Team:OUC-China/Results
1. Overview
To make riboswitch more accessible to future iGEM teams, our project focused on a standardized principle to construct modular riboswitch. The modular riboswitch we defined consists of the original riboswitch, Stabilizer and Tuner. Stabilizer can protect the structure of riboswitch from damage while Tuner can reduce the expression probability of fusion protein and make improvement of riboswitch function. When we construct a modular riboswitch, some aspects should be considered.
(1) The design principle of the ‘Stabilizer’, including the strategy for sequence selection, truncation and the modeling tool for its function prediction.
(2) The toolkit with a library of well-characterized ‘Tuners’ designed by the design principle we set up, including the strategy for sequence design and the modeling tool for its function prediction.
(3) We combined the well-designed ‘Stabilizers’ and ‘Tuners’ with different riboswitches and downstream genes.
We confirmed that our design principle could be applied in four kinds of riboswitches including three kinetic switches: Adda riboswitch, Btub riboswitch, cobalamin biosensor, and one thermodynamic switch: FourU riboswitch. Further, to ensure that our modular riboswitches will work with a variety of different proteins, three different kinds of GOI is used including sfGFP, YFP, and mRFP1. To help you enjoy our results in a more convenient and smooth way, We will demonstrate our modular riboswitches respectively.
2. RiboLego based on Adda
— The toolkits and the rational design principles
2.1 Overview
First of all, we successfully set up a basic ‘Ribo-attenuator’ device for in-depth study of its design principles.
The translation-activating riboswitch Adda from Vibrio vulnificus can regulate the expression of the adenosine deaminase by binding 2-aminopurine (2-AP). We employed this well-studied riboswitch as a platform for our research. For demonstrating that a ‘Ribo-attenuator’ device could improve the modularization and protect the function of a particular riboswitch, we design three circuits (Figure 1):
(1) A sfGFP gene controlled by an Adda riboswitch. (circuit 1)
(2) A sfGFP gene controlled by an Adda riboswitch with the ‘Stabilizer’ to insulate CDS of sfGFP and the riboswitch. The translated ‘Stabilizer’ will become a N-terminal fused peptide of sfGFP. (circuit 2)
(3) A sfGFP gene controlled by an Adda riboswitch with the ‘Stabilizer’ and a ‘Tuner’ which could avoid the translated ‘Stabilizer’ forming an N-terminal peptide on the sfGFP. (circuit 3)
Figure 1: The three plasmids were constructed to test whether our modular Adda riboswitch can express sfGFP normally.
In above designs, the first 150bp of the CDS of the adenosine deaminase gene was chosen as the ‘Stabilizer’ in the circuit 2 & 3.
Our docking matrix suggested that a functional riboswitch structure would be observed under this design. Tuner A was inserted between Stabilizer and sfGFP in the circuit 3. Adda riboswitch was under control of the anhydrotetracycline (aTc)-induced promoter.
After circuit construction, we transformed them into E. coli DH5AlphaZ1. By fluorescence microscopy imaging (Confocal Microscopy Leica TCS SP8) the strains under aTc induction, it’s obvious that the fluorescence signal from the circuit 1 is significant weaker than that of the two circuits with ‘Stabilizer’. The result reflects that the ‘Stabilizer’ could relieve the functional suppression on the riboswitch. Comparing the image result of circuit 2 & 3, we could find obvious inclusion bodies in the cells with circuit 2, which shows that the ‘Tuner’ has ability to prevent the protein folding and dysfunction effect from the translated ‘Stabilizer’.(Figure 2)
Figure 2: The fluorescence images by confocal microscopy after 6 hours. The images show E.coli with 2-aminopurine. There is no fluorescence in the E.coli when the Adda riboswitch has sfGFP introduced directly. Instead, Adda fusion shows clear inclusion bodies and the modular Adda riboswitch shows working sfGFP.
Further, a quantitative experiment was employed to exhibit the regulation dynamics of above three versions of riboswitch (Figure 3A, B), the result shows that the ribo-attenuator is necessary for the normal function of Adda riboswitch in this case.
Figure 3: (A) The intensity of sfGFP by microplate reader during the entire cultivation period. We use E.coli DH5AlphaZ1 competent cell as our negative control, which is the green line. The deep blue line refers to the recombinant strain containing the circuit where sfGFP is introduced directly behind Adda riboswitch with aTc while the pink one refers to the same strain with aTc and 2-AP (100nM). Red and purple lines refers to a strain transformed with a circuit where the modular Adda riboswitch control the expression of sfGFP, respectively with 0μM and 100μM of 2-AP. Error bars represent standard deviation of three biological replicates. It can prove that the modular Adda riboswitch can control the downstream gene expression during the whole cultivation period. (B) The function of riboswitch can be improved by Ribo-attenuator.
2.2 The rational design of the length of ‘Stabilizer’
The stabilizer needs to be long enough to insulate the riboswitch and the downstream CDS of the desired protein. However, how the length of the ‘Stabilizer’ affects its function is still unclear. After successfully testing the capability of the modular Adda riboswitch, we then want to explore the design principle of the length of the ‘Stabilizer’ by modeling method
It is not difficult to imagine that the suitable effective length is case by case for different riboswitches, so we have to focus on our Adda riboswitch platform to verify our model. Using docking matrix, we selected four lengths of Stabilizer, called STA9, STA21, STA81 and STA129, and predicted that the STA9 and STA21 could affect the functional secondary structure of the Adda riboswitch because they all have high levels of punishment score. In contrast, as the punishment score of STA81 and STA129 both close to 1, they could stabilize the secondary structure of riboswitch theoretically.
Then, we used microplate reader to test the response curve of the Adda riboswitch circuits containing the ‘Stabilizers’ with different lengths we design above (Figure 4). Tuner A was inserted between Stabilizer and sfGFP. The result indicates that the STA9 and STA21 couldn’t protect the riboswitch, leading to a failure of responsive to ligand. The STA81 and STA129 could protect the functional structure of riboswitch. (Figure 5)
Figure 4: The four plasmids were constructed to test whether our program could predict the appropriate length of Stabilizer.
Figure 5: The intensity of sfGFP by microplate reader during the entire cultivation period. For four kinds of strain, we respectively added different concentrations of 2-AP and 150μM aTc. It’s obvious that both STA81 and STA129 fail to stabilize the structure of Adda riboswitch while STA81 and STA129 can form normal response curve.
This experimental data perfectly matches the prediction result from our dry team!
2.3 Two reliable source of the sequence of ‘Stabilizer’
The wet lab’s measurements have demonstrated that the length of Stabilizer could be determined by the program developed by our dry group. As we have mentioned above, the suitable length is case by case for different riboswitches because of their sequence difference. Hence, another factor to consider for ‘Stabilizer’ design is the source of its sequence, comparing with exhaustive strategies like SELEX which may consume huge cost, we figured out 2 priority sources of the ‘Stabilizer’ sequence for particular riboswitch:
(1) We assume that the natural context sequence is best suited as the ‘Stabilizer’.
(2) We can choose the gene working well under the riboswitch from the previous work.
In previous experiments, we have used the natural context sequence as the ‘Stabilizer’ and confirmed the compatibility of that strategy with our modeling works. Now, we move on to test the second one.
By referencing past study, we found that Adda riboswitch can be constructed with gfp directly and yielded an induction response successfully. So, we use the RNAfold to predict the structure of riboswitch with gfp CDS as its adjacent downstream sequence.
The result indicated that the CDS of GFP hardly influences the function of Adda riboswitch. After running our program, we then constructed the first 150bp of gfp gene as the ‘Stabilizer’ of Adda riboswitch, named STA (GFP). Using STA (GFP), a new circuit was designed. The results measured by plate reader and flow cytometry showed that this circuit could generate a normal induction response curve to 2-aminopurine concentrations (Figure 6).
Figure 6:(A) We constructed this circuit to verify the different origin of Stabilizer. Adda riboswitch was under control of the anhydrotetracycline promoter. Tuner A was inserted between STA (GFP) and sfGFP, the report gene. (B)The fluorescence intensities of sfGFP by microplate reader. The deep blue line refers to the recombinant strain with aTc and 2-AP while the purple one refers to the same strain with aTc.(C)The results of flow cytometricy. The sky blue group represents the recombinant strain with aTc. And the pink one represents the recombinant strain with aTc and 2-AP.
2.4 Rationally designed library of ‘Tuners’ achieving diverse dynamics of ‘Ribo-attenuators’
After summarizing some basic design principles and verifying the usability of our model, we then attempted to enrich the toolbox of ‘Ribo-attenuators’ by diversifying their dynamics.
We realized that ‘Tuner’ would be a perfect part to diversify the behaviors of the Ribo-attenuators. Because it has direct effect on the function of the RBS of the downstream CDS. We created other four Tuners (B-E) and evaluated whether these Tuners are capable of affecting the response behavior of a riboswitch in a predictable way. The four modular Adda riboswitches were engineered with STA150 and these Tuners perspectivity (Figure 7).
Figure 7:The five circuits were constructed to verify that different Tuners located downstream of the same riboswitch sequence can implement diversity of response curve.
After transforming these plasmids into four strains respectively, we tested these recombinant strains individually by plate reader and flow cytometer. As we predicted, the four test groups show different characteristics (Figure 8-10).
Figure 8: The fluorescence intensities of sfGFP by microplate reader. We set the competent cell as the negative control. Each modular Adda riboswitch was compared to the original one. The results are listed in the order: from Tuner A to E. Error bars represent standard deviation of three biological replicates.
Figure 9: The summary of five Tuner groups which indicated that our system could be successfully applied to regulate the translation of another gene. The figure shows the data collected at steady state
(at least two consecutive subsequent data points do not increase fluorescence).
Besides all the works before, we also tested our Tuners by flow cytometer.
Figure 10: The results of modular Adda riboswitches containing different Tuners by flow cytometer. Tuner A: Red; Tuner B: Blue; Tuner C: Orange; Tuner D: Green; Tuner E: Dark Green.
By all the experiments mentioned above, we proved that the Tuners work as expectations successfully. They are expected to serve as a powerful and tunable tool of riboswitch for synthetic biologists and future iGEM teams.
2.5 Modular verification by replacing downstream gene
To ensure that our modular riboswitches are modularized and will work with a variety of different genes, we substituted sfGFP with EYFP. Using the eyfp gene, we tested the effect of Adda riboswitch based ribo-attenuators consisting the original Adda riboswitch, STA150 and Tuner A. The result by microplate reader has been shown below (Figure 11). An obvious induction curve was observed in response to the concentration gradient of 2-AP, which indicated that our system could be successfully applied to regulate another protein. The figure shows the data when steady state is reached. (at least two consecutive subsequent data points do not increase fluorescence).
Figure 11: (A) The plasmid was constructd to prove that our modular riboswitch could work with different proteins. (B) The result by microplate reader. The emission of EYFP was measured at a wavelength of 527nm when excited at 514nm. Error bars represent standard deviation of three biological replicates.
3. RiboLego based on Btub
3.1 Modular Btub riboswitch in E. coli based on the RiboLego design principle
By employing the rational design approach of Ribo-attenuator on the Adda riboswitch, we found it’s possible to use our design principle to optimize the function of other riboswitches in a general way. So, we chose the Btub riboswitch, a repressing riboswitch which responds to adenosylcobalamin to verify the universal applicability of our design rules and models. According to the prediction of the behavior of the ‘Stabilizer’ for the Btub riboswitch by our modeling, the first 150bp of btuB gene which is its natural downstream gene was used to serve as the Stabilizer. Besides, Tuner A was inserted between the Stabilizer and sfGFP (Figure 12).
Figure 12: The three plasmids were constructed to test whether our modular Btub riboswitch can express sfGFP normally.
We then proved the function of modular Btub riboswitch with a qualitative experiment. The fluorescence images showed that only the circuit including Btub riboswitch with the RiboLego design worked as a functional riboswitch, which responded to the VB12 (Figure 13).
Figure 13: The fluorescence images by confocal microscopy after 6 hours. All the images shows E.coli with VB12. There is no fluorescence in the E.coli when the Btub riboswitch has sfGFP introduced directly. Instead, the induced modular Btub riboswitch can show a greater induction difference than the induced Btub fusion.
This result indicates that the rationally designed RiboLego could be used on various riboswitch to protect their function.
3.2 The rationally designed library of ‘Tuners’ on the modular Btub riboswitch
As we have done on the Adda riboswitch, we utilized Tuner A and Tuner E to confirm that our ‘Tuners’ are reliable and tunable in different riboswitches. The results by microplate reader showed that the Tuners could not only be used in the Adda riboswitch, but also in the Btub riboswitch. (Figure14, 15, 16)
Figure 14: The plasmids were constructed to verify Tuners can be utilized to engnieer modular Btub riboswitches.
Figure 15: The intensity of sfGFP by microplate reader during the entire cultivation period. The results present the high fluorescence intensity in Tuner A while the low one in Tuner E with the same concentration of VB12. These constructs were compared with the original Btub riboswitch respectively.
Figure 16: The summary of two Tuner groups which indicates the capabilities of different modular Btub riboswitches, compared with Btub fushion construct. The expression of sfGFP behind Tuner A and E is showed and their intensities of fluorescence are from strong to weak. The data was selected when steady state is reached(at least two consecutive subsequent data points do not increase fluorescence).
The results above indicated that the relationship of the expression levels controlled by the modular Btub riboswitches with Tuner A or Tuner E is similar to those in modular Adda riboswitch. The data from flow cytometry shows the same trend (Figure 17).
After measuring the gene expression during entire cultivation period, we further tested the regulatory capacities of Tuner A and Tuner E with flow cytometry. As we can see, the expression of sfGFP regulated by Tuner A and Tuner E were show below, and and their intensities of fluorescence are from strong to weak.
Figure 17: The results of modular Btub riboswitches containing Tuner A and Tuner E by flow cytometer.(A)The sky blue group represents E.coli with 2-AP and the purple one represents E.coli without 2-AP. (B)The sky blue group represents E.coli without 2-AP and the orange one represents E.coli with 2-AP.
By all the experiments mentioned above, we proved that our design principle is capable to work well in two kinds of riboswitches as expectations successfully. So, it’s time to build more RiboLego!
——Resurrection of a cobalamin biosensor by employing our RiboLego design principle
Our team’s vision is a standardized and easily adaptable design principle to be used for various riboswitch. So, we hope to prove the ability of our design principles and tools by resurrecting one example of a functional defect riboswitch.
In 2015, Team Paris_Bettencourt used a riboswitch found in ,Propionibacterium shermanii, whose ligand is vitamin B12 (VB12), the VB12 could inhibit the expression of the reporter gene through this riboswitch. In their design, the egfp gene was directly inserted at the downstream of the riboswitch, and they could not detected the fluorescent signal of egfp. After that, although they substituted egfp with mrfp1 and inserted a piece of natural context sequence between mrfp1 gene and the riboswitch, the result was still negative result.
By analyzing their results, we reasonably assumed that the negative result was due to the unpredictable interference effects on the riboswitch without the protection by a Ribo-attenuator. We decided to use the set of design principles we proposed to design a RiboLego for this biosensor, to recover its function.
To test the original function of this biosensor, we repeated their experiment first (Figure 19). To our surprise, the intensity of mRFP1 was very low even without cobalamin. Then we also observed the fluorescence by confocal microscopy and got the same result.
With our principle, this construct can be optimized. We engineered a modular cobalamin biosensor consisting of the repressing riboswitch, the first 144bp of the natural context sequence of the biosensor as its ‘stabilizer’ and the Tuner A to support the modularized expression of mRFP1. To test the function of the improved circuit, we measured the fluorescence level emitted by our strain in the presence of increasing levels of ado-cobalamin by microplate reader. We could observe an obvious decrease in mRFP1 expression with the ado-cobalamin concentration increasing, which demonstrated that the riboswitch was improved successfully! (Figure 20)
Figure 18: The comparison of two parts, BBa_K1678007 constructed by Paris_Bettencourt in 2015 and BBa_K improved by OUC-China in 2019.
Figure 19: The results by confocal microscopy, which indicates that our principle can improve cobalamin biosensor successfully. It's obvious that the modular cobalamin riboswitch can express mRFP1.
Figure 20: The response curve of our improved cobalamin biosensor to Vb12. As we expected, an obvious decrease of mRFP1 expression was observed with the increased concentration of vb12. It demonstrates that the modular riboswitch functions properly.
5. RiboLego based on thermodynamic switches
Although we have verified our design tools and principles on various kinetic switches in above experiments, another main type of riboswitch known as the thermodynamic switches has not been tested.
Hence, we utilized thermodynamic switches to validate the powerful ability of our design tools. Four U, a common thermodynamic switch with a temperature threshold is 37℃. According to our previously mentioned ‘stabilizer’ sequence design rules, we found that the Four U can be combined with gene rfp directly with normal function in previous research, so we use the first 150bp of the CDS of RFP as its Stabilizer. In this circuit, Tuner A was introduced between Stabilizer and sfGFP (Figure 21).
We set three culture temperatures: 28°C, 30°C and 37°C. To measure the expression level of the reporter regulated by modular Four U riboswitch, we used the plate reader and chose 5 time points to test the fluorescence intensity of sfGFP. It’s obvious that the fluorescence intensity in 42℃ was higher than that in 37℃ but there is almost no fluorescence in 28℃ (Figure 21).
Figure 21:The results of microplate reader show the working effect of modular Four U element in different temperatures.
The results demonstrated that the Four U element is functional under our RiboLego design.
6. A novel asRNA-mediated trans-regulatory device for forced weakening of the riboswitch
6.1 Design
During lab condition, the method to turn down the function a riboswitch (activation or repression) is clearing the inducer in the culture system. Because the ligands are generally hard to be degraded.
However, in many actual industrial situations, it is very difficult and high cost to clear the ligands, which limited the application potential of riboswitch in practical field. Here, we designed and achieved an additional asRNA-mediated trans-regulatory device which could forcefully turn down the function of riboswitches without ligand-clear condition
By comparing different RNA regulation elements, we finally decided to utilize antisense RNA to tackle this problem. Antisense RNA is thought to be consist of two regions: a target binding region (TBR) containing a sequence that is complementary to the target gene, and an Hfq binding site which allows for binding of the Hfq protein.
In our work, the engineered MicF binding site (MicF M7.4) was used as a Hfq binding site, because it performed well with low off-target effect in previous studies. So, by changing TBR, we can design different asRNA with various regulation efficiency of asRNA on the riboswitches.
6.2 Proof of concept
A two-plasmid system was constructed to characterize the capability of asRNA to inactivate a riboswitch when ligand is present. One plasmid can transcribe modular Adda riboswitch system with sfGFP as reporter controlled by aTc-induced promoter, another can transcribe asRNA using the IPTG-induced Plac promoter. At first, we attempted to inhibit the expression of sfGFP by targeting the modular Adda riboswitch with asRNA. By adding IPTG and aTc at the same time, we observed a significant decrease of sfGFP level compared with the group without asRNA induction (Figure 22).
Figure 22: The intensity of sfGFP by microplate reader during the entire cultivation period. The blue group represents the recombinant strain transformed with modular riboswitch system and asRNA when adding aTc, 2-AP and IPTG at the same time. The red one represents the strain transformed with the modular riboswitch system when aTc and 2-AP is present.
6.3 asRNA on riboswitch with another ‘Tuner’.
After combining asRNA with modular Adda riboswitch containing Tuner A, we also want to verify whether it can be used to regulate the riboswitch including other Tuners. Tuner E, which can achieve the lowest expression level was selected to verify the effect of asRNA. The results by microplate reader indicated that asRNA can also decrease the expression (Figure 23).
Figure 23: The intensity of sfGFP by microplate reader during the entire cultivation period. It can verify that asRNA can be used to inactivate the modular Adda riboswitch including Tuner E when 2-AP exists.
6.4 asRNA on riboswitch with another ‘Stabilizer’.
Then we designed asRNA to deactivate modular Adda riboswitch which has another ‘Stabilizer’ as the first 150bp of gfp. A significant difference in fluorescence intensity was observed between cells with or without asRNA induction (Figure 24).
Figure 24: By microplate reader, we can observe a obvious decrease of fluorescence intensity between cells transcribing asRNA and cells transcribing both asRNA and modular riboswitch system.
6.5 asRNA on the translation-repressing riboswitch instead of the translation-activating riboswitch
After testing the translation-activating riboswitch, the translation-repressing riboswitch was also utilized to be combined with asRNA to turn on its output in the presence of the ligand. We employed modular Btub riboswitch containing Tuner E and designed asRNA to target the RBS of Tuner E. By adding IPTG and aTc, an increase of fluorescence intensity can be observed compared with the riboswitch system only induced by aTc (Figure 25).
Figure 25: The results of microplate reader can indicate that asRNA has the ability of activating the state of modular Btub riboswitch even in the presence of VB12.
Finally, we summarized the results and made a heat map in which you can see the obvious inactivation or activation efficiencies (Figure 26).
Figure 26: The heat map can summarize the inactivation or activation efficiencies.
7. A running research for further optimizing our RiboLego device
With the knowledge that the accumulation of the peptide translated from the ‘Stabilizer’ may lead to increased metabolic pressure on cells, affecting cell function and the expression of target genes. We decided to design a Tuner S containing ssrA protein degradation tag to degrade the Stabilizer-codon peptide. By wet lab experiments, we can test whether Tuner S is able to regulate the response range. But sadly, due to the lack of corresponding antibodies, we didn’t verify whether Tuner S has improved the degradation of the Stabilizer-codon peptide.
The preliminary test data from this device is showed in the Figure 27,28and 29.
Figure 27: The plasmids were constructed which contain TunerS.
Figure 28: Tuner S microplate reader experimental data plot, the results show Tuner S is able to regulate the response range.
Figure 29:The results of microplate reader present the high fluorescence intensity in Tuner A while the low one in Tuner E with the same concentration of VB12. Furthermore, Tuner S show the medium fluorescence intensity. These constructs were compared with the original Btub riboswitch respectively.
8. Future work
In the future, we will verify that Tuner S containing SsrA can degrade Stabilizer.
Besides, after successfully utilizing asRNA to regulate the on-off state of riboswitch, we’d like to achieve it in real time.
Furthermore, with the use of cell-free system, asRNA is wrapped in vesicles to regulate the on-off state of riboswitch in real time and finally the repeated reversible regulation will be expected.
