We will build systems that allow greatly improved chemically-tuned control of gene expression in mammalian cells, and make them available to other researchers. Induction of gene expression on demand is a mainstay of modern biology and cancer research. Nowadays, this is commonly accomplished by one particular approach, induction by tetracycline analogs of gene expression driven by Tn10 TetR derivatives. The TetR derivatives and doctrine for using them date from the early 1990s. These 20+ year-old methods continue to aid research on oncoproteins, tumor suppressors, cell death proteins, cell cycle proteins, and other key proteins and RNAs in cells and animals. But these approaches suffer from serious limitations. First, systems in common use operationally produce all-or-nothing induction, rather than tunable expression. Second, TetR is the only well developed chemically-inducible DNA binding moiety in wide use, and, as a consequence, investigators can typically only conditionally induce one gene product or trigger one event. Finally, these systems do not use closed- loop or feedback control. Induced gene expression is thus subject to significant cell-to-cell variation, much of it due to intracellular differences in gene expression capacity, G. A number of developments now allow us to build systems not subject to these limitations. Consideration of feedback control in bacterial systems, better understanding of artificial repression by prokaryotic repressors in eukaryotes (which we developed in the 1980s), and work in yeast has enabled systems that offer tunable induction buffered against cell-cell variation. Wholesale bacterial DNA sequencing has revealed 1000s of new TetR family repressors to make new proteins. Cas9/ CRISPR approaches now facilitate experimentation to optimize performance, by making it possible to compare expression from different test constructs integrated in single copy at the same mammalian genomic site. In this pilot phase of work, we will concentrate the risks of using lessons from yeast to build better controllers of mammalian gene expression. We will develop and distribute to other researchers at least three (3) new chemically tunable mammalian controllers that respond to different small-molecule ligands. We expect that these controllers will become broadly used in cancer research, enabling (among many examples) a) finer control of timing and amount of oncoproteins and other key proteins in cells and animals b) tighter definition of threshold levels of proteins that cause biological effects c) optimization of timing and level of expression of multiple transcription factors to best generate specific differentiated cell types d) construction of cell lines with controlled oncoprotein expression to be used in screens for anticancer drugs.
Much cancer research today relies on 20+ year-old technology scientists use to turn genes on and off. Here, we will develop controllers that allow better regulation of genes in cultured cells and lab animals, and make them available to other researchers. The controllers will allow researchers to adjust gene expression between 0% and 100%, with reduced variation, and to control multiple genes independently. If successful, our work will enable new kinds of experiments, and help make many current types of experiments more precise, enabling researchers to find answers to scientific questions more quickly.
| Brent, Roger; Boucheron, Laura (2018) Deep learning to predict microscope images. Nat Methods 15:868-870 |
| Andrews, Steven S; Brent, Roger; Balázsi, Gábor (2018) Transferring information without distortion. Elife 7: |
