Although uncovered nearly three decades ago
, the high efficiency of S. cerevisiae homologous recombination is only beginning to reveal its full potential for the assembly of large DNA constructs. In vivo assembly in yeast is predicted to have a large impact on laboratory practice, ranging from simple plasmid construction to engineering of complex pathways via automated high-throughput strain construction
[7, 8]. Despite those promising prospects, in vivo assembly has not yet become a standard technique in most academic laboratories. This offers unique possibilities for standardization and, simultaneously, for further optimization. While reported efficiencies of correct assembly of larger (over 15 kb) DNA constructs do not exceed 70%, efficiencies of 95% were reached in the present work for the assembly of a 21 kb construct
[8, 9]. Physical separation of essential elements of the plasmid backbone contributed to a strong reduction of the frequency of plasmid mis-assembly.
The high efficiency obtained with relatively short 60 bp overlaps demonstrates that, in contrast to practices and claims from recent reports
[18, 34], longer overlaps are not essential for efficient in vivo assembly. This conclusion is supported by earlier studies in which 30 to 60 bp homologous sequences were shown to lead to high recombination efficiencies in S. cerevisiae[32, 35]. Although we anticipated that the high GC content, and therefore optimal binding properties, of the optimized SHR-sequences contributed to the high assembly efficiency obtained with the present platform, our results clearly indicate that the nature of these SHR-sequences is not an essential factor for achieving efficiencies above 90%. Therefore the increase in efficiency compared to previous studies essentially originates from the implementation of a backbone-free approach, and more specifically in the physical separation of the genetic elements on a plasmid assembly that are essential for its propagation and selection in the recipient yeast cell. Earlier studies have shown that reassembly of the backbone could make up to 95% of the obtained clones
. Placing the essential yeast elements on different fragments flanked by SHR-sequences and co-transforming them to S. cerevisiae reduced erroneous assemblies by plasmid regeneration by a factor of at least 100 (Figure
1), thereby substantially increasing the fidelity of in vivo assembly. Other yet unidentified factors, such as yeast strain-dependent efficiency of homologous recombination, could also have contributed to the high efficiency of assembly in the present platform and should be considered for further development of the method.
A potential downside of the proposed system was the inherent increase in the number of fragments to be transformed. However, in a standardized transformation protocol, assembly of a plasmid from 16 fragments still generated hundreds of clones (data not shown), which is more than sufficient for metabolic engineering purposes. This result indicates that increasing the number of fragments is not a serious limitation and that use of two survival elements instead of one single backbone does not have a high impact on the overall transformation efficiency.
The considerable potential of S. cerevisiae for pathway assembly has been recently proposed
[8, 9] for short pathways up to eight genes. To obtain highly productive and balanced synthetic pathways it is evident that finding the optimal combination of (heterologous) genes and expressing them at the right levels is essential
. Combinatorial approaches are therefore necessary and hundreds to thousands of constructs carrying different alleles expressed behind various promoters will have to be constructed in high throughput platforms. The presented approach will facilitate these strain construction programs, since assembly efficiency and robustness are bound to be key variables for high throughput strain construction, as they determine how many clones will be correctly assembled. The SHR-sequences can be designed and tested for these requirements and thus contribute to the development of these systems. Moreover, the use of SHR-sequences offers an unprecedented versatility. It is a goal of synthetic biology to create versatile platforms with libraries of interchangeable parts and pieces, as exemplified by the BioBricks concept
. Using the SHR sequences libraries of standardized parts, ranging from individual gene expression cassettes to fragments that carry entire pathways, can be generated and used for combinatorial assembly and subsequent screening for high-performing strains. In contrast to existing in vivo assembly approaches, no extensive re-designing has to be performed. Libraries of survival elements, genes and SHR-sequences will enable easy in silico design, straightforward in vitro synthesis of the fragments by PCR and efficient in vivo assembly.
While implementing in vivo assembly in our research, we have been surprised by its simplicity, ease of implementation and high efficiency. Within two years, in vivo assembly has almost completely replaced standard restriction/ligation protocols for construction of plasmids carrying two or more genes in our laboratory, thereby greatly accelerating strain construction and opening possibilities for strain modification that previously would have been deemed too complicated. Looking into the future, S. cerevisiae has the potential to be developed into an even more powerful platform. Similarly to popular E. coli strains, which have been extensively optimized to become extremely efficient hosts for plasmid transformation and replication, specific modifications of the yeast chassis, such as removal of the NHEJ machinery or enhancement of DNA uptake, could further extend the efficiency and fidelity of the in vivo assembly method.