Speaker Abstracts:

_______________________________________________
S. Agarwal (MIT)
Gene Prioritization Using Ranking Methods in Machine Learning: A New Computational Approach

The problem of identifying key genes that are involved in a particular disease is of fundamental importance in biology and medicine. With the rapid growth in biological data sources containing gene-related information, ranging from gene expression data to protein interaction data, there is much interest in developing computational approaches that can analyze this data and help in the identification of important genes. In particular, a common goal is to rank or prioritize genes such that those relevant to the disease under study are likely to appear at the top of the ranking; the proteins corresponding to the top few genes can then be subjected to biological tests to elucidate their structural and functional properties, with a good chance that many of those tested will emerge as targets for drug development or find use as disease markers. Recently, the problem of ranking objects has received considerable attention in the fields of machine learning and data mining; ranking problems arise in a variety of domains ranging from web search to recommendation systems, and a variety of new learning methods have been developed that directly optimize ranking performance. In this talk, I will describe the use of such ranking methods in machine learning for prioritizing genes. In particular, using such ranking methods, we have
recently identified a new gene that is highly over-expressed in leukemia; I will describe these results, as well as some directions for future research.

_______________________________________________
H. Bolouri (Caltech)
The potential impact of inter-individual genomic variations on disease susceptibility and treatment response

Recently sequenced individual genomes have highlighted enormous inter-individual genomic variability. An analysis of the genomes of Craig Venter and James Watson suggests that most cellular signaling pathways are impacted by dozens of sequence variants with potential biochemical effects. Using models of the canonical Wnt signaling system, I argue that combinations of apparently benign/small-effect sequence variations in individual genomes can result in considerable functional diversity in signal transduction characteristics. Population-scale Monte-Carlo analysis suggests that significant numbers of apparently healthy individuals may be far more susceptible to environmental insults than expected from population-averaged models. I will conclude by exploring the implications of these observations for medical research and drug development.

_______________________________________________
A. Chakraborty (MIT)
Regulation of Ras signaling in lymphocytes

The activation of Ras proteins is a key step in activation of T and B - lymphocytes. I will describe work that brings together theoretical and computational studies (rooted in statistical mechanics) with genetic and biochemical studies (Weiss and Roose labs) that describes how Ras activation is regulated in lymphocytes. Specifically, I will discuss how the interplay between two Ras Guanine Exchange Factors results in digital signaling and hysteresis. Implications of our studies for the development of T cell lymphomas will also be presented.

_______________________________________________
C.S. Drapaca (Department of Engineering Science and Mechanics, Pennsylvania State University)
An Elastography – Mass Spectroscopy Coupling Model for Tumors’ Classification

Co-author: A.J. Palocaren

Palpation is an important medical diagnostic tool which is based on the fact that tumors tend to be stiffer than the surrounding normal tissue. None of the modern, non-invasive, imaging techniques (such as CT scan, Magnetic Resonance Imaging, or Ultrasound) used today by clinicians to find and diagnose tumors provides the critical information about the stiffness of the imaged tissues. The clinical or radiological observation of palpation can be explained theoretically by the following hypothesis: the Young’s modulus of tissues helps differentiating not only between normal and abnormal tissues but, more importantly, between benign and malignant (cancer) tumors. In this talk we present a novel technique of differentiating between benign and malignant tumors based on their corresponding Young’s moduli obtained using information about tissue microstructure provided by image mass spectroscopy.

_______________________________________________
M. Foldvari (Waterloo)
Non-viral Gene Delivery Systems based on Gemini Nanoparticles

The development of safe and effective non-viral delivery systems for non-invasive administration of nucleic acids is becoming increasingly important for both local and systemic treatments. This presentation will focus on some recent designs of non-viral delivery systems based on phospholipid-gemini surfactant composites (PL-gemini nanoparticles) and carbon nanotube-gemini surfactant composites (CNT-gemini nanoparticles). Gemini surfactants with two cationic headgroups encapsulate plasmid DNA, creating an advanced type of nanoparticle-based delivery system designed to transfect DNA both in vitro and in vivo for subsequent expression of desired therapeutic proteins. Gemini surfactants provide a basis to develop novel non-viral delivery systems for gene therapy; however, gemini nanoparticles must possess several crucial properties to overcome difficult cellular and tissue barriers. Structure-activity relationship between the chemical structure of gemini surfactants/nanoparticle internal morphological structure and transfection efficiency will be presented.

_______________________________________________
S. Haase (Duke)
Cycling Without Cyclins: New Views on the Cell Cycle Oscillator

Laura A. Simmons Kovacs1, Sara Bristow1, David A. Orlando1, Michael Mayhew1, Yuanjie Jin1, Charles Y. Lin1, Allister Bernard2, Jean Y. Wang1, Joshua E. Socolar3, Edwin S. Iversen4, Alexander J Hartemink2, John Harer5, Sayan Mukherjee4 and Steven B. Haase1. 1Dept. of Biology, 2Dept. of Computer Science, 3Dept. of Physics, 4Dept. of Statistical Science, 5Dept. of Mathematics, Duke University, Durham, NC, USA

Early work in frog and marine invertebrate embryos suggested that a biochemical oscillator centered on the cyclin dependent kinase (CDK) complex controls cell-cycle events. However, our findings suggest that the cell cycle in budding yeast is entrained to a transcription factor network oscillator that can function independently of cyclin/CDKs (Haase and Reed, Nature 1999, Orlando et al. Nature 2008). We have been investigating how the function of this transcription network oscillator is influenced by other cell-cycle control mechanisms including CDKs and checkpoints. While Boolean models indicate that the transcription factor network could function as a cell-cycle oscillator independent of CDK activity, our experimental observations indicate that CDKs contribute to the robustness of the oscillations. We are using ordinary differential equations models rather than Boolean models to elucidate how CDKs might influence network dynamics. By examining genome-wide transcriptional dynamics in cells arrested by the DNA replication checkpoint, we determined that checkpoint pathways triggered the arrest of periodic transcription as well as cell-cycle progression. We propose that checkpoints block progression through the transcription factor network oscillator in order to maintain coordination between the transcriptional program and cell-cycle progression. This coordination is likely to be important for both a robust arrest, and eventually, recovery from the arrest. Collectively, our findings support a model in which a transcription factor network oscillator cooperates with CDK activity and checkpoint mechanisms to produce robust cell-cycle oscillations.

_______________________________________________
T. Huzil (Waterloo)
Atomistic Simulation of Receptor-Ligand Interactions

Many biological processes involve the specific binding of small molecules to proteins. Modeling of these interactions can yield atomic-resolution insight into ligand-receptor complexes that play a qualitative role in applications, such as structure-based drug design. First generation protein-ligand docking algorithms made the simplifying assumption that the receptor protein is rigid. The receptor conformation(s) would generally be obtained from a structure that had been determined by X-ray crystallography or NMR spectroscopy. This effectively reduced the degrees of freedom, and therefore difficulty of the calculations, to those of only the ligand: three translational, three global-rotational, and one internal dihedral rotation for each rotatable bond. This view is too simplistic for many systems and current algorithms tend to be based on the induced-fit model, where the structure of both the receptor and ligand adapt during the binding process. This level of detail can be achieved through atomistic simulation models, such as Molecular Dynamics (MD), where the motion of atoms are simulated by evolving their atomic configuration in time according to Newton's equation (F=ma). Not only is this a more accurate view of a dynamic system, it also allows for the direct study of both the dynamical and thermodynamical evolution of the system. For this talk, I will briefly discuss the application of atomistic simulations in my research directed at the rational design of novel drugs.

_______________________________________________
Tak Mak (UHN, UToronto)
Blocking Oncogenes to Cure Cancer: Is the Paradise Lost?

Tumours arise and eventually metastasize due to the cumulative effects of multiple mutations on multiple key genes. Oncogenes undergo mutations that cause them to become active when they shouldn't, and tumor suppressor genes (TSGs) sustain damaging alterations that obliterate their protective functions. TSGs include genes that normally control cellular differentiation, regulate cell growth and the cell cycle, participate in DNA repair, and govern pathways leading to programmed cell death or survival. Knowledge of the roles of these genes in preventing or promoting tumour formation has enabled molecular oncologists to seek mechanistically-based drugs for cancer treatment. Originally, the "Oncogene Revolution" prompted these investigators to concentrate on the development of agents that block cell growth and cell cycle progression. Although therapeutics based on this approach have had some success in the clinic, it has become increasingly clear that the number of genetic aberrations are very abundant and the pathways leading to cancer are too complicated. Thus, it has become clear that, to be effective, anti-cancer agents must also target molecules involved in the metabolism, metastasis and death of tumour cells as well as proteins crucial for tumour angiogenesis.

_______________________________________________
S. Mani (M.D. Anderson Cancer Center, Houston)
Links between stem cells and EMT: A new turn in cancer initiation and progression

The majority of deaths among carcinoma patients are due to the development of metastasis. In order to metastasize, carcinoma cells must complete a complex multistep process including invasion, entering into and distribution via the circulation, extravasation, and expansion into macroscopic tumors at the new site. Recent studies have found that the activation of a latent embryonic program-epithelial-mesenchymal transition (EMT)-in carcinoma cells lead to increase in metastatic potential. EMT is a complex series of cellular reprogramming events through which epithelial cells (i.e. early carcinoma cells) lose their epithelial characteristics and acquire mesenchymal-like characteristics, such as increased migration and invasion. We have found that EMT also endows carcinoma cells with properties of tumor-initiating, cancer stem cells (CSCs). These properties may enable the expansion of single or small colonies of disseminated cancer cells into full-fledged metastatic nodules. The acquisition of stem cell characteristics has thus demonstrated an increased importance for understanding the mechanisms that regulate EMT not only to improve treatment of metastatic disease but also for better treatment of refractory disease, which have been independently linked to CSCs. My presentation will summarize the previous results, which yield insight into mechanisms of breast cancer progression as well as suggest possible new avenues for therapeutic intervention.

_______________________________________________
M. Milosevic (PMH, Toronto)
Individualized, Adaptive Radiotherapy: The Next Frontier in Radiation Medicine

Human tumors are dynamic, unstable systems that are constantly changing. There may be substantial variability in clinical behavior among tumors of the same type in different patients. In addition, the location, size, shape and biologic signature of a tumor in an individual patient may change during the six to seven weeks that are usually required to deliver a radical course of fractionated radiotherapy.

Radiation treatment plans typically are developed from a single planning CT image set obtained several days or even weeks prior to the start of therapy. Daily on-line portal imaging provides some assurance that the treatment is delivered as planned, using musculoskeletal landmarks as surrogates for the location of tumor. This is often sufficient when large fields are used, which encompass a significant volume of normal tissue around the tumor. However, with IMRT, treatment margins are usually smaller and the dose gradients between tumor and normal tissues steeper, to facilitate dose escalation. There is increasing awareness that day-to-day anatomic changes in the tumor and normal tissues relative to the adjacent bones may contribute to errors in IMRT dose delivery, and in turn a higher risk of tumor recurrence or treatment complications.

Recent advances in imaging, soft tissue deformable modelling and rapid IMRT planning have made it possible to explore the impact of anatomic changes on the dose actually delivered to tumors and normal tissues during a course of fractionated radiotherapy, and the potential benefit of adaptive re-planning at strategic points during treatment to offset the negative consequences of these changes. Evidence is accumulating from dosimetry studies in head and neck cancer, prostate cancer and cervix cancer to support the development of practical approaches to implementing adaptive radiotherapy in the clinic. Many questions remain about optimal daily online soft tissue imaging, how best to identify patients likely to benefit from re-planning, the optimal time for re-planning and the human workflow issues surrounding re-contouring and quality control. However, the benefits to some patients may be substantial. Furthermore, the development of robust approaches to anatomic adaptive re-planning will set the stage for biologic adaptation in the future, with the promise of even greater improvements in tumor control and cure rates.

_______________________________________________
H. Molavian (Waterloo)
New insight on cell metabolism and the relationship between hypoxia and acidity in solid tumors in vivo

The microenvironmental parameters affecting solid tumors in vivo can be employed to unravel the cell metabolism and help in the design of effective therapies. Specifically, the experimental measurement of microenvironmental parameters with respect to a single blood vessel and on the micrometer scale, may provide unique insight into the underlying cell metabolism. In this talk, I will show how we can combine the experimental results of pH and pO2 measurements from a single blood vessel using a mathematical model to obtain the cell metabolism in vivo. I will use this cell metabolism to describe the lack of correlation between hypoxia and acidity and to explain the heterogeous shapes of the pH and pO2 profiles on the micrometer scale observed by G. Helmlinger et al. (Nature Medicine 3, 177 (1997)). Finally, I will discuss possible mechanisms which might be responsible for the distinct behavior between pH and pO2 profiles.

_______________________________________________
L. Munn (Harvard)
Modeling tumor blood vessel dynamics

In wound healing, there is in influx of immature, angiogenic blood vessels that rapidly provide blood flow to the damaged tissue. The resulting, initial network is not very efficient, because it has not been subjected to the normal remodeling processes that lead to the well organized, optimal networks of adult tissue. During resolution of wound healing, endothelial cells in the immature networks respond to blood forces, re-organizing locally to optimize the network globally. Some segments dilate, while others are pruned, and eventually, a stable configuration is reached which is then supported and maintained by perivascular cells. In contrast, tumor blood vessels are chronically immature, probably due to the high levels of VEGF in the microenvironment (VEGF is initially high in wounds, but decreases during resolution). Interestingly, many anti-VEGF therapies can cause maturation or stabilization of tumor blood vessels through a process resembling adaptive remodeling of networks in wounds. Unfortunately this effect is temporary, and very little is known about how it affects the distribution of blood flow and the transport of nutrients and drugs into the tumor. In general, remodeling depends on blood shear forces, transvascular pressure as well as growth factors such as VEGF.

We are developing a mathematical model that incorporates: i) lattice-Boltzmann calculations of the full flow field within the vasculature and within the tissue, ii) diffusion and convection of soluble species such as oxygen or drugs within vessels and the tissue domain, iii) distinct and spatially-resolved vessel hydraulic conductivities and permeabil- ities for each species, iv) erythrocyte particles advecting in the flow and delivering oxygen with real oxygen release kinetics, v) shear stress-mediated vascular remodeling. This mathematical framework, guided by multi-parameter intravital imaging of tumor vessel structure and function, will help us understand why normalization therapy is sometimes ineffective, and may suggest alternative strategies.

_______________________________________________
G. Ostheimer (MIT)
To Sleep or Die: Cell Fate after Chemotherapy

DNA damage induces cancer cells to arrest their cell cycle and then either repair the damage and proliferate, enter a state of permanent arrest (senescence) or apoptose. Using a combination of flow cytometry and high-content microscopy we have quantitatively monitored cellular signaling and phenotypic outcomes in response to DNA damage. We have quantified signaling within the DNA damage signaling pathways, the cell cycle machinery, the apoptotic machinery and the mitogen/stress activated kinase pathways. Partial Least Squares Regression of these data demonstrate a role for the mitogen activated protein kinase, Erk, in modulating the cellular response to DNA damage.

_______________________________________________
L. Sander (UMichigan)
Brain tumor invasion: cell motility, network structure, and the spread of glioma
Highly malignant brain tumors such as gliomas spread by invasion in the extracellular matrix of the brain. We have developed biomechanical models for cell motility and for the micro-rheology of the biopolymers that make up the ECM. We have preliminary modeling results on cell invasion in a non-linear elastic network, and we attempt to understand ECM alignment and contact guidance.

_______________________________________________
S. Sengupta (MIT)
Nanomedicine in Cancer Therapy

The talk shall address the future of nanotechnology in cancer therapy. Especially, I will discuss the parameters that go into the design of nanoparticles for use in cancer chemotherapy, such as the effect of size, shape, chemistry, and staging of drugs for optimal outcome.

_______________________________________________
S. Singh (McMaster)
Brain Tumour Initiating Cells: Why the Cancer Stem Cell Hypothesis matters to patients with brain tumours

Brain tumors are the leading cause of cancer mortality in children and remain difficult to cure despite advances in surgery and adjuvant therapy. We have discovered an abnormal stem cell that may drive the formation of brain tumors. These cancer stem cells are rare and represent only a small fraction of the whole tumor, but these cells alone may be entirely responsible for the continued growth of the tumor. Therapies that focus on killing the bulk of the tumor may miss the rare stem cell fraction, allowing the tumor to continue to grow and spread. Thus, therapies that focus on killing the cancer stem cell may provide better treatment and prognosis for patients with brain tumors. Although cancer stem cells have been identified in other tumors, most notably leukemia and breast cancer, no solid prospective evidence previously existed to suggest that brain tumors arise from stem cells. Our recent work represents the first isolation of cancer stem cells from the central nervous system, which we have termed brain tumor initiating cells (BTICs). The isolation of BTICs from human brain tumors emerged from our hypothesis that brain tumors arise from the transformation of a normal neural stem cell (NSC) that resides in the brain. However, what remains to be determined is the genetic or epigenetic nature of the transformation event. How does the genetic or epigenetic state of a brain cancer stem cell differ from that of a normal neural stem cell?

The identification of the BTIC has important implications for our current understanding of the origin of brain tumors. Our current diagnostic (ie. molecular pathology), genetic (ie. gene microarray) and therapeutic (ie. chemotherapy) approaches to brain tumors focus on every cell in tumor rather than the rare cancer stem cell, and this may explain the poor response of brain tumors to current treatment. Future therapies that target the BTIC may better halt the growth and propagation of the tumor.

Our future work will focus on characterization of the genetic abnormalities of the BTIC, and we will continue to search for better surface markers for the BTIC, allowing for specific targeting of this cell. Finally, through future epidemiological analysis, our current work may give insight into patient prognosis, as patients with a higher BTIC fraction may have a shorter survival and worse prognosis. These studies will form the basis for future trials of therapy directed against the BTIC.

_______________________________________________
J. Tuszynski (Cross Cancer Institute, Edmonton)
Targeting DNA repair pathways to improve cancer therapies
Co-authors: Khaled H. Barakat (Alberta), J. Torin Huzil (Alberta), Charles Dumontet (Université Claude Bernard Lyon 1), Lars Jordheim (Université Claude Bernard Lyon 1)

The key cellular target for both radiotherapy and many conventional chemotherapeutic drugs is DNA. DNA repair systems are becoming therapeutic targets to enhance the cytotoxicity of radiation and genotoxic drugs. I will outline the essential proteins that are involved in the repair of damaged DNA as a result of radiotherapy or chemotherapy. To develop dynamic pharmacophores that can be used in the creation of specific inhibitors of DNA repair we have employed rapid and economical computational techniques to filter large libraries of both pre-characterized and untested compounds, including the NCIDS, the Cambridge Structural Database and Drug Bank. Molecular database screening, coupled with secondary and tertiary protein structure analysis and prediction supports the molecular dynamics model development of two key DNA repair pathways, namely: Nucleotide Excision Repair (NER) and Non-Homologous End Joining (NHEJ). Evaluation of binding energies and several additional parameters from this screening procedure provides us with a reduced set of potential small molecule inhibitors for further validation. We have focused on ERCC1, an essential component of the NER pathway and hPNK, which is involved in both BER and NHEJ repair pathways. We have performed an extensive screening of chemical libraries for these targets. The top hits from this study were employed to create a promising pharmacophore compound to be used in constructing novel inhibitors, some of which are currently being validated experimentally.

_______________________________________________
D. Tyson (Vanderbilt)
An EMG model of cell cycle time variability in cancer developed from large datasets of single-cell measurements

Cancer cell heterogeneity with respect to several traits (e.g., proliferation, motility, metabolism) provides a basis for clonal expansion. Cell-to-cell variability in self-sufficiency from growth factors is especially relevant to cancer progression and likely contributes to drug resistance. While it is widely expected that cells within tumors (as well as cultured cell lines) have heterogeneous proliferation rates, quantitative data have been thus far unattainable because large datasets of single-cell measurements are required. We are implementing high-throughput time-lapse microscopic imaging technology to quantify the cell cycle duration (intermitotic time, or IMT) of thousands of individual cells per experiment in an automated fashion. As a initial model, we studied genetically related human breast gland cell lines with varying degrees of tumorigenicity (none, minimal, high). Single-cell IMT distributions, representative of the underlying heterogeneity of proliferative control within each cell line population, can be fit with high statistical confidence by an Exponentially-Modified Gaussian (EMG) model. In the non-tumorigenic cell line, both the exponential (E) and Gaussian (G) components of the EMG rapidly increase to infinity upon removal of serum (i.e. cells stop dividing). In contrast, in the ras-transformed minimally tumorigenic cell line G lengthens steadily (increase in mean and standard deviation) over a period of 24-36 hours after serum removal, whereas E remains static. In the highly tumorigenic cell line, both E and G are essentially unchanged whether or not serum is provided. These results suggest that the E and G components of the EMG model of cell division control may have separate biochemical correlates. Thus, while both E and G are serum-dependent in the non-tumorigenic cell line, G remains serum-dependent and E is relatively unaffected by removal of serum in cells with constitutively active ras. In the highly tumorigenic cell line, at least one additional deregulation event has occurred to eliminate growth-factor dependency altogether. We propose that high-throughput single-cell microscopy, coupled to mathematical modeling can provide fundamental insights into decision processes affecting cell division in cancer, as well as enable quantitative approaches to understand progression and drug-resistance dynamics fueled by heterogeneous self-sufficiency from growth factor.

_______________________________________________
J. Tyson (Virgina Tech)
Mathematical Models of the Molecular Networks that Regulate Cell Growth, Division and Death

A common theme underlying the diversity of oncological diseases is dysregulation of the normal mechanisms that control growth, division and death of mammalian cells. To understand how these control systems misbehave in cancer, we must first understand the principles underlying their normal purpose and function. Because the control mechanisms are quite complicated at the molecular level, mathematical modeling has proven very useful in refining our understanding of these basic aspects of cell physiology. In this lecture I will review what we have learned about these intracellular regulatory processes by mathematical modeling and how we might put this knowledge to work in gaining a better understanding of the origin and treatment of cancer cells.

Back to top