Research Overview

A different approach to disease

Cancer has proven a difficult disease to achieve significant long-term advances in patient survival; improvements in survival are often measured in months. Diabetes has not fare much better. My laboratory has undertaken a genetic and drug screening approach targeting cancer and diabetes utilizing the fruitfly Drosophila . Our basic approach has been to use the advantages of the fly to take a whole animal approach to disease. We have been using an integrated approach: genes and drugs identified in flies are then brought to rodent and ultimately to clinical trials; sequencing and histological data from humans are then brought back to our fly models to allow us to develop increasingly sophisticated dipteran tools.

MEN2: My laboratory has developed a Drosophila model for Multiple Endocrine Neoplasia Type 2 (MEN2) and related MTC. Targeting oncogenic Ret to the eye, we phenocopied many aspects of the human syndrome (Figure). With this model in hand, we utilized a classical genetic modifier screen to identify 140 factors that function in the oncogenic process (Figure). We used this data to predict two candidate human susceptibility loci that are commonly deleted in MEN2 patients with secondary pheochromocytomas.

Adult fly eyes, Left; a normal eye, Right; An eye expressing an oncogenic (cancer-causing) form of Ret. The eye is small and 'rough', reflecting defects in the underlying


One regulator of oncogenic Ret was Csk, the major inhibitor of Src activity. Src is perhaps the original oncogene, and has recently been implicated in metastasis. Working in flies, we have found that activating Src sets up a 'metastatic boundary' cells that migrate away specifically from the tumor's edge (Figure); we have implicated multiple factors in this 'metastasis'. Working with pathologists, we have established evidence from histological sections that human solid tumors share many of these same molecular/spatial aspects present in our fly models (Figure). Currently, we are using human tumor sequencing data to alter multiple oncogenes/tumor suppressors with the goal of designing more sophisticated models of human tumorigenesis.

Fly Src-mediated 'tumors' in the wing epithelium (left) lose E-cadherin (green) at tumor edges (brackets); human Squamous Cell Carcinomas (right) share this loss of E-cadherin (brown).


We have established a collaboration with Tom Baranski's laboratory to create a model system for diabetes. Flies placed on a high-carbohydrate diet demonstrate a broad range of defects that share important properties with human diabetics: hemolymph glucose levels rise to levels similar to uncontrolled diabetics and pathways important in human diabetes progression modify diet-mediated fly phenotypes.

Our efforts are directed at three aspects of diabetes. First, our histology and lipid mass spec results indicate that changes in the lipid profiles of our 'diabetic' flies phenocopy changes in humans (Figure). Second, our genetic screens have identified 318 modifiers of the media-induced phenotype. Finally, we are screening for lead compounds that reduce the effects of a high carbohydrate diet. Our goal is an effective whole-animal approach to understanding and treating diabetes and metabolic syndrome.

Lipid profiles of flies fed a high sugar diet.

Drug screening

Recently, my laboratory has developed a novel method of high-throughput drug screening using our fly models, robotics, and compound libraries. Through our feeding paradigms, we provided whole animal validation (Figure) that helped identify ZD6474 as a useful tool for treating MEN2 patients; these trials are now entering Phase III and hold promise as the first approved chemotherapeutic for MEN2 . We have now expanded these efforts to other diseases including metastasis. Importantly, the inhibitors we identified in flies show efficacy in standard mouse models of oncogenic growth and metastasis. With these whole animal approaches, we hope to identify lead compounds that maximize efficacy and bioavailability while minimizing whole animal toxicity.

Feeding flies the chemical kinase inhibitor ZD6474 led to rescue of the Ret-mediated tumorigenic phenotype.

Patterning of the Drosophila eye

One of the fundamental interests of the laboratory is the basic understanding of epithelial patterning. How does an initially random collection of undifferentiated cells mature into a precise and functional organized epithelium? The developing Drosophila eye is an elegant model for studying epithelial patterning and, incidentally, is one of nature's most beautiful structures (Figure). Once the early pattern of photoreceptors are laid down, a progressive stepwise program of recruitment gathers the other 13 cells required to create the core of each unit eye or 'ommatidium' Remaining is a 'sea' of undifferentiated interommatidial precursor cells IPCs): these IPCs will differentiate as 2 cells that form an interweaving hexagonal lattice around the ommatidia, and the rest are killed off by PCD to tighten the pattern. We have used genetics, biochemistry, histology, laser ablation studies, disc culturing, microarrays, etc.

The adult fly eye.

Wolff and Ready (1991) and Reiter et al (1996) demonstrated that mutations in rst led to a failure of IPCs to move into their correct niches, leading to a loss of patterning and cell death. We have closely examined the role of Rst, a transmembrane adhesion molecule expressed in IPCs. We find that it directs hexagonal pattern in the pupal retina by binding to the transmembrane ligand Hibris on the surface of cells within the neighboring ommatidial cores. Our results suggest a model in which the drive to maximize Rst/Hbs binding drives cells into their proper niche.

We are currently testing whether simple adhesion is sufficient to direct cells into a honeycomb pattern through experiments and through computer modeling. Implicit in this model is the idea that adhesion, not signal transduction, is paramount. However, we have identified several candidate genes that might mediate a signal; their knockdown phenotypes are consistent with this possibility. For example, we have tied together Rst and Dpp/Tkv: loss of the latter yields a patterning defect, and loss of the former prevents normal progression of Mad activity. Other factors including fly orthologs of CD2AP and ZO-1 are also active in the patterning process. We are currently working to understand whether Hbs/Rst really sends a signal, and what this means in terms of patterning. As we further explore the developing fly eye, we are beginning to integrate others' and our work on adhesion, signal transduction, and cell biology to achieve a more complete and useful understanding of the mechanisms that direct epithelial patterning.