Abstract
An inability to reliably predict quantitative behaviors for novel combinations of genetic elements limits the rational engineering of biological systems. We developed an expression cassette architecture for genetic elements controlling transcription and translation initiation in Escherichia coli: transcription elements encode a common mrnA start, and translation elements use an overlapping genetic motif found in many natural systems. We engineered libraries of constitutive and repressor-regulated promoters along with translation initiation elements following these definitions. We measured activity distributions for each library and selected elements that collectively resulted in expression across a 1,000-fold observed dynamic range. We studied all combinations of curated elements, demonstrating that arbitrary genes are reliably expressed to within twofold relative target expression windows with ~93% reliability. We expect the genetic element definitions validated here can be collectively expanded to create collections of public-domain standard biological parts that support reliable forward engineering of gene expression at genome scales. One main goal of synthetic biology is to make the engineering of biology easier 1,2. DNA synthesis and assembly has progressed to the point where entire metabolic pathways, chromosomes and genomes can now be synthesized and transplanted 3-5. However, our capacity to rationally design increasingly complicated genetic systems as enabled by improvements in DNA construction methods has not kept pace 2,6. One of the greatest claimed barriers to efficient and scalable genetic design is the lack of standard parts that can be reused reliably in novel combinations 6,7. Many examples instead highlight, even within well-studied organisms such as E. coli, how seemingly simple genetic functions behave differently in different settings 8,9. For example, a prokaryotic ribosome-binding site (RBS) element that initiates translation for one coding sequence might not function at all with another coding sequence 10. If the genetic elements that encode control of central cellular processes such as transcription and translation cannot be reliably reused, then there is little chance that higher-order objects encoded from such basic elements will be reliable in larger-scale systems 6,11. Standard biological parts could, in theory, enable hierarchical abstraction of biological functions 1,2,12,13. The behavior of integrated genetic systems could then be represented via simpler models of individual elements and ultimately mapped to underlying genetic sequences whose encoded functions are dependent on a limited number of measurable or calculable intrinsic variables. Such abstraction of function seems necessary to manage biological complexity and to allow the engineering of increasingly sophisticated genetic systems 6,12,14. We engineered ~500 transcription and translation initiation elements that are compatible within a standardized genetic context, or expression operating unit (EOU), that enables predictable forward engineering of gene expression over a wide dynamic range. We characterized representative parts for each type by testing more than 1,200 part-part combinations to establish and validate functional composition rules while quantifying scores for part activity. From this data we also estimated the 'quality' of each part, a second-order statistic that represents the extent to which the activity of a part varies across changes in context 15. Our results demonstrate how, when combined with standardized transcription control elements, a more physically complex design for the control of translation initiation creates simply modeled parts enabling reliable forward engineering of gene expression. results Prioritizing part composition puzzles In related work, we systematically assembled and tested all combinations of frequently used prokaryotic transcription and translation control elements to quantify average part activities and also variation in activities as parts are reused in novel combinations 15. Here we focus on developing rules for a genetic layout architecture underlying gene expression cassettes that eliminate