The long-term objective of our proposal is to greatly improve anti-cancer therapy and dramatically reduce the side effects of anti-cancer drugs by engineering sophisticated regulatory mechanisms that govern the release of a drug in a patient. Cancer is caused by multiple changes in the molecular profile of a healthy cell, and a mechanism for detecting these changes based on the notion of logic functions forms the basis of our approach for cancer diagnostics in individual cells. Existing cancer therapies usually focus on single biological cues with occasional but limited success. As a result, there is an increasing appreciation of the fact that multiple disease markers must be considered in order to achieve precise detection of a disease state. The crux of our proposal is a new method where the therapeutic agent samples multiple aberrantly-expressed molecular markers in individual cells and uses pre-programmed diagnostic criteria to control the response based on the combinatorial over- or under-expressed state of the markers. We focus on microRNAs, a family of molecules whose expression is often altered in cancer, as the biomarker cues whose expression profile ultimately controls the activity of the drug. We engineer hybrid RNAi/transcriptional regulatory networks where network elements interact with the intracellular microRNA cues and with each other to implement the desired control mechanism. For example, for selective targeting of HeLa cells we require expression of a pro-apoptotic protein only when microRNAs miR-21, miR-17, and miR-30a are over-expressed and miR-141, miR-142, and miR-146 are not expressed, all at the same time. In our experiments we use human cell culture as a test-bed for the proposed approach. The specific aims of our proposal (1) conjecture that a relatively small number of microRNA markers is sufficient for precise diagnosis of specific cancer lines at the single cell level, (2) hypothesize that a 'logic circuit' designed to detect the microRNA expression profile of a specific cancer cell line can efficiently destroy these cell, and (3) verify that the operation of this circuit inflicts minimal damage to healthy cells. To design the circuits, we use published microRNA expression data and data that we collect with our own measurements using single-cell readouts. We will develop computational tools to elucidate the best possible diagnostic criteria aimed at maximizing selectivity and specificity while minimizing the number of markers used, even in the presence of cell-to-cell variability in phenotypically uniform populations. The diagnostic criteria, i.e. the expression profile of selected markers, will guide construction of the circuits. The circuits use a combination of transcriptional and post-transcriptional regulatory elements that respond to the microRNA markers and together control expression of the "killer" protein. The first criterion for our system's success is the efficiency with which cancer cells are eliminated from a mixed culture containing cancer and healthy cells. The second criterion is the absence of side-effects on healthy cells in the mixture. We will analyze side effects by a combination of observations, including phenotypic assays, gene expression analysis and functional assays.