Project Description

---

Introduction

In our project we were inspired to enable a targeted protein translation of an mRNA molecule. That is due to the many advantages it offers while progressing towards unlocking the true potential of mRNA-based technologies. Technology that may revolutionize personalized medicine, bioproduction, vaccines and much more. Our approach is to generate a solution which is based upon the sequence of the mRNA, rather than its delivery mechanism. This approach enables easier designing and an increased modulation in comparison with the current methods. Therefore, we have produced a computational tool, which is directed on generating optimized toehold switches for any specific need. A toehold switch is a secondary mRNA structure in the 5' UTR of the mRNA, enabling/disabling translation based on the physical accessibility of the mRNA to approaching ribosomes, induced by the presence/lack of presence of a trigger RNA molecule that might interact with the switch and change its conformation. The ability to generate an adapted switch for any use might substantially enhance the ability of research facilities and industrial companies to induce an efficient specific translation. By that, the developing field of RNA-based therapeutics might be scaled up.

---

Choosing Our Project

As we have gathered together as a team, from the very first moment the theme of specificity has been brought up to the table: divided into pairs, we have introduced iGEM projects developed throughout past years competitions. Even though each pair has chosen its project independently of the others, it turned out that each selected project has dealt with a very specific challenge of the growing field of Synthetic Biology: while one has struggled to develop a rapid kit for the detection of pandemics, another one has been aspired to develop a reliable biological sensor in a generic manner; a third project has tried to develop a synthetic ATP recycling system, while a fourth one has engineered an intercellular approach to deal with CKD (Chronic Kidney Disease). We then realized that if we desire to form a successful and beneficial iGEM project, we have to dive into a biological problem and adapt a specific solution to it using the generic tools of Synthetic Biology.

The next stage was to come up with ideas by ourselves for a project. Still divided into pairs, the projects which have been offered were of various sources: from a kit for a rapid diagnosis of tuberculosis, throughout preventing an immune response against the Gluten protein by modification of its sequence (as a therapy for the Celiac disease), up to detecting water contamination utilizing bacterial communication systems. However, as discussion progressed, it has been realized that not only we have to be focused on a specific biological problem we'd like to solve, but it could be of importance also to think of specificity itself as the very solution to that problem. Therefore, the idea of Nitay and Shai, whose theme was to enable differential expression of an mRNA in a specific tissue (using a secondary structure formed in the 5'-UTR of the molecule), has been chosen as the one upon which our project will be based on.

---

The Problem Our Project Deals With

Our project deals with the problem of specific translation of an mRNA molecule in a target tissue. Most of the nowadays approaches to it focus on the delivery itself of the molecule to Eukaryotic tissues, not on restricting the expression of it to a specific one. For example, the current mRNA vaccines developed against SARS-CoV-2 have utilized the delivery technology of liposomes: cationic lipid-based particles which encapsulate the anionic mRNA, protecting it from degradation. While engaging the target tissue, the mRNA is injected to it and the lipid-based particle is dissolved (Vahed et al., 2017). However, in case one would like to derive a certain protein to be expressed in a certain tissue only, it cannot be obtained while using that approach solely: once injected, liposomes will reach systemically the organism's tissues and the mRNA will be found at each. Our solution does not intend to replace the liposome-based approaches (or any other delivery technology), which has many advantages and benefits. Instead, we desire to enable an mRNA molecule, delivered one way or another, to be expressed just in a target tissue and not in another. We plan to achieve that end by adapting a solution which is based upon the sequence of the mRNA prone to be translated into a protein and the secondary structure which is expected to be formed as a result.

---

What Inspired Us to Choose This Project

RNA-based therapies hold the potential to revolutionize the ways by which future medical institutions and companies will handle treatment and prevention of diseases. As already stated, many cutting-edge technologies encompassing various fields, such as vaccines development, cancer therapy and personalized medicine, have utilized the advantages of RNA – safe, easily-manufactured, cost effective – in order to develop new therapies, successfully addressing issues that have been considered to be untreatable or hard to approach (Zhu et al., 2022).

However, precautions have to be taken even when dealing with an allegedly safe tool as RNA. In order to form a successful RNA-based therapy, it is required that in most cases, once injected, the translation of an mRNA into a functional protein will be restricted to a target tissue or to a specific cell type.

There are many medical applications desiring to achieve specificity, although they do not have the means for it. Many strategies have been taken to address this issue, most of which have put their faith on a switch module, usually formed by a secondary structure element in the 5' UTR of an mRNA molecule (an hairpin, for example) and designed to sense and respond to an endogenous RNA trigger that is mostly expressed in the target tissue/cell type. Typically, the module switches between on and off states, enabling or disabling translation, respectively: for example, the off state could be represented by a self-folding of the mRNA so a ribosome cannot reach it and induce translation, while the on state could be represented by an hybridization between the mRNA and the RNA trigger, so the first is exposed to approaching ribosomes(Green et al., 2014). However, developing an optimal switch is a complex process which depends upon many parameters, and is also a resource-consuming one (Green et al., 2014).

Even though the concept of a secondary mRNA structure forming a stem loop was well-known for decades (Singh et al., 2019), generating one in a synthetic manner was challenging due to the low number of biological components which may regulate its activity (Green et al., 2014). It is the pioneered work of Green et al (Cell, 2014) that has opened the field of de-novo designed riboregulators. Developed in Prokaryotes, the newly-developed construct, known as a Toehold Switch, has enabled the expression of desired genes based on the presence of an RNA molecule, the last acting in trans, in order to regulate translation by binding to the switch (which itself is an RNA strand, regulating a biological activity, such as translation) (Green et al., 2014).This work has demonstrated well how to overcome obstacles implemented within biological environments. Not only that the construct has yielded a very wide dynamic range* (the average one has exceeded 400), but it also helped to form a low crosstalk system. Up to this work, RNA-based regulators have not been close to reach their full potential serving as orthogonal regulatory components, mainly because most of the engineered riboregulators devices had been developed earlier were based on base pairing of the trans-acting RNA molecule to the ribosome binding site (RBS) itself.

By that, the trans-acting RNA was imposed to contain an RBS-binding sequence, constraining the range of RNA molecules that could have been of use to act in trans upon the switch and induce biological regulation. Toehold constructs, in contrast, put their faith upon sequestration of a region around the start codon, so a translation could be activated without restricting the trans-acting RNA molecule to contain a sequence interacting with the RBS or even the start codon itself. This space of the RNA sequence acting in trans (an mRNA molecule in that specific work of Green et al) has allowed the formation of libraries of components, constructed as such that unprecedented orthogonality could have been achieved – a set of twenty six systems which demonstrate less than 12% crosstalk (Green et al., 2014).

This canonical work in Prokaryotes has inspired many others to think of implementing the Toehold construct in Eukaryotes, so endless applicative developments might be formed. While the work of Green et al. has dealt with the inherent difficulties of a biological system, others have planned to expand that knowledge to the much more complex realms of therapeutics. One such work is the one of Wang et al (ACS Synthetic Biology, 2019), which attempted to utilize Toehold switches to detect presence of microRNA (miRNA) molecules in mammalian cells. Throughout this work, a synthetic Toehold switch, adapted for the Eukaryotic system, has been designed. Although, it is important to note that the translation induced by the construct was not the end of it, but rather a tool for detection of miRNAs, serving as the trans-acting RNA molecule (a trigger RNA molecule). The switch itself has been designed in a restrictive manner in order for it to hybridize with the required miRNA and induce translation of a reporter gene (Wang et al., 2019). Since so, the indices of that switch were quite different from the ones of the construct generated by Green et al: the dynamic range was much lower (about 2), and the orthogonality levels (as they rise up they represent a lower crosstalk between the system's components) were rather modest (Wang et al., 2019).

Researchers were not the only ones which have tried to utilize the advantages of the Toehold's construct, but also former iGEM teams took up the challenge. As a part of our former iGEM projects presentations, Ilan and Shir have chosen to introduce The House of Toeholds project, developed by the Sastra team throughout the 2019 iGEM competition. The goal of this project was to detect cervical cancer in a non-invasive manner and at an early stage in order to reduce mortality. Even though this project has not yielded a proof of concept, it has gained many achievements, including nomination for the Best Software Tool Prize.

Given all that, it could be summed up that the Toehold construct is a very advantageous riboregulator, which has the potential to generate powerful applications of synthetic biology scattered around different fields. However, even though its potency has been well demonstrated in Prokaryotes, its performances among Eukaryotes systems were much more modest. It is very important to note that the designing of the switch was specific for each case. While the work of Wang et al. has produced a construct that was suitable for the detection of the specific miRNA molecules, the iGEM 2019 Sastra team has designed a Toehold switch adjustable for the specific miRNA molecules expressed throughout the early stages of cervical cancer. In relation to our initial insight, we have come to realize that producing a generic tool for the specific need of designing Toehold constructs will be a perfect fit for our project. Even though the Toehold switch is only one out of many riboregulators, it could be inferred from the Green et al. work that it yields an enormous potential to serve as a classical synthetic biology application.

However, in order for it to reach its full potential as an easy-to-use tool, it must be optimized to be as generic as possible, so research groups around the world will be spared the need to design it by themselves for each specific application. In addition, it could be inferred that the applications of the Toehold switch generated so far in Eukaryotes were less convincing than the work in Prokaryotes of Green et al.: a much lower dynamic range (if any) while using only specific miRNAs as trigger RNA molecules. Those realizations have led us to set up our goals, in light of the need to get computational models into action so generic Toehold switches could be formed, and the vision of bringing the Toehold construct to serve as an integral part of therapeutic applications involving mammalian organisms.

---

How We Plan to Address the Problem

We have decided that the main goal of this project is to develop a user-friendly software that will ease the struggle of creating a Toehold switch module, enabling specific translation for any Eukaryotic application. Our tool should be based upon computational models so it would be as generic as possible in order to fit the most-optimized Toehold for a specific application. By generating such a tool, researchers and companies may utilize easily the advantages of Toeholds switches as riboregulators in order to generate selectively translated mRNAs or to provide solutions to any other applicative need. The tool will be generated by developing a sophisticated AI algorithm which finds the optimal RNA trigger in a given cell type/tissue and optimizes the switch module sequence accordingly. The algorithm will be backed by data gathered from the literature as well as data generated by us in the lab for testing and verification.

However, since we aspire to assist with the development of Eukaryotic applications, it is rather important that the output optimized Toehold construct will yield higher performances than the ones witnessed up to date, so we have set up another goal: achieving a better dynamic range concerning the translation ratios while dealing with Eukaryotic systems. Since the dynamic range was low in the cases where miRNAs were utilized as the trigger RNA, we have thought that not only it would be much more applicative to use an mRNA which is specific to the target cell type/tissue as a trigger RNA, but it also may yield much higher dynamic range. As mentioned before, the trigger RNA in the work of Green et al. was an mRNA molecule. Even though they dealt with a Prokaryotic system, the same principle described earlier of allowing the trigger RNA molecule to be more spaced might work to our benefit and allow a higher dynamic range even in Eukaryotes.

In order to achieve those goals, we might first define an optimized sequence by the extent of achieving high dynamic range. However, high dynamic range will not be achieved solely by a suitable physical design of our construct, but it also depends upon the environment. The trigger could serve well in differentiating between tissues, by enabling a maximal translation rate in one and minimal translation rates in others. Following that, our goals might be achieved by taking into account two considerations when generating our tool: what is the best trigger we may find in a given tissue, and given we have found that trigger, how we might design our toehold construct so it could be efficient in a manner of interacting with its environment and generate translation.

The first consideration, of finding the best trigger, should take care of our defined goal to use an mRNA molecule as a trigger as a part of Eukaryotic applications. By using a larger RNA molecule we may have more degrees of freedom, so a lower crosstalk and a higher dynamic range might be achieved. We plan to find the best trigger of a certain tissue by using statistical search. By scanning big data sources, our algorithm will find the best RNA molecules candidates to serve as a trigger and subsequently rank them according to various characterizations: their existence and abundance at the target tissue in comparison with other tissues and so forth. In case the chosen trigger RNA is quite long, then only a sub-sequence out of the molecule would serve as a trigger. That sub-sequence will be found by using a window search algorithm, so an open and a stable sequence could be chosen to serve as a trigger which interacts efficiently with the Toehold construct.

After the trigger is chosen, we might proceed to the design of the Toehold switch itself. Structural properties of the Toehold switch and thermodynamical characterizations concerning its folding and hybridization with the trigger, might be taken into account. It is also desired to examine both local and global features of its sequence, that will be achieved by detecting GC contents and minimum free energies of certain positions in the sequence at the local stage, in addition to verifying the presence of nucleotides at specific positions of the sequence at the global stage. After relating to all of those features, we might use a regressor trained upon an experimental database in order to simulate the performances of the Toehold construct and test it. In case it surpasses the required threshold, we might proceed to wet lab experiments, performed on both yeast and mammalian cells, so more data could be gathered and be used to optimize our tool even more. While experiments on yeast are suitable for molecular work and represent a more "low-leveled" Eukaryotic system, mammalian cells represent the applicative quality of our project, directing it towards the realm of therapeutics.

We plan first to insert the optimized Toehold construct sequence generated by our tool to a plasmid containing a reporter gene (GFP) using Molecular Biology methods. After that, we will examine in-vitro the efficiency of our optimized Toehold construct. In both yeast cells and mammalian cells, we plan to measure the fluorescence levels generated by the expression of the reporter gene, based on the activity of the switch. By using that integrated approach, of generating a computational tool and examining its predictions throughout wet lab experiments, all of our three goals could be reached.

In order to validate our tool's capabilities, it is of a special interest to test it upon cancerous setting. Since current oncological therapies face a hard time trying to address cancer cells, it may be essential to tackle this problem using our Toehold switch generator. A switch might be designed in a way that enables the activation of it just when the unique settings induced by cancerous cells are sensed. Those unique settings, which differentiate between healthy and cancerous cells, are formed primarily by cancer-related mutations. The design process may be divided into two parts, just as earlier: finding the optimal trigger sequence and designing the switch itself in an optimal manner. Concerning the former, the optimized trigger might be found by utilizing a specific window space, contained within the trigger, which itself contains mutations specific for cancerous genome and not for a healthy one. Those mutations, of course, must induce a significant measure of variance between the healthy and cancerous genomes, while their variance among oncological patients has to be minimal. Once the optimized triggers will be formed, the switches themselves will be designed, based on the expected interactions between those two structures, in a way which is similar to the designing process described earlier.

---

The Importance of the Problem

The importance of the problem could be easily inferred just by watching the news on television: the SARS-CoV-2 pandemic had a gigantic effect upon the world citizens, both healthily and mentally. As already stated, the technology of mRNA vaccines developed in order to face the pandemic was based upon a delivery of an mRNA molecule encapsulated by a lipid-based particle. However, as mentioned earlier, the systemic injection of it could not allow a differential expression of the mRNA in a certain tissue. Enabling the expression of the protein translated from that mRNA to be restricted to a certain tissue could have a major effect upon the efficacy and side-effects of the vaccine. However, differential expression of an mRNA molecule is not relevant only for vaccination, but also holds the potential to generate many other major advances in the field of personalized medicine: chemotherapy treatments aimed to destroy cancerous malignant foci might have deleterious effect not only upon the tumors, but also healthy tissues could suffer a substantial damage. Now let us suppose we would have the ability to limit the expression of a chemotherapy drug to a target cancerous tissue and not others, or even to a specific cancerous cell within a tissue – we may target the tumors directly while the patient gets free from the harsh well-known side-effects.

---

Why Our Project is an Application of Synthetic Biology

Our project utilizes a very basic and influential work performed in the field of Synthetic Biology, allowing it to spread and affect many research projects around the world. However, this work of Green et al. was relevant for Prokaryotic systems, while other works that have attempted to make use of Toehold constructs in Eukaryotic settings have reached much less efficacy. Not only that our project takes an unprecedented approach by combining computational models with wet lab experiments in order to form efficient Toehold switches, but it also tries to do it in the more applicative (yet, more complex) Eukaryotic system. As already explained, the ability to enable differential translation within tissues could have major implications upon the fields of personalized and translational medicine, so it could be understood why we define this ultimate goal as an applicative one. The way by which we address this problem is a one that is based upon a well-based former Synthetic Biology work and upon the very basic principles of Synthetic Biology – utilizing a certain intracellular construct while taking into advantage a certain property of it in order to synthetically gain a function to the welfare of many, a function which could not be achieved endogenously.

---

References

1. Barbier, A. J., Jiang, A. Y., Zhang, P., Wooster, R., & Anderson, D. G. (2022). The clinical progress of mRNA vaccines and immunotherapies. Nature Biotechnology, 1-15.
2. Green, A. A., Silver, P. A., Collins, J. J., & Yin, P. (2014). Toehold switches: de-novo-designed regulators of gene expression. Cell, 159(4), 925-939.
3. Singh, J., Hanson, J., Paliwal, K., & Zhou, Y. (2019). RNA secondary structure prediction using an ensemble of two-dimensional deep neural networks and transfer learning. Nature communications, 10(1), 1-13.
4. The House of Toeholds, Sastra_Thanjavur iGEM 2019 Group.
5. Vahed, S. Z., Salehi, R., Davaran, S., & Sharifi, S. (2017). Liposome-based drug co-delivery systems in cancer cells. Materials Science and Engineering: C, 71, 1327-1341.
6. Wang, S., Emery, N. J., & Liu, A. P. (2019). A novel synthetic toehold switch for microRNA detection in mammalian cells. ACS synthetic biology, 8(5), 1079-1088.
7. Zhu, Y., Zhu, L., Wang, X., & Jin, H. (2022). RNA-based therapeutics: an overview and prospectus. Cell Death & Disease, 13(7), 1-15.