When it comes to synthetic biology, it is important to ensure that the results of our project align with our expectations. However, before we successfully achieve our desired goals, there is a long way to go due to the unpredictability of the results. For this situation, we introduce an important principle: the engineering cycle, often abbreviated as the DBTL (Design, Build, Test, and Learn) cycle, to divide the entire experiment into four sections. The DBTL cycle helps us pinpoint where potential issues may arise more easily. When it goes wrong, we can redesign and rebuild a cycle to lead our biological system in the right direction and finally reach its destination. Beyond the four stages in the DBTL cycle, we include a 'recap' section after the learning stage to summarize actions for enhancing our experiment.

Figure 1. Engineering cycle.

    Cycle 1

Design:

In the attempt to circularizing linear RNA, we designed a DNA oligo, known as splint. The splint has a complementary sequence to the linear RNA. With this characteristic, it enhances the circularization ability of our RNA, by forming a double-stranded RNA (Figure 2) to circularize RNA. After the ligation process is done, the splint will be degraded by treating with DNase I. (See page design for the design of circularization). [1]

Figure 2. A diagram shows how splint helps the formation of circular RNA.

Build:

In the very beginning, we only noticed the basic principle of splinting. Such a splint needs to be a complementary sequence to linear RNA that acts as a platform that can ligate the ends of RNA, helping it to form the covalently-closed loop structure. Initially, we thought that the longer a complementary splint, the better its ligation ability, so we designed splint sequences ranging from 12 to 30 base pairs long. However, there would be too many candidates (10 exactly) if we designed and produced all the splints, which means a higher cost would be required, and this was not possible. Thus, we took a previous study [2] into account. The paper mentions a method for circularizing DNA from its linear form. However, this is the closest idea aligning with our project, so we had no option but to take their previous experience and parameters into consideration. In this study, it said that the best circularization efficiency is at 12 bp, nevertheless, the Tm value of our splint with 12bp was too low for the reaction. As a result, we mimicked their experiment and decided to design a 14 bp splint, Splint_1802_14 (BBa_K4636015) and Splint_4771_14 (BBa_K4636018) for circularizing RNA. (See page parts for detail information)

Test:

Here only shows the result of circularization using Splint_1802_14. To make sure the products of each step in synthesizing circRNA (See page design for details) were produced as expected, we decided to sequentially verify the effectiveness of each step and conduct E-Gel EX gel electrophoresis tests. Firstly, the presence of a band at lane 1 indicated a successful in vitro transcription (IVT) process, as we treated the reagent with DNase I. Secondly, we added XRN-1, a chemical capable of degrading monophosphorylated RNA [3] (lane 2: Without XRN-1, lane 3: With XRN-1 added). Lastly, given the circular form of RNA's resistance to RNA exonuclease activity, we could verify the yields in the back-splicing experiment by treating them with RNase R [4]. Similarly, we also treated the back-splicing product with XRN-1 in an attempt to degrade monophosphorylated linear RNA for further validation. (Lane 4: Without RNase R; lane 5: With RNase R added; lane 6: With XRN-1 added).

Figure 3. Lane 1: IVT product, Lane 2: Monophosphorylation product, Lane 3: Monophosphorylation product + XRN-1, Lane 4: Circularization product, Lane 5: Circularization product + RNase R, Lane 6: Circularization product + XRN-1

Learn:

From the result of E-Gel EX electrophoresis, lane 1, 2, and 3 displayed the outcomes we anticipated. In other words, in vitro transcription and monophosphorylation were confirmed to be successful. However, lanes 4, 5, and 6 exhibited such shallow bands that they were almost imperceptible. This indicated a low yield in our back splicing procedure.

Recap:

Knowing that the failure came from circularization, we then tried to search for more resources to explore any detail we might have ignored before.

    Cycle 2

Design:

After looking through papers and experiments done by previous people, we realized that the ligation efficacy of circular RNA depends on the CG content, the structure, the Tm value, and the entropy value. [5] Besides, we also noticed that the Tm value is a cofactor that affects the ligation ability. Splint's Tm value should be more than 25°C, so that it would be able to combine with the DNA more firmly. Last but not least, we thought of testing our designed splint sequence to predict entropy value and structure by running through a website [6]. We consider that the reason why our previous circularization experiment with 14 base pair splint didn't succeed in circularizing RNA lies in two main reasons: Tm value and GC content.

For the 14bp splint (For linear RNA of Insert_0101802, BBa_K4636040), the Tm value was lower than the 16bp splint, which means it may not correctly bind to our target RNA since the Tm of the splint should be higher than 25 °C to make it anneal with the DNA more firmly.

While for GC content, 14bp splints are 50% and 16bp splint are 43.8%. Since the higher GC%, the more difficult it will be to unscrew the double helix, the 14bp splint may have relatively poor ability compared with the 16bp splint.

The reason mentioned above and the previous 14bp experiment failure made us want to try a longer splint length, so we designed two more splint lengths, which were 16 and 18 base pairs long.

Build:

We finally build our splint sequence through filter criteria mentioned above, like CG content, avoiding secondary structure, splint Tm, and entropy values. This also included conducting the previous paper, so we designed two more different lengths, which were 16 (Splint_1802_16 ,BBa_K4636014) and 18 (Splint_1802_18, BBa_K4636013) base pairs long.

Test:

To prove the success of our new design, circularization product (With RNase R added) using Splint_1802_14 (BBa_K4636015) , Splint_1802_16 (BBa_K4636014) and Splint_1802_18 (BBa_K4636013) were tested together using agarose gel electrophoresis.

Figure 4. Lane 1: low range ssRNA ladder. Lane 2: Negative control. Lane 3: Back-splicing product using Splint_1802_14. Lane 4: Back-splicing product Splint_1802_16. Lane 5: Back-splicing product Splint_1802_18.

Learn:

Based on the result of agarose gel electrophoresis (Figure 4), we could identify that the splint with 16bp and 18bp have higher back-splicing efficiency than 14bp as we expected.

    Cycle 1

Design:

Since both the circular RNA we selected present low concentration in the human serum, we choose to amplify them through an isothermal amplification method called rolling circle amplification (RCA) (see page design for details).

Figure 5. A diagram shows how RCA amplify circRNA by producing long repeated single-stranded complementary DNA.

Build:

Followed by paper studying, the Protoscript II reverse transcriptase (New England BioLabs) seems to be the the most commonly used enzyme when conducting RCA using circRNA as template [7]. Therefore, we decided to use Protoscript II reverse transcriptase to amplify our target biomarkers and conducted agarose gel electrophoresis to present the result.

Test:

In theory, reverse transcriptase can repetitively amplify the circRNA to produce a long-repeated single-stranded DNA (ssDNA) (Figure 5). Conversely, the linear RNA can only be reverse transcribed into complementary DNA (cDNA) with a length identical to that of the RNA template. Consequently, we conducted RCA using both linear and circular RNA to evaluate the effectiveness of RCA.

Figure 6. Lane1: ssRNA ladder. Lane 2: RCA product using circularized RNA as template. Lane 3: RCA product using linear RNA as template.

Learn:

From the result of agarose gel electrophoresis (Figure 6), though the single-stranded cDNA was successfully synthesized, the length remained the same as the linear form of RNA even though the reaction time was extended. In this case, we assumed that there might be some inherent limitations to protoscript II reverse transcriptase.

Recap:

To deal with the question we hypothesized, all of us spared no effort to collect and integrate related literatures.

    Cycle 2

Design:

After extensive data querying, we found that the amplification ability of Protoscript II reverse transcriptase is limited to around 500 base pairs. On the other hand, we also discovered a promising reverse transcriptase called Induro® Reverse Transcriptase (New England BioLabs)with an amplification ability of around 12,000 base pairs [8].

Build:

We followed the protocol provided by the manufacturer. Since the manufacturer suggested a reaction time between 15-30 minutes, we conducted the reaction with durations of 15 and 30 minutes respectively to determine the optimal condition.

Test:

We used a ssRNA ladder (New England BioLabs) (lane 1) to recognize the length of our reaction product. To determine the optimal reaction time, we loaded RCA products using circular RNA (circRNA) as a template with reaction times of approximately 15 minutes (lane 2) and 30 minutes (lane 3), respectively. RCA products using linear RNA with the RNA concentration remaining the same as in lane 2 and 3 were used as templates. Both reaction times of 15 minutes (lane 4) and 30 minutes (lane 5) were tested. For lane 6 and lane 7, the components were essentially the same as in lane 4 and lane 5, with the only difference being that the RNA concentration was 10 times higher. (Figure 7)

Figure 7. Lane 1: ssRNA ladder. Lane 2: CircRNA as template with reaction time of 15mins. Lane3: CircRNA as template with reaction time of 30 mins. Lane 4: Linear RNA with the concentration identical to lane 2 and lane 3 as template and the reaction time of 15mins. Lane 5: Linear RNA with the concentration identical to lane 2 and lane 3 as template and the reaction time of 30 mins. Lane 6: Linear form RNA with concentration 10 times higher and the reaction time of 15mins. Lane 7: Linear form RNA with concentration 10 times higher and the reaction time of 30 mins.

Learn:

According to the results (Figure 7), we observed that lane 2 and lane 3 displayed significantly larger RCA products (approximately 1000 to 2000 nt) compared to lane 4 to lane 7 (approximately 450 nt). This indicated that the Induro reverse transcriptase successfully amplified the RNA signal when circRNA was used as the template.

    Cycle 1

Design:

For the colorimetric assay, we need to attach DNA probes with sequence complementary to our RCA products on the surface of gold nanoparticles (see page description for details of gold nanoparticle based-colorimetric assay) [9].

Figure 8. A diagram shows the mechanism of gold nanoparticle based-colorimetric assay.

According to the literature, there will be a redshift in wavelength when probes are attached [10]. Therefore, in this section, our aim is to explore suitable experimental method to synthesize probe-conjugated gold nanoparticles with red shift in wavelength.

Build:

We referred to the protocol provided in the literature and labeled it as Option 1 (see page experiment for protocols) [9].. Subsequently, we determined the redshift in wavelength using the nanodrop technique.

Test:

Figure 9. The comparison of wavelengths between the pure gold nanoparticle solution and solution after processing Option 1.

Learn:

Based on Figure 9, we can observe that there was no red shift in wavelength, which indicated we failed to attach probes on the surface of gold nanoparticles.

Recap:

We hypothesized the failure of our attempt to synthesis probe-conjugated gold nanoparticles was a result of unsuitable experimental conditions. Consequently, we tried to find an alternative method to produce the complex.

    Cycle 2

Design:

Learned from the cycle 1, our adviser suggested us to study the references [11] provided by the previous paper to figure out if there were other solution to synthesize the probe-conjugated gold nanoparticles we desired.

Build:

Fortunately, we had found out another protocol, named it Option 2 (see page experiment for protocols) [11], for conjugating probes and gold nanoparticles. As mentioned in the cycle one, we could determined the red shift in wavelength using the nanodrop technique after completing the synthesis procedure.

Test:

Figure 10. The comparison of wavelengths between the pure gold nanoparticle solution and solution after processing Option 2.

Fortunately, we had found out another protocol, named it Option 2 (see page experiment for protocols) [11], for conjugating probes and gold nanoparticles. As mentioned in the cycle one, we could determined the red shift in wavelength using the nanodrop technique after completing the synthesis procedure.

Learn:

According to Figure 10, they show significant red shift in wavelength. In the other word, we successfully attached probes on the surface of gold nanoparticles.

    Cycle : Difussion

Purpose:

According to scientific research, we know that the segregation of the gold nanoparticles may influence the observed color of the solution. Therefore, in order to predict the color of the solution, we need to investigate the diffusion process of the gold nanoparticles beforehand.

Design:

We use the function and the geometric factors of the container to simulate the diffusion of the particles. After getting the equilibrium condition of the diffusion result, we then try to comprehend the distribution of every particle to further derive the wavelength shift of the liquid.

Redesign:

Initially, we assume the velocity inlet to be constant, the result is show in Figure 11:

Figure 11. Diffusion under constant velocity

We then use the Gaussian distribution to fit with Figure 11 to see the distribution extent. At first, we use the variances of 1, 5 and 20 then choose the most similar case to fit with Figure11 The results are given as follows:

Build:

However, the constant descending velocity does not fit the actual condition since the velocity may decrease along the descending process. Then we change descending velocity into a time-depending function to more explictly simulate the real condition. The diffusion result can be shown as:

Test:

Figure 12. Gaussian distribution of variance 1, 5 and 20

However, even with the most familiar in those three cases, there is still a significant difference between the simulation and the result. In order to retrieve the most familiar variance, we use the following code to compare the pixel between distribution under different variances and Figure 11 Hence, the variance can be elaborated as follows:

Figure 13. Difference of pictures under different variances

Worth mentioning, when the variance is greater than 1 then the segregation area will be much smaller than Figure 1. Therefore, we choose the lowest point in the graph to do the distance calculation. The comparison between the most familiar variance and Figure 1 are shown as follows:

Figure 14. Comparison between the most familiar variance and Figure 11

Moreover, we also retrieve the value directly from the Simscale:

Figure 15. Turbulent kinetic energy of every particle

In order to calculate the distance between particles, we assume that the particle in the container is homogenous and distribute uniformly, therefore the kinetic energy is proportional to the number of particles. The result is given as:

Volume (m3)	Moles (mol)	Max distance (m)	Min distance (m)
7.5×10-8	9.75×10-13	5.04×10-7	5.03×10-7

Sheet 1. Calculation of distance between particles in every region

Build:

However, the constant descending velocity does not fit the actual condition since the velocity may decrease along the descending process. Then we change descending velocity into a time-depending function to more explictly simulate the real condition. The diffusion result can be shown as:

Figure 16. Diffusion under time-depending velocity

Learn:

After determinating that diffusion are mainly cause by the velocity inlet and the osmotic pressure which caused by the difference of the concentration, we know can further derive the relationship as follows:

Figure 17. Derivation of the velocity relationship

Redesign:

After knowing the velocity relationship, we apply it back to the simulation to understand the influence of different concentration. The simulation results are given in the model page.

    Cycle : Amplification

Purpose:

The establishment of RT-RPA model is to assist Wetlab in determining whether the current experimental design can yield the expected results through simulation before the actual experiment.

Design:

We first select the crucial reactions of RT-RPA and converted these reactions into differential equations. Then, we use SimBiology to execute these differential equations.

Build:

At the beginning, we intended to directly utilize the differential equations from the paper and incorporate them with the differential equations for reverse transcriptase. However, after running simulations, we found that we were unable to obtain results consistent with the paper.

Figure 18. Simulation Results of cDNA after a 20 minutes RPA Reaction

Learn:

After referring to the modeling approach of previous iGEM teams, we decided to independently select the crucial reactions of RT-RPA and input them into SimBiology for simulating these reactions.

Redesign:

After determining the new approach to constructing the model, we referred to the relevant paper and selected the crucial reactions of RT-RPA. We then inputted these reactions into SimBiology to establish the model.

G + F ⇌ FG (1)

FG + R ⇌ FGR' (2a)

FGR' → FGRi + G (2b)

FGRi + R ⇌ FGRc (2c)

FGRc → FRc + G (2d)

FRc + D ⇌ FRcD (3)

FRcD + ATP → FD + R + AMP + PPi (4a)

FRcD + ATP → FD + ADP + H3PO4 (4b)

FD + P ⇌ PFD (5)

PFD + dNTPs → P + PPi + D (6)

This is what the complete RT-RPA model look like in simbiology:

Figure 19. RT-RPA reaction pathway in MATLAB Simbiology

After identifying those crucial reactions of RT-RPA, we referred to relevant paper to obtain the required parameters and initial values. In addition, we also obtained initial values for some key reactants from Wetlab. Ultimately, we successfully get the simulation results we expected.

Figure 20. Simulation Results of cDNA after a 20 minutes RPA Reaction

After obtaining the simulated results we expected, we aim to apply the model in practical scenarios. We decided to use this RT-RPA model to validate Wetlab's proposition that the quantity of the forward primer influences the lower limit of the initial values of cDNA. The simulation results showed that the primer quantity does influence the lower limit of the initial value of cDNA. This not only confirmed our proposition but also boosted our confidence in the RT-RPA model.

Figure 21. simulation results of cDNA after a 40 minutes RPA reaction when the forward primer quantity is 4.8e-7

Figure 22. simulation results of cDNA after a 40 minutes RPA reaction when the forward primer quantity is 4.8e-8

Reference

[1] Chen, H., Cheng, K., Liu, X., An, R., Komiyama, M., & Liang, X. (2020). Preferential production of RNA rings by T4 RNA ligase 2 without any splint through rational design of precursor strand. Nucleic acids research, 48(9), e54. https://doi.org/10.1093/nar/gkaa181

[2] An, R., Li, Q., Fan, Y., Li, J., Pan, X., Komiyama, M., & Liang, X. (2017). Highly efficient preparation of single-stranded DNA rings by T4 ligase at abnormally low Mg(II) concentration. Nucleic acids research, 45(15), e139. https://doi.org/10.1093/nar/gkx553

[3] Jinek, M., Coyle, S. M., & Doudna, J. A. (2011). Coupled 5' nucleotide recognition and processivity in Xrn1-mediated mRNA decay. Molecular cell, 41(5), 600–608. https://doi.org/10.1016/j.molcel.2011.02.004

[4] Boss, M., & Arenz, C. (2020). A Fast and Easy Method for Specific Detection of Circular RNA by Rolling-Circle Amplification. Chembiochem : a European journal of chemical biology, 21(6), 793–796. https://doi.org/10.1002/cbic.201900514

[5] Cui, Y., Han, X., An, R., Zhou, G., Komiyama, M., & Liang, X. (2018). Cyclization of secondarily structured oligonucleotides to single-stranded rings by using Taq DNA ligase at high temperatures. RSC advances, 8(34), 18972–18979. https://doi.org/10.1039/c8ra02804d

[6] http://rna.tbi.univie.ac.at//cgi-bin/RNAWebSuite/RNAfold.cgi?PAGE=3&ID=zcRAvdofnB&r=65

[7] Boss, M., & Arenz, C. (2020). A Fast and Easy Method for Specific Detection of Circular RNA by Rolling-Circle Amplification. Chembiochem : a European journal of chemical biology, 21(6), 793–796. https://doi.org/10.1002/cbic.201900514

[8] Maguire, S., & Guan, S. (2022). Rolling circle reverse transcription enables high fidelity nanopore sequencing of small RNA. PloS one, 17(10), e0275471. https://doi.org/10.1371/journal.pone.0275471

[9] Ali, M. M., Kanda, P., Aguirre, S. D., & Li, Y. (2011). Modulation of DNA-modified gold-nanoparticle stability in salt with concatemeric single-stranded DNAs for colorimetric bioassay development. Chemistry (Weinheim an der Bergstrasse, Germany), 17(7), 2052-2056. https://doi.org/10.1002/chem.201002677

[10] Li, F., Zhang, H., Dever, B., Li, X. F., & Le, X. C. (2013). Thermal stability of DNA functionalized gold nanoparticles. Bioconjugate chemistry, 24(11), 1790-1797. https://doi.org/10.1021/bc300687z

[11] Liu, J., & Lu, Y. (2006). Preparation of aptamer-linked gold nanoparticle purple aggregates for colorimetric sensing of analytes. Nature protocols, 1(1), 246-252. https://doi.org/10.1038/nprot.2006.38

[12] R. Tsekov, P. Georgiev, S. Simeonova and K. Balashev (2017), Quantifying the Blue Shift in the Light Absorption of Small Gold Nanoparticles https://arxiv.org/ftp/arxiv/papers/1702/1702.04513.pdf

[13] Perez-Juste J, Liz- Marzán LM, Carnie S, Chan DYC, Mulvaney P (2004) Electric-field-directed growth of gold nanorods in aqueous surfactant solutions https://doi.org/10.1002/adfm.200305068

[14] Morris AM, Watzky MA, Finke RG (2009) Protein aggregation kinetics, mechanism, and curve-fitting: a review of the literature 10.1016/j.bbapap.2008.10.016

[15] Lewis GN (1908). The Osmotic Pressure of Concentrated Solutions and the Laws of the Perfect Solution. Journal of the American Chemical Society. https://pubs.acs.org/doi/abs/10.1021/ja01947a002

[16] Liu, Jie ; Berger, Christopher L ; Morrical, Scott W (2013). Kinetics of Presynaptic Filament Assembly in the Presence of Single-Stranded DNA Binding Protein and Recombination Mediator Protein. Biochemistry (Easton), 2013, Vol.52 (45), p.7878-7889 https://pubs.acs.org/doi/full/10.1021/bi401060p

[17] Moody, Clint ; Newell, Heather ; Viljoen, Hendrik (2016). A mathematical model of recombinase polymerase amplification under continuously stirred conditions. Biochemical engineering journal, 2016, Vol.112, p.193-201 https://www.sciencedirect.com/science/article/pii/S1369703X16301152?casa_token=nv3pO9UHxvYAAAAA:9kA9Wc27RzRSbEKsF5uC_q44f_6Z-g4MGT-T63K7gns8uvCOv2T9ttnD-UmuxCmZKDGIWak_roM#bib0090

[18] Farb, Joshua N. ; Morrical, Scott W. (2009). Role of Allosteric Switch Residue Histidine 195 in Maintaining Active-Site Asymmetry in Presynaptic Filaments of Bacteriophage T4 UvsX Recombinase. Journal of molecular biology, 2009, Vol.385 (2), p.393-404 https://www.sciencedirect.com/science/article/pii/S0022283608014125