Results

The objective of wet lab will be to generate large amounts of data characterizing the binding affinity of various pairs of RNA sequences. The most straightforward way to quantify the binding affinity would be to build RNA switches.

RNA switches are characterized by the binding of a target RNA to the secondary structure of a switch RNA, altering the structure and triggering downstream translation. Thus, we first had to establish a model RNA switch to work on.

Experiment Objective

The first RNA switch we examined was the toehold switch, as it has been shown to be highly modular and specific(Green et al., 2014). As outlined in our description, we plan to use an oligo pool to generate pairwise variants of different complimentary RNA sequences. This requires the target and the switch RNA to be physically linked on a single strand, whereas the toehold switches in literature have relied on expressing the switch and target from different plasmids.

Thus, the first experiment carried out was to integrate the switch and target expression cassettes into a <300bp stretch of DNA on a single plasmid.

---

Experiment Design

To begin, the toehold sequence specified in "Toehold Switches: De-Novo-Designed Regulators of Gene Expression" (Green et al.,2014) along with it's cognate target was designed as a single plasmid. Expression cassette for the target was placed under the control of an IPTG-inducible pLac promoter, while the toehold switch was placed under the control of an ATC-inducible ptet promoter.  These expression cassettes were placed back to back such that the target to the toehold spanned <300 bases, allowing different pairs of toeholds and targets to be ordered as a single oligo in future. The whole target and toehold switch sequences were ordered and synthesized by IDT and TWIST respectively and they were then assembled into a plasmid called TH01 as shown in the figure below.

It was expected that only after inducing both promoters would the downstream GFP be expressed, thus cells were cultured for 6 hours in selective LB media with no inducers, only IPTG, only ATC, or with the addition of both inducers.

---

Experiment Result

Results show the cells exhibited similar levels of GFP under all four conditions. Since the sequences of the toehold switch and the target are identical to Green et al.(2014) , some other aspect of the 1-Plasmid system must have caused the expression to fail. Green et al.(2014)used a constitutive promoter to express their target and switch. RNA switches may be highly sensitive to their secondary structure, it could be possible that swapping out the promoter altered the transcription start site and the 5' end of the RNA changed their secondary structure and prevented the target and switch from interacting.

---

Conclusion

The experiment was unsuccessful in showing that a single-plasmid system would work for toehold switches. Hence, we decided to pivot to STAR as it was well characterised. In the meantime, we focused on characterising the SacB negative selection marker, as we wanted a selection marker to select for the successful RNA switches.

Experiment Objective

To try a different RNA switch, we examined the STAR system (Chappell et al., 2015).

Similar to the toehold switch, STAR had only been tested on a 2-plasmid system. Thus, we tried a similar approach to integrate the switch and target into a single plasmid.

---

Experiment Design

We obtained the two separate target and switch plasmids from the Addgene. As plasmid copy number and selection markers can greatly affect expression, we strove to stay as close as possible to the system outlined in Chappell et al. (2015). Thus, the target expression cassette was cloned into the plasmid harbouring the switch.   Learning from the previous experiment with the toehold switch, we cloned the target and switch under the constitutive promoters originally found in the paper. Cells were cultured with our new one plasmid system for 6 hours. As a negative control, strains either only containing the switch(plasmid S5) or the target(plasmid A5) were tested, and as a positive control the original 2 plasmid system(S5+A5) was used.

---

Experiment Results

Lack of expression observed from the individual switch and target was expected. Our 1-Plasmid System had shown increased expression compared to the negative controls, however, it was not as high as the original 2-Plasmid System, likely due to an altered ratio between the target and switch.

---

Conclusion

We can conclude that the 1-Plasmid System was successful, but that it was also important to consider the ratio between the target and switch. Due to a lack of time, we did not focus on optimisation. Instead, we next moved on to the STAR Oligo 1-Plasmid System test.

---

Experiment Objective

Now that the STAR system has been consolidated to a single plasmid, we wanted to design a plasmid that could generate a library of different sense-antisense pairs by accepting an oligo pool of various target and switch sequences. The library generating system consists of two parts.

Part 1 would be the oligo pool carrying all the variable regions. A python program was written to randomly generate pairs of complimentary switches and targets along with a unique barcode. This oligo pool was flanked with standardized BsaI golden gate restriction sites to allow for easy assembly. Part 2 was a vector with restriction sites that could accept the BsaI digested oligo strands, such that a golden gate reaction could assemble the generated oligo pool into a library

---

Experiment Design

Before ordering an oligo pool with pairwise variation, we first constructed the vector and a single instance of the designed oligo. Because a larger oligo would take longer to order, in the interest of time we tested it a scaled down version of the oligo library and vector, that would use a constant switch but with variable target sequences.

To test the assembly, we produced a PCR construct with a single target sequence but with the barcoding and restriction sites of our designed oligo pool. Golden gate assembly was carried out between our modified target sequence and the vector and trasnformed into E.coli.

---

Experiment Results

After golden gate assembly and transformation, GFP of resulting oligo insert was shown to be not significantly different from the original 1 plasmid STAR system(checked using a t-test).

---

Conclusion

This experiments shows that the addition of the golden gate sites and barcoding, the GFP expression is not affected. This affirms that the oligo pool can be ordered with the same design and assembly protocol.

---

Experiment Objective

Now that it had been established that the designed assembly was functional, we wanted to test whether a library could be created with an oligo pool of target sequences, and characterize how well preserved the variation of sequences would be in liquid culture. This is crucial as our deep learning model relies on data from a large amount of sequences, and a liquid, heterogenous population of cells is required for our proposed ON/OFF diamteric selection with sucrose and antibiotics.

---

Experiment Design

An oligo pool of various target sequences was ordered according to our design. This library would be made to test strandwise variation, and thus a variety of sequences were generated by taking the cognate target of our current RNA switch, and introducing mismatches in varying amounts and position[insert list of variants with mismatches highlighted here, need a figure] This oligo pool amplified with PCR and the same golden gate experiment was carried out and transformed into E.coli. After transformation, instead of spreading on an agar plate, transformants were inovulated into a 5mL liquid culture of selective LB broth. A PCR reaction amplifying the oligo region was carried out after 1 hour, 2 hours, 3 hours and 24 hours. A FLO-FLG114 Flongle Flow Cells (R10.4.1) kindly sponsored by Oxford Nanopore was then used to do high-throughput sequencing and assess if the culture still maintained a large variety of sequences from the oligo library.

---

Experiment Results

Nanopore sequencing showed a variation of target sequences, even after several hours of culturing. This was important in demonstrating that the assembly preserved the different variants in the oligo pool, such that there would be a sufficient population of different phenotypes to undergo screening and sorting into different bins.

---

Conclusion

Oligo pool assembly would produce a heterogenous culture, which would then provide a variety of sequences to link to GFP outputs of RNA-RNA interactions. This would allow us to produce a large amount of data to train the model on.

---

Experiment Objective

Now that it had been established that the oligo pools could be assembled into our vector to produce a variable library of different switch-target pairs in one culture, we wanted to swap out the constitutive promoter controlling the target RNA for an inducible one. This is key to being able to measure an ON/OFF ratio, which would be the main classifying metric for our model.

---

Experiment Design

A new plasmid (IndP1) was cloned with the target placed under the control of the pTet promoter, thus we expected that the target would only be transcribed in the presence of ATC, unfolding the switch and producing GFP.

---

Experiment Results

Results showed that no expression was observed with or witout ATC compared to the system containing constitutive promoters. To perform a diagnostic, the plasmid was recloned without the the STAR terminator. From the result, we could see that without the STAR harpin, GFP was expressed properly, indicating the specific change in the constitutive promoter controlling the target to the pTet promoter caused the switch to be insensitive towards the target. A further test was conducted on the pTet promoter. It was cloned upstream of the GFP gene, and tested with and without aTc.

As can be seen from the graph, GFP expression was successfully induced in the presence of ATC.

---

Conclusion

We were unable to turn the STAR system to an inducible system for the measurement of ON/OFF ratio. We suspect that this may result from the same issue as the toehold switch, where changes in promoters change the overall transcript of the target, affecting it's interaction with the switch. This demonstrates that the system may not be as modular as we initially thought, and that the RNA target cannot be viewed separately from the promoter. More importantly, this further highlights the need for better tools to fully understand RNA-RNA interactions, as small changes such as transcription start site can cause criticial downstream failure for RNA tools if not properly characterized.

---

Experiment Objective

As part of our comprehensive workflow to capture ON/OFF information, we developed the idea of a diametric marker that could directly screen for that characteristic despite ON and OFF states being mutually exclusive. This marker would consist of a fusion of a negative selection marker, a positive selection marker and GFP. To develop this tool, we first wanted to characterize a negative selection marker fused to GFP. We chose the sacB gene as a well characterized negative selection marker. In the presence of sucrose, sacB converts sucrose into levan, which is toxic to E.coli. 

---

Experiment Design

The sacB gene was fused to the C-terminal of GFP, and the fusion protein was placed under the control of the IPTG inducible pLac promoter. We expected that varying levels of expression through different IPTG concentrations would give different responses to scaling sucrose concentrations.

Cells were cultivated in 0%, 0.5%, 1%, 2.5%, 5% of sucrose and 0, 25, 50, 100μM of IPTG, and their OD600 as well as GFP fluorescence taken over a period of 6 hours to track cell growth.

---

Experiment Results

GFP expression levels across different concentrations of IPTG without sucrose demonstrated that we would expect IPTG and marker expression to be positively correlated. GFP cannot be used as an accurate indicator in the presence of expressed marker as cell death will influence protein production. As can be seen from Figure 6a, even at 0μM IPTG, growth curve of the cell deteriorates with increasing concentrations of sucrose. This can indicate 1) that there is significant leakiness of the pLac promoter, expressing enough marker to cause cell death even in the absence of the inducer; or 2) sucrose has a negative effect on cell growth independent of sacB. In the presence of any concentration of IPTG, concentration of sucrose 1% and above caused immediate stunting of growth. However, for 0.5% sucrose, all curves had a distinct peak shape, where cells grew for a period of time before entering a death phase with decreasing OD600. Importantly, the speed at which the population entered death phase was linearly correlated with the IPTG concentration.

---

Conclusion

The negative selection marker component of our diametric marker is meant to screen against leaky variants by killing them in culture. We now have found the ideal concentration of sucrose to use, 0.5%, which would allow us to tune the level of 'acceptable' leakiness. This acceptable leakiness can be quantified using the IPTG concentrations as a proxy. For example, based on the graphs, allowing cells to grow in 0.5% sucrose for 2 hours would guarantee that surviving cells had constructs that had leakiness equivalent to or less than a pLac promoter's expression at 25uM IPTG, while growing it for only 1 hour would loosen the upper limit of leakiness, allowing any cell that had leakiness up to the equivalent of a pLac promoter's expression at 100uM IPTG to survive in the media.

---

Experiment Objective

In order for the selection marker to be able to select for individual variants in a culture, it must directly link the surviviability of the cell to the variant. As the mechanism of sacB is not well understood, it is important that sucrose converted to levan in one cell does not affect the surviviability of other cells, i.e by levan crossing the cell membrane/leaked by lysed cells and killing cells that do not have leaky RNA switches. Thus, we conducted an experiment to verify the effect that intracellular sacB could have on a non-sacB producing population of cells. 

---

Experiment Design

Cells expressing our sacB-GFP fusion protein under an IPTG inducible promoter were cocultured with cells producing RFP. Several cocultures were made, with different starting proportions of the 2 cell populations. All populations were exposed to either 100uM IPTG(to induce sacB-GFP production) only, 1% sucrose only, both conditions or neither, and their growth tracked over 10 hours.

---

Experiment Results

Figure 8 and 9 show the expression of RFP over time in each of the different conditions. As the OD600 cannot be used to directly differentiate the cells expressing RFP and the cells expresssing SacB-GFP, RFP was assumed to be directly proportionate with the amount of RFP-expressing cells. From the condition without any additives, we can see that the amount of RFP produced over time is correlated with the seed proportion of RFP-expressing cells. From the condition where 1% Sucrose and 100uM IPTG are added, it can be seen that there is significantly more RFP expression independent of the seed proportion.

Figure 10 and 11 show the overall OD600 of each coculture seed proportion in each condition over 6 hours.  Across all conditons the addition of sucrose caused the OD to drop slightly, even for cultures that did not have any sacB-GFP producing cells. As expected, the addition of sucrose and IPTG to cultures that were fully sacB-GFP expressing cells led to a collapse in OD600, similar to what was observed in the sacB-GFP characterization. However, for cocultures, the addition of both IPTG and sucrose saw a restoration of normal growth compared to the addition of sucrose only, except for cultures that were fully RFP-expressing cells.

---

Conclusion

As RFP-expressing cells continue to grow in 1% sucrose and IPTG, conditions that, from our data, evidently produce sufficient levan to kill the sacB-GFP producing cells, we can conclude that levan only negatively affects cells that sacB-GFP was produced in. In addition, it seems that sucrose has a negative effect on cell growth, independent of sacB expression. This effect is relieved in the presence of IPTG and sacB-GFP producing cells. Our hypothesis is that sacB-GFP producing cells metabolize the sucrose in the media into levan, which is evidently kept intracellular. Even though this kills the sacB-GFP producing cells, it removes the sucrose from the media, alleviating the negative effect on other cells. Thus, not only does sacB not negatively affect the growth of surrounding cells, it provides a growth advantage to non-sacB producing cells, which assists in our use of the marker to distinguish leaky and non-leaky RNA switches.

---

Experiment Objective

Our workflow for screening ON/OFF strength of the RNA switch should in theory be applicable to any switch that controls protein expression. In order to save time, we developed our workflow in parallel to our attempts to create a library generation system for the STAR RNA switch. To test our workflow independently of the RNA switch, we focused our efforts on the pLac promoter, a well characterized and commonly used promoter in E.coli.  

---

Experiment Design

To simulate a variety of switches with different ON/OFF ratios and leakiness levels, we decided to conduct random mutagenesis on the lacI gene in the plasmid expressing IPTG inducible sacB-GFP, followed by gibson assembly back into the sacB-GFP expression vector. The resulting transformants were inoculated into liquid culture, and it was expected that the culture would contain a heterogeneous population of different lacI variants with different properties which could be screened by the sacB-GFP marker.

---

Experiment Results

Firstly, the result of the error-prone PCR was sequenced with FLO-FLG114 Flongle Flow Cells (R10.4.1) to verify that variants were being created.

Gibson assembly was then carried out with the mixture of mutants and transformed into E.coli. Transformation was plated, and 10 colonies cultured and plasmid extracted. After sequencing, it was verified that transformations of this protocol all had different mutations in the lacI gene, and thus all 10 plasmids were cultured in varying concentrations of IPTG to judge their response.

Different variants were shown to have different responses to IPTG, ranging from increased expression to increased leakiness or insensitivity to IPTG. Thus we established that error-prone PCR could produce a variety of phenotypes to check our screening against.

---

Conclusion

Experiment demonstrates that mutagenesis produces different phenotypes for the lacI protein, and they all have different responses to IPTG. This would allow us to use the various mutants of lacI as a base pool to test the capability of our diameteric selection marker to screen out the leaky variants of switches.

---