Discussion

Combinatorial design and construction using multi-part DNA assembly is a very useful approach to quickly and easily obtain effective designs of DNA constructs encoding metabolic pathways or other systems. However, in standard designs, hierarchical DNA assembly systems generate intermediate constructs with complete expression units, leading to expression of CDSs and metabolic burden in the typical assembly host E. coli, reducing library size, quality, and exploration of design space. Here we tested Clostridium synthetic promoters in E. coli, and found they were too strong and thus burdensome in E. coli to allow standard combinatorial assembly, in keeping with previous observations. This issue may be a common barrier to combinatorial assembly for Clostridium and other organisms. To address this problem, here we developed integration-coupled activation of promoterless sequences (ICAPS), assembling constructs or libraries as promoterless operons which are inactive during assembly, and activating their expression only upon ACE chromosomal integration downstream of a promoter.

As an ICAPS proof of concept, we constructed a combinatorial library for an extended CoA-dependent alcohol pathway in C. acetobutylicum and activated it via ACE integration, achieving detectable hexanol production in a strain not previously engineered for this product. Functional variants were obtained at a good frequency (15% of colonies screened) from a relatively small library of 216 possible designs, indicating the usefulness of the approach. Our results also identified bktb as a potential bottleneck, showing how combinatorial approaches can readily reveal expression trends and targets for improvement.

The titres we observed were modest compared to native producers such as C. carboxidivorans [31, 33] or engineered C. ljungdahlii [34]. This suggests that improvements could be made, and that the range of expression levels used did not maximize production of hexanol. The system could be improved by developing further RBSs, and by testing further chromosomal integration sites with promoters providing different levels and patterns of expression. RBS strength is known to be context-dependent, especially to depend upon CDS sequence. Other strategies which aim to make translation less context-dependent, like bicistronic design (BCD), might be usefully combined with ICAPS [35]. The choice of C. necator bktb, its codon optimization, competition between branches of the pathway, and the strength of flux towards butanol might contribute towards the low hexanol titres.

ICAPS, and ACE integration which ICAPS builds upon, have similarities with classic promoter trap/gene trap methods, and the more recent Tn-Seq. Promoter trap/gene trap methods use transcriptional activation of a reporter or marker gene upon chromosomal integration of a recombinant transposon to identify active promoters [36]. Tn-Seq is a high-throughput parallel method enabling fitness and genetic interaction studies [37] and identification of essential genes [38]. Here we applied a similar principle but in a synthetic biology context, to facilitate design and construction, rather than analysis.

New vectors and parts constructed for this study include a special set of nine alternative Start-Stop Assembly Level 1 vectors for simple, efficient assembly of promoterless operons (omitting the standard promoter and terminator positions, thus with β-δ acceptor fusion sites); a set of context-resistant synthetic RBSs, and E. coli–C. acetobutylicum ICAPS shuttle vector pPM65. These vectors and parts may be useful to others either directly, or as an example of how the ICAPS approach could be implemented in other cases.