10.2 Synthetic Control of Transcript Stability 199[21, 78, 88]. Phosphodiester bond cleavage within the 5′ UTR also removes the
5 ′-P(PP) recognized by RppH or RNase E, which can lead to increased transcript
persistence and gene expression [7, 21, 88] (see Section 10.1.2). We speculate
that mRNA terminated with a 5′-OH is degraded via comparatively slow RNase
E direct entry, consistent with increased half-lives measured for transcripts
cleaved by hammerhead ribozymes [21].
Carothers et al. [7] formulated a model-driven process that uses UTR cleavage
to engineer devices that regulate transcript stability and quantitatively program
gene expression. Static ribozyme-regulated expression devices (rREDs) and
dynamic, metabolite-controlled aptazyme-regulated expression devices (aREDs)
were constructed that employ transcript stability, via 5′ UTR cleavage, as the
underlying genetic control mechanism. With mechanistic understanding of RNA
degradation pathways as a starting point [21, 88], a coarse-grained biochemical
model of device function was created to simulate global device functions from
local, measurable, and tunable component characteristics. The combinatorial
space of design variable inputs was then mapped to the space of device outputs
with a sampling-based approach, providing data for global sensitivity analysis
(GSA) and identifying functional designs that meet targeted performance crite-
ria. To physically implement functional devices, a novel method for designing
transcripts with kinetic RNA folding simulations [89] was created that enables
the assembly of individually characterized components parts.
To demonstrate that variations in tunable design parameters generate quanti-
tatively predictable outputs, genetic devices were constructed to program
amounts of a reporter protein and production levels of p-aminophenylalanine
(p-AF), a chemical precursor of bioactive compounds and advanced polymers,
from a 12-gene engineered biosynthetic pathway. In total, 28 E. coli expression
devices were assembled from component parts that were generated and charac-
terized separately in vitro, in vivo, and in silico. Excellent quantitative agree-
ment between the design specifications and the device functions (r = 0.94) was
observed, experimentally validating the underlying models and simulation tools
and the overall approach. rREDs and aREDs have immediate utility as program-
mable biosensors and controllers for metabolic pathways and genetic circuits.
And, notably, this work also provides a conceptual and experimental framework
for investigating and engineering complex RNA functions through the applica-
tion of fundamental biochemical understanding.
Using this framework, we envision a model-driven design process for creating
RNA-based dynamic control systems for applications in metabolic engineering
and biosensing (Figure 10.4). As a testbed for RNA-based control circuit design,
we are engineering E. coli to produce p-aminostyrene (p-AS), a component of
polymer composites with optical and mechanical properties favorable for
advanced applications in photonics, photolithography, and biomedicine [90, 91].
Substituted styrenes have been difficult to chemically synthesize in high yields
[92], and the cytotoxicity of key intermediates and products has prevented effi-
cient microbial production [93]. The proposed p-AS pathway is an ideal testbed
because it has 15 well-defined gene products and measurable intermediates yet
presents a full complement of canonical control problems that must be addressed
to obtain efficient production.