The Power of Collaborations: Engineering Human Tissues to Facilitate

Drug Discovery and Development

In order to increase the efficiency and productivity of the drug development process, a broad in vitro platform of

engineered human “tissue” is proposed to enhance (i) evaluation of toxicity of NCEs, (ii) the drug discovery process

itself, and (iii) studies of disease mechanism and patient stratification.

Goal: Bridge the void between preclinical models and clinical trials.

The problem: “The drug discovery process is too costly and inefficient.” too numerous to cite

• Animal models have limited predictive value as a pre-clinical platform for drug discovery. This may be due to molecular differences in protein structure or larger scale differences in cell and tissue physiology between mouse (e.g.) and man.

• In vivo models are also very labor intensive and expensive to study and are not suitable for HTS approaches.

• In addition, current in vitro testing strategies (using mouse or human cells lines, or primary cells from multiple donors) while compatible with HTS, display limited fidelity to the function of complex tissue in situ. Primary cells, while more accurate indices of in vivo function, introduce too much variability into a screen.

• “A principal component of the failure [of in vitro systems] results from our lack of understanding of, and inattention to, how to culture cells specifically so that they represent their in vivo counterparts.” Bhadriraju (Drug Disc Today 2002. 7:612)

• “The value [of cell-based screening] in predicting clinical response to new agents is limited. This lack of predictability of commonly employed 2D cellular assays is attributable to the fact that they do not mimic the response of cells in the 3D micro- environment present in a tissue or tumor in vivo.” Ebner (J Biomolec Screen. 2004. 9:273)

Examples of failures of current systems

A. Mouse is not always a good model for human disease A Role for the CHC22 Clathrin Heavy-Chain Isoform in Human Glucose Metabolism. Vassilopoulos Sci. 2009. 324: 1192.

CHC22, an isoform of the membrane protein clathrin that plays a central and previously undescribed role in glucose trafficking in humans, isn't even expressed in mice.

B. Isolated human cells may not be a good model for organized human tissue Human occludin is a hepatitis C virus entry factor required for infection of mouse cells. Ploss Nature 2009. 457: 882

Because occludin (a tight junction protein) is lost on dissociated cells, even human cell lines are an inappropriate model to study Hep C infection.

The proposed strategy: • This proposal is based on promoting interactions,

across both academia and Biopharma, which address the critical need for more efficient drug discovery and evaluation.

• Specifically, existing (based on previous work at AMRF) and planned academic collaborations in bioengineering and tissue modeling will be utilized to faithfully mimic critical functions of a broad range of human tissues in vitro.

Alternatively, Pharma could pursue this on a tissue-by-tissue basis with individual labs. This approach would forfeit the power of the collaboration and most likely result in inefficiency due to each group “rediscovering” the key steps in tissue construction rather than sharing their insights with the group.

Strategy (ii): • Utilizing primary human stem or stem-like cells and

creating biologically relevant matrices and environments, these platforms will be designed to represent the tissue niche and preserve tissue function far more accurately and reproducibly than in vitro systems currently in use.

• These models will be constructed in a miniaturized format to accommodate HTS, and can also incorporate increasing cellular and structural complexity over time, as needed.

• An additional advantage of this platform is that it serves as a catalyst to facilitate academia-pharma interactions, where the directions of the university-driven platforms will evolve to reflect the specific needs of individual pharma partners. It is also very cost effective compared to a similar strategy developed internally.

It is proposed that all of the critical stages of drug discovery - target identification and validation, efficacy of drug candidates, toxicity and off-target effects of drug candidates, even mechanism of disease - will be facilitated by application of these platforms.

How is the approach unique?The inefficiency of drug discovery is recognized as a fundamental problem in Biopharma. Tissue engineering is being studied in many academic and some biotech settings as a means to address this concern. Our approach differs from those currently being pursued in several critical ways:

• It addresses the significant complexity of building these models by tapping the power of collaborations. It brings together the best minds in bioengineering, cell and molecular biology, stem cell biology, immunology, and intestinal microbiology (for the intestinal platform, as an example) to tackle a problem that simply can not be addressed by one lab. In short, it builds powerful translational research teams.

• The value and power of the plan are based on assembling outstanding scientists with low cost to Pharma for salary or infrastructure. Financial support is limited to actual costs to perform experiments and manage collaborations, and to an “investment component” that would be performance-based and milestone-driven.

• It recognizes the need for a uniform and reproducible platform to assess large compound libraries over time. This can only be accomplished with a stem cell (or iPS) based platform. This also makes the platform amenable to HTS.

• It stratifies the platform into two components: (i) simpler tissue constructs to evaluate toxicity and (ii) more complex tissue constructs to evaluate drug function and mechanism of disease. This allows it to be built out logically and sequentially, with distinct short term and long term milestones, and to appeal to universal, as well as more focused, needs of Biopharma.

Examples of the strategy: (a) Two programs had been initiated at AMRF to

develop natural or bioengineered 3D in vitro systems:

• Melanoma - Human melanoma cells grown on human skin give rise with very high efficiency to “tumors” that are virtually indistinguishable in appearance from those that develop in patients. Introducing vascular endothelium as substrate for skin allows for study of directed transmigration/ metastasis.

• IBD - Human intestinal stem cells (SC) are seeded onto bioengineered matrix to recreate the specific architecture of the large or small intestine. Immune and other cell components can be integrated to increase the fidelity of the structures.

IBD team• Clinical evaluation, patient stratification, clinical

trails, systems biology - D Podolsky, S Targan, S Hanauer, R Xavier

• Development of mouse models informed by human GWAS - C Terhorst, Targan, Xavier

• Intestinal bioconstructs - B Langer, H Clevers, C Reinecker

• Studies of the Biome - J Gordon, D Relman, D Kasper

• Immune pathways and inflammation – Terhorst, Targan, Kasper

Proposed platform: Human in vitro models for health and disease

LIVER

SKIN/MELANOMA

INTESTINAL EPITHELIUM/IBD

ISLET/DIABETES

KIDNEY NERVES HEART

To assess:

drug efficacy(platform 2)

drug toxicity(platform 1)

•Develop human tissue model from stem cells (or tumor cells)•Reproduction of environmental stress (e.g., physical destruction, immune cells, cytokines, bacterial stimulation, etc) that may promote transition from normal homeostasis to disease

Human in vitro model of intestinal epithelium

Epithelium

Lamina propria

Submucosa

Dendritic cell

T lymphocyte Basement membrane

Intestinal stem cells

Neuron

Porous scaffold +ECM Hydrogel

Muscularis mucosae

Sphere or cylinder ?

Lumen in or out ?ECM targeting the stem cells at the base of the crypt?

Addition of goblet cells? Part of SC repertoire.

1) 2D representation of an engineered colonic fold

Components in red represent the basic structure which can be made more complex over time

2D and 3D representation of section of the engineered in vitro structure

Questions

2) Design of miniaturized 3D porous scaffolds forming circular folds and imprinted with ECM

(b) This approach could be extended to other diseases for broad use in drug discovery.

• Other possibilities based on prior AMRF affiliations include breast CA (Joan Brugge, Harvard), cardiac tissue (Bob Langer, MIT) and neural tissue (Ken Kosik, UCSB).

• Efforts to develop functional islets could be pursued in collaboration with the JDRF.

• Programs would not be limited to previously contacted investigators or collaborations that we have already initiated. For example, Sangeeta Bhatia (MIT) would be a logical candidate with whom to develop a liver system.

• (c) Summary of platforms: platform 1: Tissue models directed towards toxicity studies: kidney, liver, neuro, heart, lung, etc. “Simpler” systems; level of complexity can be increased over time. Maintaining optimal cell viability along with a critical function, e.g., cardiac muscle contractility, may be sufficient to effectively address more subtle aspects of toxicity. platform 2: More complex tissue constructs developed for drug discovery and study of disease mechanisms. Tissues can “move” from platform 1 to 2 as needed.

Moreover, in addition to determining direct hepatotoxicity of a drug candidate on a liver “bioconstruct”, effects of drug metabolites (using conditioned medium from [liver + drug]) could be tested on the other systems in the platform, providing an assessment of “systemic” toxicity.

• Taken together, this platform is envisioned to evolve into a “disarticulated” human model.

• (d) A bioinformatics/ systems biology guided approach, perhaps enhanced by in silico strategies, will be utilized to interrogate the data in a meaningful manner.

• (e) It is anticipated that the original POC for drug discovery would be a limited validation screen using perhaps 100 compounds containing examples with (i) known efficacy in the disease, as well as clinical trial failures, e.g., (ii) efficacy in mouse but not in man, (iii) documented off-target effects, etc. The goal will be to determine the efficacy of this model as a predictor of drug function (and dysfunction) in patients.

Paths to development of the BHTP

1. Foundation funding

2. VC support for virtual company

3. Industry support (as single entity or as consortium with other Pharma)