Team:Peking R/Project

Softcoding of Genetic Program

Background of softcoding

There are basically two design principles in computer programming: hardcoding and softcoding. Hardcoding [1] refers to the practice of embedding parameters and functions into the source code of a program, whereas soft coding [2] obtains values and functions from external source. Hardcoding would be convenient when no dynamic parameters are required in the program, but the source code should be rewritten anytime the input data or functions change. On the contrary, softcoding enables users to customize the function of software to their needs by altering external input, without having to edit the program’s source code time after time.

In the exciting field of synthetic biology, various genetic programs have been developed to perform certain functions in living organisms. For instance, a genetic toggle switch in vivo was developed that could be switched between two states by chemical or thermal induction [3]. Another example is that an oscillatory network was constructed in which the synthesis of green fluorescent protein was periodically induced [4]. Yet genetic programs need optimization to achieve ideal performance, especially when several genetic modules are coupled. Besides, optimization of metabolic pathways is also a key issue in the field of metabolic engineering.

Traditional methods to optimize genetic programs or metabolic pathways generally involve construction of libraries that contain large amounts of mutants, and multi-round screening or iterative design is usually required. Apart from the obvious drawback that the constructing and screening procedures are laborious and time-consuming, these methods could only generate mutants with a fixed configuration, and to fine-tune their performance would require another round of mutagenesis and selection, which resembles “hardcoding”. Therefore, a platform for “softcoding” of genetic programs is urgently needed.

Softcoding Approach

Based on the principle of softcoding, this year our team established an extensible and versatile platform for softcoding of genetic program, which is composed of “genetic rheostats” and an “RBS calculator”: genetic rheostats consist of interoperable and truly modular ligand-responsive riboswitches/ribozymes, while the RBS calculator is automated design of synthetic ribosomal binding sites (RBS) with customized relative translation strength. When combining them together, a quantitative correlation between the concentration of specific ligand and synthetic RBS’ relative translation strength can be established. Therefore, when tuning genetic program, customized RBS’ relative translation strength at multiple sites can be high-throughputly achieved without having to conduct laborious mutagenesis and characterization, followed by easily determining the configuration of RBS(s)’ relative translation strength. Then RBS sequences that meet this configuration will be automatically designed via computer algorithms.

Genetic Rheostats (Learn more...)

Recently, RNA devices have emerged as powerful tools to regulate gene expression in vivo, and particularly, ligand-responsive riboswitches enable us to manipulate translation strength of specific genes upon different concentrations of ligands [5]. Ligand-responsive riboswitches regulate the translation rate of downstream gene by changing conformations, cleaving or splicing upon external addition of ligand. Compared with transcriptional and post-translational regulation, riboswitches function through allostery of RNA structure, which requires little or no assistance from proteins, so the regulation mechanism is relatively simpler and their functions are more decoupled from native biological activities. In our project, we selected riboswitches as the genetic rheostats.

We characterized some existing riboswitches, namely thiamine pyrophosphate (TPP)-responsive hammerhead ribozymes and theophylline riboswitches. By altering the upstream promoter and downstream coding sequence of the RNA controllers, we demonstrated that their performance was independent from sequence context, which proved that these RNA devices are modular.

To further extend the range of application of our genetic rheostats, we created a ribozyme that functions with a different mechanism, which has an extreme low level of backgrounds. We substituted the aptamer domain of c-di-GMP group Intron to theophylline-responsive aptamer, thus invented a group I intron that senses theophylline to perform splicing function. Moreover, we introduced a general method to evolve hammerhead ribozyme that senses a new ligand. By coupling an adenine aptamer with hammerhead ribozyme and randomizing nucleotides in the linker domain, we evolved new hammerhead ribozymes through dual selection, whose self-cleavage could be regulated by adenine. Our project provided a new design principle for rational or semi-rational design of genetic rheostats.

Figure 1． A) The tertiary structure of natural TPP aptamer. B) Measure the translation ratio curves of P1G1 and 1G1 in the absence of arabinose with a theophylline gradient. C) The design of the engineered group I intron with theophylline ribozyme. D) Schematic illustration of the dual genetic selection process of gene switches

RBS Calculator (Learn more...)

In bacteria, ribosomal binding site (RBS) sequence is one of the most important determinants of translational initiation/translation strength. Therefore manipulating RBS sequence would significantly affect the relative translation strength of downstream gene. Salis et.al. [6] used Gibbs energy (∆G) of the “docked” state of the mRNA-30S ribosomal subunit complex to predict the translation strength of RBS sequence. Based on their pioneering work, we developed an RBS calculator that correlated the performance of the RNA controllers under certain concentration of ligand to relative translation strength met by corresponding RBS sequence. Combining this RBS calculator with our genetic rheostats, we can generate an RBS sequence through automated design once we achieved an ideal configuration of genetic programs through genetic rheostats.

figure 2 Schematic representation of genetic soft-coding approach and the role of RBS calculator.

Demonstration of softcoding platform (learn more...)

We utilized the platform to fine-tune performance of two genetic circuits to acquire AND gate performance and bistable switch behavior, respectively.

AND gate

	Figure 3. AND gate performance regulated by different concentration of thiamine pyrophosphate (TPP). The on/off ratio of AND gate increases with ligand concentration, while the single induction of arabinose is diminished, resulting in a well performed AND gate.

Bistable switch

	Figure 4. Fluorescence images of E.coli DH5α strain populations with different plasmids from bistable switch mutant library. Each plasmid contains different ribosome binding sites (RBSs) which control the expression of cI434 gene, demonstrating that the ratiometric of green cells to red cells is correlated with translation strength.

violacein biosynthetic pathway

We further applied this platform to optimize a segment of violacein biosynthetic pathway, and achieved producing purer desired products.

	Figure 5. E. coli producing pigments. When induced by arabinose, the engineered E. coli produced dark-green pigments. Upon addition of different concentration of thiamine pyrophosphate (TPP), the color of the bacteria gradually shifted from dark-green to dark-brown.

Reference:

[1] http://en.wikipedia.org/wiki/Hard_coding
[2] http://en.wikipedia.org/wiki/Softcoding

[3].Gardner, Timothy S. et. al. (2000). Construction of a genetic toggleswitch in Escherichia coli. Nature 403, 339-342 [4].Elowitz, Michael B. and Leibler, Stanislas (2000). A synthetic oscillatory network of transcriptional regulators. Nature 403, 335-338 [5].Breaker, Ronald R (2004).Natural and engineered nucleicacids as tools to explore biology. Nature 432, 838-845
[6].Salis, Howard M et.al. (2009). Automated design of synthetic ribosome binding sitesto control protein expression. Nat. Biotech. 27, 946-950