June 15, 2005 (Vol. 25, No. 12)

Developing Predictive Models of Human Disease with Systems Biology

Participants in Cambridge Healthtech’s conference, “Systems Biology: Validation of Multi-Variate Biology,” held recently in San Francisco, described multiple approaches to putting biology back into drug discovery and development as well as extending it to predictive medicine.

The term “systems biology” describes a range of enterprises and technologies aimed at understanding the operations and interrelations of complex biological systems to develop predictive models of human disease. Systems biology is driven by the realization that the reductionist approach to drug discovery, including the identification of drug targets through genomics, has failed to produce the anticipated windfall of new drugs.

While enterprises such as the Human Genome Project have generated vast amounts of information, presenters at the conference echoed the common theme that in order to inform drug development decisions efficiently, unique combinations of existing and new technologies incorporating both sophisticated computer modeling and biological information will be required.

BioSeek (Burlingame, CA) founder Eugene C. Butcher, Ph.D., says that the integration of three distinct approaches is required to practically apply systems biology concepts to drug discovery.

These include the “bottom up” approach, in which genomics, proteomics, and metabolomics data creates large data sets to form a basis for modeling; the “top down” approach in which computer models of diseases or organ system physiology built from existing data are applied to drug discovery and clinical trial design; and a third approach in which the disease-relevant cellular responses can be recorded directly from primary human cells in culture to model cell regulation in response to drugs.

BioMAP Systems

BioSeek has developed quantitative automated primary human cell systems modeling several pathological conditions including inflammation, autoimmunity, and cardiovascular disease. The company’s technology platform, BioMAP (Biologically Multiplexed Activity Profiling) Systems, replicate the cell and pathway interactions normally found in disease states.

Depending on their mechanisms of action, test compounds induce specific changes in protein expression, captured as readouts to produce a BioMAP profile, which is then compared to reference profiles in BioSeek’s database. Using its proprietary algorithms to characterize drug function, the company can characterize drug mechanisms of action and of off-target activities.

As a specific example, using four model systems containing primary human endothelial cells and peripheral blood mononuclear cells in which multiple signaling pathways associated with inflammation and immune responses were simultaneously activated, BioSeek scientists reported that they could detect and distinguish among a wide range of drugs covering not only inflammation, but also mechanisms of metabolic and cardiovascular diseases and cancer.

This breadth of disease coverage makes possible rapid “therapeutic area scanning,” in which unexpected therapeutic opportunities can be uncovered for novel or known compounds or drugs. It also allows the firm to screen compounds against hundreds or potentially thousands of drug targets simultaneously, while providing biological insights into their potential utility at the same time.

Dr Butcher says, “I believe this cell systems biology’ approach will complement or eventually even largely replace current target-directed discovery, which is just not working well as an avenue to drug innovation.

Dr. Butcher comments that, “Our goal was to bridge the perceived gap between omics and modeling to collect cell biology data incorporating the complexity of cell regulatory systems, and that could produce highly reproducible data.

“We developed robotic cell systems to measure and catalog multiple cell type responses in diverse disease-related environments and have collected responses to compounds in many disease states in our database.

“Statistical comparisons of quantitative BioMAP profiles and analysis of profile features allowed correlation of the observed induced biological effects with chemical structure and mapping of biological responses to compounds with similar chemical characteristics, as well.

“We can use this database to identify the mechanisms or targets of new compounds, as well as off target activities; for example, we have identified the targets of several compounds selected as cytokine inhibitors in standard cell-based assays.”

Because of its automation, throughput, and reproducibility, and the informatics that allow us to compare any new activity with hundreds of compounds and drugs already in the firm’s database.

Network Biology

Merrimack Pharmaceuticals (Cambridge, MA) focuses on developing drugs for cancer and inflammatory disease using its Network Biology drug discovery platform. This platform integrates experimental and computational biology with the goal of defining the ideal mechanism for modulating the biological networks that underlie disease.

Merrimack president and CEO Robert Mulroy explains, “Our fundamental approach is to generate a new type of data, or network data, that captures the quantitative contributions of multiple and parallel protein-protein interactions within functional protein networks and allows us to better understand the signal dynamics that govern cellular behavior.”

Merrimack’s network oriented approach is designed to overcome the barriers of traditional research by understanding potential targets within the context of the complex signaling environments in which they function.

At the core of the Network Biology platform is a set of integrated engineering and computational technologies that enable a high throughput biology research engine. Central to the company’s approach is the use of its own protein and antibody microarray technologies to generate network data.

Merrimack’s microarray technologies are robust, quantitative, and automated to allow the high throughput profiling of specific protein networks. The resulting proprietary data is then used to build detailed computational models of disease that can predict, screen, and validate targets.

“We have the ability to sort out which interactions are meaningful and which targets make sense in the context of multiple and often competing signalsand often contradictory data from publications.”

Merrimack has also been able to use its Network Biology platform to move upstream from individual networks to identify which of several competing pathways are the most critical to driving signaling within a given tumor. Mulroy further explains that Merrimack’s tools not only have the potential to help the company define and develop better drugs, but to enable a better diagnosis and the right mechanism based treatment in the doctor’s office.

Merrimack has used its microarray and computational approach to produce a quantitative model describing how various ligands such as EGF and HRG induced ErbB receptor activation and the resulting downstream signaling that regulates cell growth.

The activation of the ErbB receptor group induces structural changes that trigger ERK and AKT activation characteristic of tumor cell signaling. The company has developed several other network models in other cancer and inflammation pathways.

Merrimack’s lead product, MM-093, is currently in Phase II clinical trials in patients with moderate to severe rheumatoid arthritis. MM-093 is a transgenically-produced recombinant version of human alpha-fetoprotein, a protein produced by the fetus during pregnancy and associated with remission of autoimmune diseases in the third trimester of pregnancy.

Top Down Approach

Entelos’ (Menlo Park, CA) PhysioLab platform exemplifies the “top down” approach to biological modeling of complex disease processes.

PhysioLab platforms are computer-based mathematical models that integrate the major physiological systems involved in a disease in order to predict human clinical outcomes. Using information from multiple sources, including human clinical trial results, the company has developed in silico models for asthma, obesity, diabetes, and rheumatoid arthritis.

These platforms consist of a disease map that connects all the physiologic systems involved in a disease using equations that describe the relationships between components; a knowledge management environment in which researchers can define virtual patients; and virtual drugs, targets, and experimental protocols.

For example, the Obesity PhysioLab platform incorporates virtual systems involved in human metabolism and food intake control, and includes normal and overweight individuals with differing genetic backgrounds. By simulating changes on disease-relevant pathways, their effects on overall metabolism and weight are measured and extrapolated to likely outcomes in real patients.

Entelos announced in May, that it had successfully completed an in silico model of the non-obese diabetic (NOD) mouse, its first animal model, and the primary animal model used to study human type 1 diabetes.

Entelos president and CEO, James Karas, comments that, “Diabetes is a complicated disease that occurs over a long period of time and manifests itself in a variety of ways. We believe there are potential drug targets that haven’t been investigated yet. The approach to developing the type 1 diabetes model was to mathematically describe the general biology of a healthy system, then perturb it to create the disease.”

The result of a two-year collaboration between the American Diabetes Association (ADA) and the company, the NOD mouse model is expected to provide a way to “translate” what is seen in animals to likely human responses, for example, the influence of diabetes drug dosing and timing on safety and efficacy.

Entelos has also applied its virtual patient technologies to simulate the effects of patient heterogeneity on clinical trial results. By creating virtual patient populations representing the gamut of biological characteristics and drug responses seen in an actual clinical trial patient population, the company has been able to optimize clincial trial designs.

According to Entelos’ Karas, “We don’t have to build a perfect, whole human model. If we can take one year off of the time needed to develop a drug, we’ve succeeded.”

Contextual Biology

The “bottom up” approach to systems biology is exemplified by ongoing efforts at the Institute for Systems Biology (ISB; Seattle). ISB’s president and founder, Leroy Hood, Ph.D., believes that systems biology will be required to move medicine toward the “predictive, preventive, and personalized.”

Dr. Hood points out that the Human Genome Project revealed that humans differ from one another by six million DNA variations, a few of which predispose to certain diseases such as cardiovascular disease and some cancers.

But, he says, without a context in which to place the expression of such genes, knowledge of how to prevent or treat disease remains unknown. The ISB enterprise approaches the issue of contextual biology by using all of the hard and software platforms produced by the “genes to drugs” vision and applying them to the generation of clinically relevant knowledge that will support the development of new treatments.

ISB focuses on making substantial improvements in these technologies and combining them with biological assays and significant computer-based computational and analytical fire power.

For example, ISB’s approach to understanding the biology of prostate cancer, underwritten by a $23.5 million grant to ISB, the University of Washington and the Fred Hutchinson Cancer Research Center, is to use a variety of available technologies to identify markers for prostate cancer stratification (that is, to allow early identification of the transition from androgen dependence to androgen independence, for example), to identify and characterize novel genes as diagnostic markers, and to ultimately understand the androgen-regulated network in this form of cancer.

ISB uses monoclonal antibody production, microarray gene expression analysis, comparative proteomics using mass spectrometry, a three-dimensional tissue culture system comprised of epithelial progenitor cells, protein factors derived from prostate gland mesenchymal cells, extracellular matrix, and animal studies to model biological systems controlling the development of this cancer.

ISB scientists believe its approach will allow the identification of different types and stages of prostate cancer early as well as understand the relationships between genetic and environmental factors impacting their development.

Ultimately, ISB’s goal is to apply its new tools in development to profiling diseases in patients early enough for prevention or intervention, for example, point of care messenger RNA and proteome analysis. The “bottom up” approach to systems biology is exemplified by ongoing efforts at the Institute for Systems Biology (ISB; Seattle). ISB’s president and founder, Leroy Hood, Ph.D., believes that systems biology will be required to move medicine toward the “predictive, preventive, and personalized.”

Dr. Hood points out that the Human Genome Project revealed that humans differ from one another by six million DNA variations, a few of which predispose to certain diseases such as cardiovascular disease and some cancers.

But, he says, without a context in which to place the expression of such genes, knowledge of how to prevent or treat disease remains unknown. The ISB enterprise approaches the issue of contextual biology by using all of the hard and software platforms produced by the “genes to drugs” vision and applying them to the generation of clinically relevant knowledge that will support the development of new treatments.

ISB focuses on making substantial improvements in these technologies and combining them with biological assays and significant computer-based computational and analytical fire power.

For example, ISB’s approach to understanding the biology of prostate cancer, underwritten by a $23.5 million grant to ISB, the University of Washington and the Fred Hutchinson Cancer Research Center, is to use a variety of available technologies to identify markers for prostate cancer stratification (that is, to allow early identification of the transition from androgen dependence to androgen independence, for example), to identify and characterize novel genes as diagnostic markers, and to ultimately understand the androgen-regulated network in this form of cancer.

ISB uses monoclonal antibody production, microarray gene expression analysis, comparative proteomics using mass spectrometry, a three-dimensional tissue culture system comprised of epithelial progenitor cells, protein factors derived from prostate gland mesenchymal cells, extracellular matrix, and animal studies to model biological systems controlling the development of this cancer.

ISB scientists believe its approach will allow the identification of different types and stages of prostate cancer early as well as understand the relationships between genetic and environmental factors impacting their development.

Ultimately, ISB’s goal is to apply its new tools in development to profiling diseases in patients early enough for prevention or intervention, for example, point of care messenger RNA and proteome analysis. The “bottom up” approach to systems biology is exemplified by ongoing efforts at the Institute for Systems Biology (ISB; Seattle). ISB’s president and founder, Leroy Hood, Ph.D., believes that systems biology will be required to move medicine toward the “predictive, preventive, and personalized.”

Dr. Hood points out that the Human Genome Project revealed that humans differ from one another by six million DNA variations, a few of which predispose to certain diseases such as cardiovascular disease and some cancers.

But, he says, without a context in which to place the expression of such genes, knowledge of how to prevent or treat disease remains unknown. The ISB enterprise approaches the issue of contextual biology by using all of the hard and software platforms produced by the “genes to drugs” vision and applying them to the generation of clinically relevant knowledge that will support the development of new treatments.

ISB focuses on making substantial improvements in these technologies and combining them with biological assays and significant computer-based computational and analytical fire power.

For example, ISB’s approach to understanding the biology of prostate cancer, underwritten by a $23.5 million grant to ISB, the University of Washington and the Fred Hutchinson Cancer Research Center, is to use a variety of available technologies to identify markers for prostate cancer stratification (that is, to allow early identification of the transition from androgen dependence to androgen independence, for example), to identify and characterize novel genes as diagnostic markers, and to ultimately understand the androgen-regulated network in this form of cancer.

ISB uses monoclonal antibody production, microarray gene expression analysis, comparative proteomics using mass spectrometry, a three-dimensional tissue culture system comprised of epithelial progenitor cells, protein factors derived from prostate gland mesenchymal cells, extracellular matrix, and animal studies to model biological systems controlling the development of this cancer.

ISB scientists believe its approach will allow the identification of different types and stages of prostate cancer early as well as understand the relationships between genetic and environmental factors impacting their development.

Ultimately, ISB’s goal is to apply its new tools in development to profiling diseases in patients early enough for prevention or intervention, for example, point of care messenger RNA and proteome analysis.

Previous articleIntegrated Workflow for Development Testing
Next articleAmerican Thoracic Society – Gene Express