Adam Margolin's Home Page

Cancer cells are determined by genetic mutations that cause the aberrant expression of oncogenes and tumor suppressor genes. Thus cancer cells can be discriminated from healthy cells based on a unique "molecular signature". A futuristic, yet conceivable, therapeutic approach would be to design intelligent, nanoscale devices that can be transfected inside of cells and can detect these discriminative molecular signatures and elicit a response, such as the release of a toxic agent, upon encountering the cancer cell. In contrast to traditional therapy approaches that seek to inhibit enzymatic activity in all cells, this approach would selectively target only the cancerous cells. Towards this goal, I am working on an engineered gene expression control system in which an arbitrary gene can be placed under the control of molecular logic gate that is activated only the presence of specific, cancer-related, molecules within the cell.

Figure Caption: The AND gate recognizes two oligonucleotides, I (blue) and II (red) and has four possible states. Three states exhibit no self-cleavage activity (inactive or output 0) owing to misfolded catalytic modules (boxed). Only the state with both oligonucleotides bound to the recognition module contains the properly folded hammerhead catalytic module that exhibits self-cleavage activity (active or output 1). b, Operation of a rudimentary therapeutic automaton implementing AND logic and analyzing two disease markers. Two upstream sensor (or YES) gates recognize two protein disease markers (ellipse and rectangle), become active (1) and release oligonucleotides (blue and red) through autocatalytic action. The two released oligonucleotides activate a downstream AND gate, and release an oligonucleotide (green) that activates a downstream drug delivery element.

 

 

Research Interest 2 -- Computational Deconvolution of Regulatory Networks in Human Cancer Cells

adapted from Basso, Margolin, et al. Nature Genetics, 2005

My second research interest focuses on the development of statistical methods to identify genetic regulatory networks that are disregulated in cancer cells. As a first step towards this goal, we have developed an algorithm that uses microarray data to infer which genes are most likely to exert direct regulatory influences on each other, and define putative regulatory circuits, as depicted in the figure on the left. Current efforts focus on defining a unified statistical framework, using data from gene expression measurements, DNA sequences, ChIP-on-chip, and orthologous promoters to identify regulatory networks disregulated in cancer cells, and to define potential therapeutic targets.