Specific RNA-Binding Proteins and Sequences Program Events in the Life of a Messenger RNA
Most of the genome's instructions that specify the characteristics of each cell are transmitted in the form of molecules called messenger RNA (mRNA). Each mRNA molecule carries a copy of a gene's DNA sequence from the genome to the cell's "translation" machinery, which decodes the nucleotide sequence to produce a specific protein. In addition to the nucleotide sequence that specifies its protein product, each mRNA molecule also carries mysterious sequences from the genome that program the key events in its life history—where in the cell it is to be translated into a protein sequence, when and at what rate it gets translated, and when it should be destroyed.

We are working to systematically define the sequence of events in the life of each RNA molecule, the molecular system that regulates them, and the sequences in each mRNA that encode this program. We have recently found evidence that hundreds of specific RNA-binding proteins act as critical regulators in a multifaceted molecular system that controls the fate of each RNA. The regulation of mRNA fates by these RNA-binding proteins has many parallels to the regulation of the production of the mRNA molecules by the DNA-binding transcription factors.

To investigate the molecular mechanisms by which the regulatory RNA binding proteins receive information and carry out their regulatory roles, we have developed genomic methods for quantitatively monitoring the interactions between every mRNA in a cell and the translation machinery, and tracking its eventual destruction. To study how RNA molecules move from the nucleus to the specific places in the cell where they are translated or destroyed, we are developing methods that use fluorescence microscopy to track the movement of individual RNA molecules in living cells. Using DNA microarrays, computational methods, and high-throughput sequencing, we have begun uncovering the sequences in RNA molecules that enable their recognition by specific RNA-binding proteins and thereby specify features of their life history.

Gene Expression Patterns as Molecular Portraits of Cancer
It has been known for a long time that faults in the gene expression program of cancer cells play a fundamental role in cancer pathogenesis. Characteristic patterns of gene expression might therefore be related to mechanisms of pathogenesis and perhaps to patterns of clinical behavior in human cancers. By using DNA microarrays to examine gene expression patterns in hundreds of human tumors, we found that these patterns can vary dramatically between tumors—even tumors that would classically be considered the same type.

The gene expression patterns, which we now can readily observe, provide distinctive molecular portraits of each patient's tumor and clues to the mechanisms underlying the biological differences between cancers. One important medical application of this work is the identification of molecular patterns that help predict the behavior of a tumor and thus help guide treatment decisions. Another is the identification of molecular signatures that might be detectable by a noninvasive test that could reveal the presence of a cancer at an early stage, when the prospects for successful treatment are good. We are now working to develop new tools to exploit the molecular characteristics of cancers for early detection, including development of small-molecule substrates for key enzymes involved in oncogenic signaling pathways, which could be used as diagnostic imaging agents.

Systematic studies of gene expression patterns in cells purified from diverse normal human tissues, by our lab and others, have shown that the number of distinct specialized cells in the human body is far greater than previously believed. Even cells that appear indistinguishable under the microscope can have strikingly different gene expression programs and molecular characteristics. These differences are likely to have important consequences for their behavior and significance in human diseases. Because of the inability to distinguish functionally distinct cells by conventional histopathology, however, the roles those differences play in health and disease are yet to be defined. To overcome this limitation and improve the precision and information yield of the histological methods commonly used for diagnosis and classification of cancer in clinical practice, we are developing imaging and analysis methods for precise molecular classification and characterization of cells and microenvironments in normal and diseased human tissues.