Team:CSMU Taiwan/Model

CSMU - iGem

Model

Toehold Switch Design

Initially, we designed our toehold switches based off of Wang et al.'s (2019)¹. They have designed a toehold switch for miRNA 21 which is also one of the miRNAs that we plan to detect. However, we were still having trouble with designing our own version of the toehold switch as the analysis with NUPACK showed unoptimal results. Therefore, we consulted with Alexander Green from Arizona State University and he advised us to design the toehold switch based off of one of his more recent studies on toehold switch for ZIKA virus (2016)². Therefore, we started to design our toehold switches and ended up with 14 toehold switches in total for our target miRNAs. To confirm our designs, we ran NUPACK analysis alongside Vienna RNA Package to determine the RNA-RNA interaction between the toehold switches and the triggers.

As we can see from the figure above, a toehold switch is composed of serveral different compartments:

Trigger Binding Site (TBS): This is the site where the desired target binds to. Generally, it is complementary to approximately 30 nucleotides from the target RNA³.

Ribosome Binding Site (RBS): This is where the ribosome binds and allows for the start of translating RNA into proteins. It is usually rich in AG for bacteria.

Linker: This linker acts as a spacial barrier for the toehold structure and the downstream reporter protein. This allows the protein domains to be separated and allowing the downstream protein to be uninterfered by the toehold structure.

Protein Coding Sequence: This is the region where we can insert the reporter protein. This allows us to detect the presence or quantify the amount of targets presence in the sample indirectly. This is usually a fluorescent protein such as Green Fluorescent Protein.

For an animated verison on the functionality of a toehold switch, please check out our promotion video on our home page.

For the purpose of designing our toehold switches, we broke the toehold switch into three sections.

Trigger Binding Site
The loop structure, which contains the upper stem, RBS, and the start codon.
Linker

Trigger Binding Site (TBS)

Due to the limitation of the length for miRNA only having 21-23 nucleotides, the trigger binding site (TBS) was solely designed upon reverse translating the sequence. For instance,

miRNA	Sequence	TBS
miR-21	UAGCUUAUCAGACUGAUGUUGA	UCAACAUCAGUCUGAUAAGCUA
miR-31	AGGCAAGAUGCUGGCAUAGCU	AGCUAUGCCAGCAUCUUGCCU
miR-146	UGAGAACUGAAUUCCAUGGGUU	AACCCAUGGAAUUCAGUUCUCA

To form the bottom stem of the toehold switch, we reverse complemented 11 nucleotides after the start condon (underlined in the table above) to bind with the TBS sequence.

The Loop Structure

For the loop structure, it should form a stable hairpin loop structure to avoid expression of the downstream reporter protein in the absence of the trigger RNA. We looked at one of Green’s papers² for our choice of design. In the paper, he designed two seperate loop structures and tested their effects on the ON/OFF ratio (a ratio is calculated by the expression of reporter protein in the presence and absence of the trigger) for the toehold switches. One of the designs, termed the A series, modified the original toehold switch³ in which the size of the loop domain was reduced from 18 nts to 11 nts. The reduce in length discourages loop-mediated docking of the ribosome which led to reduced leakage in the OFF state. The other design, termed the B series, uses a 12-nt loop and incorporates a more thermodynamically stable upper stem to prevent the leakage problem with the toehold switch. We have adopted the B series design from Green to generate our toehold switch and named it Zika loop. In addition, we adopted a loop design from the original toehold switch as our control group and called it Old loop. For miRNA 21, we kept the loop structure from Wang et al. (2019)¹ and named it Paper loop.

Name	Abbreviation	Source
Old loop	O	Toehold Switches: De-Novo-Designed Regulators of Gene Expression (Green et al., 2014)
Zika loop	Z	Rapid, Low-Cost Detection of Zika Virus Using Programmable Biomolecular Components(Green et al., 2016)
Paper loop	P	A Novel Synthetic Toehold Switch for microRNA Detection in Mammalian Cells (Wang et al., 2019)

Linker

The linker’s main purpose is to minimize the interference between the domains of toehold switch and the reporter protein when they form secondary structures. When designing the linker, we were looking for minimal bond formation between the linker and the TBS to prevent unwanted secondary structures. The linker design was an adaptation of the linker from Wang et al. (2019)¹ and through randomization of linkers and testing out the thermostability and secondary structure formations with NUPACK. Thus, we named the linker designed this way "Random linker". Besides, we also kept the linker from the toehold switch for ZIKA virus (2016)² and Wang et al. (2019)¹. Their names and abbreviations are shown in the table bellow.

Name	Abbreviation	Source
Random linker	R	IGEM_CSMU_2020
Zika linker	Z	Rapid, Low-Cost Detection of Zika Virus Using Programmable Biomolecular Components(Green et al., 2016)
Paper linker	P	A Novel Synthetic Toehold Switch for microRNA Detection in Mammalian Cells (Wang et al., 2019)

Our Toehold Structure Design

The assembly of different TBS, loops, linkers and their toehold code names are listed in the following tables.

For miR-31,

TBS	Loop	Linker	Name
31	Zika loop	Random linker	zr31
31	Zika loop	Zika linker	zz31
31	Old loop	Random linker	or31
31	Old loop	Zika linker	oz31

For miR-146,

TBS	Loop	Linker	Name
146	Zika loop	Random linker	zr146
146	Zika loop	Zika linker	zz146
146	Old loop	Random linker	or146
146	Old loop	Zika linker	oz146

For miR-21,

TBS	Loop	Linker	Name
21	Zika loop	Paper linker	zp21
21	Zika loop	Zika linker	zz21
21	Old loop	Paper linker	op21
21	Old loop	Zika linker	oz21
21	Paper loop	Paper linker	pp21
21	Paper loop	Zika linker	pz21

NUPACK Analysis

Take one of our design, zr31, as an example. We ran NUPACK to ensure the spontaneous formation of the secondary structure at 37 degrees Celsius, which is -22 kcal/mol. Another factor to consider while designing a toehold switch is to minimize the secondary structure formation from the upper stem all the way to the linker (red part shown in the sequence) of the design as this region is for the beginning of translation for ribosomes. Therefore, we need the RNA to remain single-stranded. In order to achieve that, the closer we can get the free energy to 0 while designing the sequence for the linker and loop structure, the more protein will be transcribed.

Vienna RNA Package

We used the RNAup webserver from the Vienna RNA Package to predict the interaction between our trigger and the toehold switch. The toehold switch-trigger RNA duplex needs to have a more favorable energy state compared to the unbound toehold switch and trigger RNA. If it does not, then the hairpin loop would not unfold. Therefore, we need to ensure the toehold switch has a more favorable energy state.

Take the simulation result of zr31 as an example again. The black line in the figure indicates the amount of energy required to open the secondary structures of the TBS. The line red indicates the amount of energy required to open the secondary structure after the binding of the trigger. As a result, this indicates that zr31 will successfully bind and open the locked structure spontaneously.

CLICK HERE for all results from NUPACK and Vienna RNA pack for all our sequences!

Invertase Activity

We started with a simple model of the Michaelis-Menten mechanism based on Keramat et al (2015) ⁴ to model the enzyme activity for the invertase we used in our experiment. The low substrate concentrations (Equation 1 and 2), which the Vm is calculated from the equation one above based on previous data from Keramat et al (2015) ⁴, and the substrate inhibition model for high concentrations of substrate (Equation 1, 3 and 4) were both used in the determination of the enzyme activity.

E = Enzyme concentration ; S = Sucrose concentration ; ES =Activate complex of enzyme–sucrose molecule ; P = Product ; V = Reaction rate ; Vm = Maximum reaction rate ; Km = Michaelis–Menten parameter ; ESS [-] Inactive complex of enzyme-two sucrose molecules ; Ki = Substrate inhibition parameter

As we can see from above the velocity rapidly accelerates as the concentration of the initial sucrose increases, but then it slowly reaches a maximum and eventually starts to decrease. The apparent decrease in reaction rate is caused by the folding of sucrose molecules from the intramolecular hydrogen bonds which is characterized by Mathlouthi et al (1998)⁵. From the paper, we were able to determine the inhibiting effect from the substrate reaching saturation using equation 5 and 6 which is known as the aggregation effect. The aggregation effect from the sucrose molecules can be considered using the Ks value calculated from equation 6.

Ks = Sucrose association parameter

After considering the Ks value from equation 6, we are able to combine it with equation 2 to form the following equation:

However, according to Keramat et al (2015)⁴, there are still several weaknesses in this model as it does not match their experimental data.⁴

For instease, the addition of an excess Arrhenius type term where it represents the probability of enzyme deactivation by the inhibition of products during hydrolysis. This enzyme deactivation parameter is shown by Kd from below,
EI* = Irreversible ; Kd = Irreversible enzyme deactivation parameter

However, this model still has a low accuracy for the initial time lag at the beginning of the reaction. This is considered for the enzyme sites of sucrose molecules to become available. ⁴ This was deduced by adding in a initial time lag facter (1-e^(-Ka * t)) which results in our final model:
Ka = Activation parameter of enzymes

Ka (= 0.1171 for the S3(III) model) and Kd (= 0.0058 and 0.0054 for S3(II) and S3(III) models, respectively) are two empirical parameters in Eqs. 11 and 12 which are calculated by the presently available data (Kermat et al. 2015).

This model is capable of predicting the reaction rate at the beginning and also at the remaining reaction pathway which includes the initial time lag and irreversible enzyme deactivation through the remaining pathway. More importantly, Keramat et al. (2015)⁴ have confirmed the similarity in this model and in their experimental data. Therefore, this can be a great model for kinetic investigations of the enzymatic reactions and act as good conformity when compared to our experimental data.

Figure 1. The predicted invertase reaction rate of the MM model using MATLAB.

Figure 2. The predicted invertase reaction rate of the S^2 model using MATLAB.

Figure 3. The predicted invertase reaction rate of the S^3 model using MATLAB.

Figure 4. A comparison between the predicted invertase reaction rate using the all three model.

Reference

Shue Wang, Nicholas J Emery, Allen P Liu. A Novel Synthetic Toehold Switch for microRNA Detection in Mammalian Cells. ACS Synthetic Biology 2019; 8 (5): 1079-1088.
Pardee K, Green AA, Takahashi MK, et al. Rapid, Low-Cost Detection of Zika Virus Using Programmable Biomolecular Components. Cell 2016; 165(5): 1255-66.
Green AA, Silver PA, Collins JJ, Yin P. Toehold switches: de-novo-designed regulators of gene expression. Cell 2014; 159(4): 925-39.
Kermat A et al. Kinetic Model for Invertase‐Induced Sucrose Hydrolysis: Initial Time Lag. Chemical Engineering Technology 2015; 40(3): 529-536.
Mathlouthia A, and Genotelleb J. Role of water in sucrose crystallization. Elsevier 1998; 37(3): 335-342

Tools Used

NUPACK: http://www.nupack.org/

Vienna RNA Pack: http://rna.tbi.univie.ac.at/