish Muppirisetty

The generic ordinary differential equation (ODE) model of biological signaling, when cast in terms of elementary processes, takes a form that has time derivatives on the left-hand sides, and is linear in the parameters on the right-hand sides. Upon time integration, however, the parameters become complex non-linear functions of the model states. Since models of biological signaling typically comprise tens to hundreds of states and the same numbers of unknown parameters, solving non-linear parameter estimation problems and designing experiments for identification of signaling models is computationally challenging. Traditional biological assays used for modeling purposes (e.g. western blot, ELISA, immunofluorescence, etc.) rely on fixed or lysed cells. As a result, obtaining high frequency measurements from such assays is not feasible, and likewise nor is obtaining accurate estimates of time derivatives. This forces researchers to estimate model parameters after time integration, giving a difficult, high-dimensional non-linear optimization problem. However, a host of Forster resonance energy transfer (FRET) probes for detecting biochemical activities in live cells have recently been developed which allow for high frequency measurements and therefore more accurate estimation of state time derivatives. Thus, modeling elementary processes based on FRET probe data allows one to perform parameter estimation and identifiability analysis without time integration, yielding problems that are linear in the parameters. Furthermore, by considering time derivatives as an available measurement, one obtains an experimental design problem that is also linear in the parameters. We are currently exploring to what extent this possibility of using time-derivative data for doing parameter estimation, parameter identifiability analysis, and model-based experimental design is valuable, and deriving theoretical frameworks to support the use of such derivative-based methodologies.