Optimizing Multi-Gene Biological Systems Using High-Throughput DNA Assembly, Sequencing, and Model-Guided Search Strategies

We have developed a framework to modularize and re-optimize massively multi- gene pathways or biological systems that increases the size and complexity of genetic systems that can be engineered. Engineering and optimizing the performance of large multi-gene systems requires combinatorially tuning gene expression through many design iterations. Here, new high-throughput DNA assembly strategies were developed that leverage advances in DNA sequencing to speed this process and also enable statistical combinatorial design for targeted searches through the high-dimensional space of possible genetic designs. We demonstrated these strategies developed here by optimizing a pathway of 16 genes. This 16-gene nitrogen fixation (nif) gene cluster is entirely modular and synthetic (â??refactoredâ?) with ~100 parts. The synthetic nif pathway encodes
16 essential genes that have tightly coordinated action to convert atmospheric nitrogen into ammonia, a critical process for global agriculture productivity. Using our novel assembly strategy, we constructed and sequenced a library of >55,000 unique nif transcriptional units (TUs). This is a 100-1,000-fold improvement in library size over our and comparable previous studies. Each TU is composed of characterized genetic parts (5-
7 each), including one refactored nif gene. These nif TU modules can be directly assembled into whole 16-gene, 100-part permuted cluster designs (~1045 unique cluster designs possible from this library). Additionally in this study, we sequenced approximately 7.5 million nif TUs to characterize DNA assembly error modes and quantify their frequencies. Next, from the library of nif modules we built and tested systematically permuted refactored clusters to investigate optimizing nitrogen fixation activity by tuning expression within the refactored cluster using characterized biological parts. Combinatorial design and statistical design of experiments were applied to constrain the part substitutions integrated into the cluster designs fabricated and tested in each round. Together, these methods enable model-guided redesign and optimization of genetic system performance with minimized DNA assembly required (~10-fold reduction in cloning). Looking ahead, we believe the assembly and optimization strategies demonstrated here will accelerate transferring complex functions to engineered systems and dramatically expand the complexity of biological systems that can be written, engineered, and optimized.