Team:OUC-China/Design
1. Background
Riboswitches have been discovered and characterized across numerous prokaryotes and eukaryotes. They are RNA parts that bind small molecules to regulate gene regulation. Riboswitches contain aptamer domain, comprising highly specific pockets in the 5’ untranslated region (5’-UTR) of the mRNAs that bind small molecules or ligands. Once a ligand selectively binds the aptamer, a conformational change in the RNA structure will leads to a change in gene translation.
For synthetic biology application, riboswitches should be modular devices. However, the riboswitch’s specific RNA secondary structure is influenced not only by its own sequence, but also by the surrounding context including the proximal open reading frame (ORF) under the control of the riboswitch. Thus, substituting the original ORF with a new one can Interfere with its function, which strongly limits its applications scope. To overcome this lack of modularity, previous study created insulator for riboswitches, , the first few hundred base pairs of the ORF was changed to an inserted standard sequence, which named ‘Stabilizer’. A particular riboswitch and the sequence of the ‘Stabilizer’ are fixedly matched to insulate the riboswitch from the replaceable parts.
However, this approach fails in many circumstances as the N-terminal fused peptide translated from the ‘Stabilizer’ region may affect the folding and function of the desired protein, leading to unpredictable results. In 2017, scientists invented a ‘Riboattenuator’ device which separates the translation of the ‘Stabilizer’ and the desired coding sequences (CDS) by a ribosome sliding-mediated chain triggered RBS exposure. Thus, the ‘Ribo-attenuator’ keeps the translational regulation ability of the riboswitch, but avoid the peptide translated from the ‘Stabilizer’ becoming the N-terminus of the desired protein.
2. Our Pace
The goal of our work is to propose a standardized design principle named “RiboLego”, based on the research of ‘Riboattenuator’ mechanism. We planned to make design and construction of modular riboswitch faster, easier and achieve more diverse regulation regulatory dynamics. A typical ‘Riboattenuator’ contains three functional elements, including the original riboswitch, ‘Stabilizer’ and ‘Tuner’ arranged from 5'to 3' direction of the mRNA.
The ‘Stabilizer’ can protect the structure of riboswitch from interference. Two factors need to be considered when designing a ‘Stabilizer’, the source and length. There are many sources for users to choose, such as sequences from high-throughput screening methods, acquisition from the natural context of the riboswitch or an available sequence has been characterized in previous studies.
We assumed that the natural context sequence of a particular riboswitch is the best sequence source for the ‘Stabilizer’ and planned to research that if this principle is universally applicable to various riboswitches. Then, as for its length, dry group members will help us determine it by docking matrix and RNAfold. We hope to verified the modeling result by wet experiments.
The ‘Tuner’ can couple the translation process of the ‘Stabilizer’ and the desired CDS, and prevent the translated ‘Stabilizer’ becoming the N-terminus of desired protein which may cause its unpredictable folding and dysfunction. By researching some key parameters of the ‘Tuner’, we will provide a series of ‘Tuners’ to regulate the response curve of riboswitch, achieving various dynamics.
To this end, we planned to designe an innovative software named RiboLego according to our model and experimental data, which could be used to design a modular riboswitch in the later stages based on the user's CDS sequence and desired expression level.
In order to reach our ambitious goal, we employed Adda riboswitch to demonstrate the usability of the basic design principle. Based on the design principle from our research, more riboswitches will be engineered into RiboLego to indicate the versatility of our design tools and rules.
3. RiboLego based on Adda
3.1 Tuner
3.1.1 The structure of the Tuner
Adda riboswitch from Vibrio vulnificus is an riboswitch for translational activation which is responsive to 2-aminopurine. When 2-aminopurine exists, it can bind the aptamer domain of riboswitch, causing a structural rearrangement which can open up RBS, so GOI can be translated. In Vibrio vulnificus, the gene which locates downstream of the Adda riboswitch is adenosine deaminase. To protect the structure of Adda riboswitch from being destroyed by GOI, we truncated the first 150bp of this gene as the ‘Stabilizer’ of modularized Adda riboswitch. Because our docking matrix suggested that a normal riboswitch structure would be observed when using 150bp as its length. Then a typical ‘Tuner’ was created and utilized to deal with the potential dysfunction of desired protein.
In Vibrio vulnificus, the gene which locates downstream of the Adda riboswitch is adenosine deaminase. To protect the structure of Adda riboswitch from being destoryed by GOI, we truncated the first 150bp of this gene as the 'Stabilizer' of modularized Adda riboswitch. Because our docking matrix suggested that a normal riboswitch structure would be observed when using 150bp as its length. Then a typical ‘Tuner’ was created and utilized to deal with the potential dysfunction of desired protein.
We defined a ‘Tuner’ element as a part including a repressing region, an RBS region and a coupled junction region. The repressing region is the reverse complement of a subsequence of the RBS region so that the ‘Tuner’ can form a hairpin with appropriate ΔG. The stop codon and start codon fused in the junction region. Ribosomes recruited by the upstream riboswitch can open up the hairpin of the ‘Tuner’ before dissociation at the stop codon in the junction region. Additional ribosomes can then assemble at the Tuner’s RBS and initiate translation at the first start codon of the introduced downstream gene of interest. Therefore, the ‘Tuner’ can couple the riboswitch’s response and the translation of desired protein.
The superfolder green fluorescent protein (sfGFP) is the reporter gene to verify our modular Adda riboswitches, which is under control of the tetracycline promoter(Ptet). The results indicate that the protein dysfunction of the sfGFP reduced obviously.
3.1.2 More Tuners
In practical applications, the various application scenarios and associated systems may need different dynamic characteristics of the riboswitch. Obviously, the limited number of the wild type riboswitches cannot provide optional regulatory parts as a toolkit.
Talking with Professor Zhang Dawei in the 5th Synthetic Biology Young Scholar Forum, we found it was necessary to make the function curve of riboswitch diverse. For example, in different environments, the same riboswitch is required to have different response ranges. The yellow response curve with high output range may be more appropriate for the regulation of enzymes with low catalytic activity, the blue medium one may be more appropriate for layered regulatory networks that require a high output dynamic range. The red one with low output range may be more appropriate for the regulation of cytotoxic genes.
According to the mechanism of ‘Riboattenuator’, We realized that the ‘Tuner’ could be a perfect target to create the dynamics diversity. In order to meet this requirements, modeling help us to create more 'Tuners'. Five Tuners with various dynamic characteristics were selected as a part collection. According to the expression strength, we named Tuner A to E.
To demonstrate whether the ‘Tuners’ can make the response dynamics diverse, we created five modular Adda riboswitches by combining the original Adda riboswitch, Stabilizer and Tuner A to E respectively.
Wet experiments show that our system can work well! Click results for more details!
3.2 Stabilizer
3.2.1 Source of Stabilizer
To stabilize the structure of riboswitch, many studies create a fixed sequence for a particular riboswitch in front of the GOI, which we named ‘Stabilizer’. Stabilizer can protect the structure of riboswitch from unpredictable interference by changeable downstream CDS. The source and length are two factors need to be considered when design the ‘Stabilizer’, as we mentioned above.
There are many ways to select the source of Stabilizer. However, we assumed that the natural context sequence of a particular riboswitch is the best sequence source for the ‘Stabilizer’, and this principle is universally applicable to various riboswitches.
In order to initially verify this idea, we utilized blast to catch the nature gene downstream of Adda riboswitch as the ‘Stabilizer’
Furthermore, in order to verify that the unnatural downstream context of a riboswitch could also be available, if it has been proved that there is no interference problem between the riboswitch and the context. We also chose GFP as Stabilizer of Adda riboswitch because Adda riboswitch can control GFP expression directly in the previous study. By experiments, we can verify that the above two sources of Stabilizer is reliable.
By experiments, we can verify that the source of Stabilizer is diverse.
3.2.2 Length of stabilizer
Stabilizer should be long enough to insulate the riboswitch and the downstream CDS in the most case, but short enough to minimize the overall size of the system. So, we explored the suitable length of the ‘Stabilizer’.
According to using docking matrix, we can get Stabilizer of appropriate length. Detailed methods can be referred to model and software.
In order to verify our modeling, we chose two ‘good’ Stabilizers which are able to prevent the structure from interference and two ‘bad’ Stabilizers whose length were too short to protect their structures.
The results show that our software could predict the result of wet experiments and it is useful and reliable!
3.3 GOI
To ensure that our modular riboswitch will work with a variety of different CDSs of proteins, we substituted sfgfp with eyfp. Using the new gene, we tested the effect of modular Adda riboswitch consisting the original Adda riboswitch, STA150 (Stabilizer) and Tuner A. The result verified that the target gene is indeed replaceable, the modular riboswitch secondary structure was not affected!
3.4 The design principle of modular riboswitch
Fortunately! So far, we have been able to summarize a small set of modular riboswitches with below design principles: modular riboswitch consisting of original riboswitch, well-characterized Stabilizer (wild type context preference), and a dynamic-predictable ‘Tuner’ arranged from 5' to 3' end of the mRNA.
We have applied this design principle to Adda riboswitch. The experience verified that GOI is replaceable and the modular riboswitch can regulate the expression of different GOIs. Further, we will use this design principle to create more RiboLego!
4. More RiboLego
4.1 RiboLego based on Btub
After introducing an translation-activating riboswitch, a translation-repressing riboswitch was employed to test our design principle. In absence of ligand, the repressing riboswitch can expose RBS, making GOI express. When ligand exists, GOI translation decreases. Taking it into consideration, we employed Btub riboswitch responsive to VB12 from E.coli. By our program, the first 150bp of the CDS of gene btuB which is the wild type corresponding CDS of the Btub riboswitch, was used to serve as Stabilizer. We selected Tuner A and E to design two modular Btub riboswitches and used sfgfp as the reporter gene to test whether the expressions of these two systems match our prediction.
Another concern is that, although introducing the ‘Stabilizer’ can protect the structure of the riboswitch from interference, we worried that the accumulation of the nonsense peptide translated from the ‘Stabilizer’ may lead to increased metabolic burden on cells, which may affect cell growth and the expression of desired genes. Therefore, we decided to design a Tuner S containing SsrA protein degradation tag to degrade the nonsense peptide from the ‘Stabilizer’.
The results show that the ‘Stabilizer’ and Tuner constructed on Btub riboswitch can work well.
4.2 IMPROVE: RiboLego based on cobalamin Riboswitch
In order to verify the universality of our design principle, we improved the part BBa_K1678007 by the principle we set up.
By referencing the previous iGEM project, we found that Paris_Bettencourt has created a cobalamin biosensor to measure vitamin B12. The cobalamin biosensor is based on a riboswitch taken from a transcribed fragment upstream of a cobalamin biosynthesis gene, cbiB, which is found in Propionibacterium shermanii and has been demonstrated to be sensitive to VB12. At first, they used egfp as their reporter gene, however, even in the absence of cobalamin, they had no GFP expression at all. Then they substituted egfp with mrfp1 and inserted the first 24 bases of the natural downstream context of cbiB between them, the result was still negative.
Using our software, we selected the first 81bp of the natural downstream context cbiB as the ‘Stabilizer’ and used Tuner A to control the expression of mRFP1.
The results proved that we successfully designed the modular cobalamin biosensor by our design principle!
4.3 IMPROVE: RiboLego based on Four U
Riboswitches can further be classified into thermodynamic and kinetic switches. Different from kinetic switches, thermodynamic switches can reversibly and repeatedly toggle between "on-" and "off-" states, depending on temperature. Thermodynamic switches are temperature-sensing RNA sequences in 5’UTR of their mRNAs. At low temperature, they can fold into the structure, blocking access of ribosome; at high temperature, it will open its conformation, increasing the efficiency of translation initiation.
After successfully creating modular riboswitches with our method on kinetic switches and demonstrating that the ‘Tuner’ can control the output ranges of riboswitches, we started to build modular thermodynamic riboswitch, and here we made part BBa_K115002(Four U) improvement based on TUDelft team’s work in 2008.
Four U is an RNA thermometer that can be used for temperature sensitive post-transcriptional regulation that initiates translation at 37°C. Fortunately, we found that OUC-China team had used Four U to successfully express RFP in 2015, which provided us with great convenience! So, we selected the first 132bp of the natural context of the Four U as its Stabilizer. Then we used Tuner A to regulate the gene translation. The superfolder green fluorescent protein (sfgfp) is the reporter gene to verify our modular riboswitch.
Experimental results show that we have successfully constructed a modular thermodynamic riboswitch and changed its response curve.
5. AsRNA
Thermodynamic switches are found being energetic equilibrium between their on- and off-state. If switching is triggered, the equilibrium distribution shifts towards the new energetically best conformation. This implies that thermodynamic switches can reversibly and repeatedly toggle between on- and off-states. In contrast, kinetic switches are trapped in one state, depending on whether the ligand was present at the time of folding. Because ligands are hard to be degraded, the state of the dynamic riboswitch is difficult to be changed without ligand washing. However, the ligand washing or medium replacement is almost impossible for many actual industrial application scenarios.
With the help of Professor Li Yun, we finally utilized an antisense RNA to tackle this problem. Antisense RNA is endogenous in E. coli that do not require heterologous proteins to function. Owing to its simple design principles, small size, and highly orthogonal behavior, the engineered genetic parts of asRNA has been incorporated into genetic circuits. Antisense RNA can be thought to 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. Hfq is a native chaperone protein that mediates RNA interactions by binding to a particular RNA binding site on the asRNA molecule. In our work, the engineered MicF binding site (MicF M7.4) was used as Hfq binding site because it performed well with low off-target effect in previous studies.
By using model to design artificial TBR, we hope to utilize asRNA to turn off the activated riboswitch. When targeting the RBS of Adda riboswitch, asRNA can close modular Adda riboswitch even in presence of 2-AP. At the same time, when targeting RBS of Tuner E, asRNA is able to open modular Btub riboswitch even in presence of VB12.
The results show that we've been able to regulate the on-off state of riboswitch!
6. In the future
This year, OUC-China proposed the design principle of the ‘Attenuator’ for rational design of modular riboswitch. And created genetic toolbox “RiboLego”. Besides, we introduced asRNA as an additional control Method on the riboswitch, which extend the application scenarios of riboswitch to a wider range of practical situations.
In the future, there are some works to improve our project. Firstly, we want to create more Tuners based on this design principle and models to achieve more predictable designed riboswitch for various applications. Secondly, we expect our software can help synthetic biologists and future iGEM teams to easily use riboswitches to control GOI in an anticipated way. Finally, we hope to use the cell-free system to optimize RiboLego from the perspective of engineering, and use vesicles to wrap asRNA in the in-vitro system, so that it can carry out accurate real-time regulation of riboswitch state for many times.
REFERENCE
[1] Folliard T , Mertins B , Steel H , et al. Ribo-attenuators: novel elements for reliable and modular riboswitch engineering[J]. Scientific Reports, 2017, 7(1):4599.
[2] Zhu X , Wang X , Zhang C , et al. A riboswitch sensor to determine vitamin B12 in fermented foods[J]. Food Chemistry, 2015, 175:523-528.
[3] Dixon N , Duncan J N , Geerlings T , et al. Reengineering Orthogonally Selective Riboswitches[J]. Proceedings of the National Academy of Sciences of the United States of America, 2010, 107(7):2830-2835.
[4] Berens C , Groher F , Suess B . RNA aptamers as genetic control devices: The potential of riboswitches as synthetic elements for regulating gene expression[J]. Biotechnology Journal, 2015, 10(2):246-257.
[5] Lee Y J , Moon T S . Design rules of synthetic non-coding RNAs in bacteria[J]. Methods, 2018:S1046202317303389.
[6] Hoynes-O’Connor, Allison, & Moon, T. S. . (2016). Development of design rules for reliable antisense rna behavior in\r, e. coli. ACS Synthetic Biology, acssynbio.6b00036.
[7] Tian T, Salis HM. A predictive biophysical model of translational coupling to coordinate and control protein expression in bacterial operons. Nucleic Acids Res. 2015;43(14):7137-51.
[8] Seeliger J C , Topp S , Sogi K M , et al. A Riboswitch-Based Inducible Gene Expression System for Mycobacteria[J]. PLOS ONE, 2012, 7.
[9] Dixon N , Duncan J N , Geerlings T , et al. Reengineering Orthogonally Selective Riboswitches[J]. Proceedings of the National Academy of Sciences of the United States of America, 2010, 107(7):2830-2835.
[10] Allner O , Nilsson L , Villa A . Loop-loop interaction in an adenine-sensing riboswitch: A molecular dynamics study[J]. RNA, 2013, 19(7):916-926.
[11] Lee Y J , Kim S J , Amrofell M B , et al. Establishing a multivariate model for predictable antisense RNA-mediated repression[J]. ACS Synthetic Biology, 2018.
[12] Shiori S , Jung-Eun S , Razika H , et al. Paradoxical suppression of small RNA activity at high Hfq concentrations due to random-order binding[J]. Nucleic Acids Research(17):17.
