Optimization and automation of the ligase cycling reaction

Modern approaches in the field of biology, especially the field of synthetic biology, are based on design-build-test-learn (DBTL)-cycles to, e.g., generate optimized genetic constructs for various applications. The overall speed of one cycle mainly depends on the ability to physically assemble DNA with a high efficiency in a short time period. This is achievable by automation approaches. Several automated DNA assemblies are described in the literature but are related to laborious in silico planning, are usually limiting the reusage of the DNA parts or are introducing scars into the final sequence. To overcome those restrictions and to enable a rapid automation, an easy-to-plan and scar-less assembly method has to be utilized. The ligase cycling reaction (LCR) is the most promising candidate and is the scope of this thesis. For the LCR, no construct specific DNA parts have to be designed and no scars are incorporated into the desired construct. Single-stranded bridges made of DNA (bridging oligos (BOs)) are utilized to specify the order of up to 20 parts in a one-pot reaction. A thermal cycling protocol enables the strand separation of the DNA parts, the annealing of the BOs and the in vitro ligation. The desired product is finally derived by transforming the LCR mixture into Escherichia coli (E. coli). Due to the benefits of this assembly technique, the LCR was already implemented from Robinson et al. on a robotic platform in 2018. According to the authors and the results presented in this thesis, the assembly efficiency for some LCR reactions is low. This issue is associated with a tremendously increased effort to obtain the desired sequence. The reasons for low efficiencies are determined within this thesis to optimize the LCR. Furthermore, an alternative in vitro method for the cumbersome in vivo approach is developed and is based on a cell-free system to screen for correctly assembled constructs. In the end, a workflow for an automated LCR is designed and initially validated on a robotic platform. Overall, up to ten DNA parts were assembled in a one-pot reaction by applying the LCR protocol described in the literature by de Kok et al. (2014). Nevertheless, some assemblies were not successful and an improved LCR assembly protocol was developed. In contrast to the literature, to omit the secondary structures inhibitors dimethyl sulfoxide (DMSO) and betaine had tremendously increased the efficiency and the total number of colonies for the assembly of various plasmid designs. Furthermore, to shift the annealing temperature to the activity optimum of the utilized ligase was beneficial. The new LCR protocol was implemented in an automated assembly and plating workflow for a robotic platform to theoretically build 96 DNA constructs within 19 h and needs to be further validated. To support the automation process, a software for the BO design of combinatorial LCR assemblies was build. As an additional scope, an in vitro LCR method was designed and validated by using a cell-free system for the test-learn steps of the DBTL-cycles. In comparison to the automated in vivo workflow, ca. 5× more LCR assemblies are screenable within the same time frame of 19 h and the same hardware setup. In future, the optimized LCR protocol and the robotic workflow for the in vivo LCR can be adapted for the in vitro LCR approach and represents the method of choice for automated high-throughput DBTL-cycles of genetic switches and circuits