26 2 Trackable Multiplex Recombineering (TRMR) and Next-Generation Genome Design Technologies
2.4.1 TRMR and T^2 RMR are Currently Not Recursive
One challenge for the current TRMR and T^2 RMR designs is that only a single
round of recombineering is currently implemented. This limitation is due to
the fact that the TRMR and T^2 RMR substrates are dsDNA, and because of the
current low efficiency of dsDNA recombineering, it is essential that an antibi-
otic selection step is used to ensure the removal of non-recombineered cells.
While multiple different antibiotic markers may be used for successive rounds
of recombination with a TRMR or T^2 RMR library, the limited number of mark-
ers restricts the number of additional cycles that can be performed. Another
option is to remove the resistance gene (via flanking FRT sites) and reintro-
duce the same library in the next recombineering round, but this will greatly
extend the time required for every cycle. Alternately, a new technology called
CRISPR-enabled trackable genome engineering (CREATE) has been developed
that allows clustered regularly interspaced short palindromic repeat (CRISPR)-
based markerless selection of recombineered cells [34]; this technique could be
combined with the expression-level cassettes from TRMR or T^2 RMR as dis-
cussed in Section 2.5.2.
An additional concern is the rapid increase in the number of recombinants
that result from every TRMR or T^2 RMR cycle. In the ideal case, every possible
combination of mutations should be represented in the cell population, which
would require a volume of cells that exceeds the capability of current equip-
ment. Lastly, barcode identification is being performed at the whole population
level and thus does not distinguish whether the barcode originated from a
single or multiple cells. This limitation could be overcome by using new single-
cell sequencing technologies including (i) single-cell linkage PCR, which allows
for the sequencing of millions of barcoded individual cells [35] and (ii) tracking
combinatorial engineered libraries (TRACE), which gives the ability to track
combinations of mutations from a single cell [36, 37].2.4.2 Need for More Predictable Models
Mathematical modeling of a metabolic pathway can be a valuable tool for further
optimization of that pathway [reviewed in [38]]. Once the metabolic flux through
a pathway is accurately modeled, bottlenecks in that pathway can be identified,
and further engineering efforts can be directed toward removing that bottleneck.
TRMR and T^2 RMR can aid in the development of models by identifying genes
that are involved in a pathway and that would have been difficult to predict
a priori [22, 26]. Once these genes have been identified, new TRMR-like libraries
that are predicted to be enriched for better performing strains can be designed.
Although metabolic models can be useful, unfortunately they often lack pre-
dictive power [reviewed in [39]]. This can be due to a number of factors includ-
ing lack of mechanistic detail about the pathway, inconsistent behavior of
synDNA parts, or failure to account for epistatic interactions [25]. Epistatic
interactions can make both TRMR and T^2 RMR data particularly difficult to
model. The development of more predictive models is an active and ongoing part
of metabolic engineering research.