Team:OUC-China/Results
How can we build a biological computer? Since the basic functional unit of an electronic computer is implemented through logic gates, we think that we can start by optimizing and perfecting the logic gate. We incorporated a series of logic gates which were built by toehold and 3WJ repressor, and designed the IMPLY gate and XOR gate that are still devoid. We made some improvements to these logic gates to improve their stability, versatility, composability, and designed multiple-input logic gates and swapping gates. In order to show the wide application prospect of these logic gates, we used them in the construction of adder and subtractor, as well as virus detection. We will introduce our results in the following aspects: preparations, logic gates, improvements and applications.
1. Preparations
In order to build logic gates better, we need to do some preparations. Therefore, we have conducted a series of tests on promoters, the ON RNA switch and the OFF RNA switch. As we introduced in DESIGN, we chose toehold as the ON RNA switch and three-way Junction (3WJ) repressor as the OFF RNA switch. The results of these basic tests provided references for the construction of logic gates and the design of some more complex circuits.
1.1 Promoters
It is very important for us to select the inducible promoter with low transcriptional leakage and high dynamic range. These promoters are essential parts of building the logic gates. These promoters also played a crucial role in the subsequent construction of multiple-input logic gates and swapping gates. After a series of experimental tests, we selected Tet promoter and Lux promoter with good performance to build the double-input logic gate. Affected by COVID-19, we don’t have enough time to test all promoters, so we obtained the experimental data of the pBAD from the NEFU_China.
|
|
Figure 1. The fluorescence data of promoters.
(A) The Tet promoter fluorescence data in E. coli BL21 (DE3) at different aTc concentrations (0mg/mL, 0.01mg/mL, 0.1mg/mL, 0.25mg/mL, 0.5mg/mL and 1mg/mL). The Tet promoter showed the highest fluorescence intensity when the aTc concentration was 0.25mg/mL. (B) The Lux promoter fluorescence data in E. coli BL21 (DE3) at different HSL (N-(Ketocaproyl)-L-homoserine Lactone) concentrations (0mg/mL, 0.001mg/mL, 0.002mg/mL, 0.01mg/mL, 0.1mg/mL and 1mg/mL). The Lux promoter showed the highest fluorescence intensity when the HSL concentration was 0.01mg/mL. Error bar: SD (n=9).
We confirmed the optimal concentration of inducers for Tet promoter and Lux promoter. The Tet promoter showed the highest fluorescence intensity when the aTc concentration was 0.25mg/mL, and the Lux promoter showed the highest fluorescence intensity when the HSL (N-(Ketocaproyl)-L-homoserine Lactone) concentration was 0.01mg/mL. In subsequent experiments, the optimal concentration of inducers was used to being inputs to logic gates, in addition to verify the performance of toehold and 3WJ repressor.
1.2 Toehold
Toehold switch is composed of cis-acting element RNA hairpins and trans-acting factor trigger RNA. The binding of a trigger RNA to the toehold sequence allows for a branch migration process, exposing AUG and RBS for translation initiation.
Figure 2. Validation of toehold
Compared with the blank control (IPTG=0 M, aTc=0 mg/mL), the fluorescence of GFP was low when only the promoter before toehold sequence was turned on (IPTG=0.1 M, aTc=0 mg/mL). This indicates that toehold has the advantage of low leakage. When the trigger was expressed (IPTG=0.1 M, aTc=0.25 mg/mL), it showed a high GFP fluorescence of up to 32-fold due to the destruction of the toehold hairpin structure. Error bar: SD (n=9).
Because toehold has the characteristics of low leakage and high ON/OFF ratio, we choose it as an ON switch. And we used it in the construction of logic gates and the design of some more complex circuits.
1.3 Three-way Junction (3WJ) Repressor
3WJ repressor switch RNA employs an unstable hairpin secondary structure. This unstable hairpin was previously demonstrated to be translationally active. When a complementary trigger RNA is expressed, the trigger will bind to the switch RNA, making the originally unstable 3WJ structure stable, and represses translation.
Figure 3. The inhibitory effect of 3WJ repressors
In the validation of 3WJ, as with toehold, we also set blank control (IPTG=0 M, aTc=0 mg/mL). Compared with the group without trigger expression (IPTG=0.1 M, aTc=0 mg/mL), the group with trigger expression (IPTG=0.1 M, aTc=0.25 mg/mL) showed inhibitory effect. Error bar: SD (n=9).
Figure 4. Orthogonality of 3WJ repressors
Crosstalk was determined by dividing the arithmetic mean of the GFP fluorescence from a given trigger switch pair by the arithmetic mean of the GFP fluorescence for the cognate trigger switch interaction. GFP fluorescence was measured from n=9 biologically independent samples.
3WJ repressor showed remarkable ON/OFF ratio and had good orthogonality, we choose it as OFF switch. And we used it in the construction of logic gates and the design of some more complex circuits.
2. Logic Gates
In the verification of the logic circuit, the inductors corresponding to the promoter before the switch structure were taken as the inputs (after adding aTc, trigger A will be expressed and after adding HSL, trigger B will be expressed), and the expression quantity of fluorescent protein was taken as the output, so as to test the existing logic operations and the logic operations designed by us (IMPLY gate and XOR gate).
2.1 OR Gate
OR gate composed of two switch RNA hairpins. Each switch module has an input RNA recognition site and its own RBS and start codon. Input RNA binding unwinds the corresponding switch stem to activate translation.
Figure 5. Two-input toehold OR gate
The left figure shows that when INPUT A=0, INPUT B=0, the fluorescence of GFP is low. And the fluorescence intensity of GFP was high in the other three groups. This corresponds to the situation described in the truth table on the right. INPUT A=1 means that aTc (0.25 mg/mL) is added, INPUT B=1 means that HSL (0.1 mg/mL) is added. Error bar: SD (n=9).
2.2 AND Gate
AND gate constructed from two input RNAs that bind to yield a complete trigger RNA. When either input RNA is expressed, it is incapable of activating the switch because neither trigger sub-sequence alone can unwind the repressing hairpin. The toehold switch can only be turned on when the two input RNA species hybridize and form a complete trigger sequence.
Figure 6. Two-input toehold AND gate
The left figure shows that when INPUT A=1, INPUT B=1, the fluorescence of GFP is high. And the fluorescence intensity of GFP was low in the other three groups. This corresponds to the situation described in the truth table on the right. INPUT A=1 means that aTc (0.25 mg/mL) is added, INPUT B=1 means that HSL (0.1 mg/mL) is added. Error bar: SD (n=9).
2.3 NOT Gate
NOT gate is made up of 3WJ repressor. Its unstable hairpin was previously demonstrated to be translationally active. When a complementary trigger RNA is expressed, the trigger will bind to the switch RNA, making the originally unstable 3WJ structure stable, and represses translation.
Figure 7. Two-input 3WJ repressor NOT gate
The left figure shows that when INPUT=0, the fluorescence of GFP is high. And the fluorescence intensity of GFP was low when INPUT=1. This corresponds to the situation described in the truth table on the right. INPUT=1 means that aTc (0.25 mg/mL) is added. Error bar: SD (n=9).
2.4 NIMPLY Gate
In the NIMPLY gate, a deactivating RNA (INPUT A) uses direct hybridization or strand displacement to abolish trigger RNA (input B) activity. So when only the correct trigger RNA is expressed, the switch can be turned on.
Figure 8. Two-input 3WJ repressor NIMPLY gate
The left figure shows that when INPUT A=0, INPUT B=1, the fluorescence of GFP is high. And the fluorescence intensity of GFP was low in the other three groups. This corresponds to the situation described in the truth table on the right. INPUT A=1 means that aTc (0.25 mg/mL) is added, INPUT B=1 means that HSL (0.1 mg/mL) is added. Error bar: SD (n=9).
2.5 IMPLY Gate
We combined the 3WJ switch and toehold switch to realize the IMPLY Boolean calculation. When no trigger is expressed, this logic gate just likes a 3WJ switch. When trigger A expressed, the trigger will bind to the switch RNA. The binding allows for a branch migration process, exposing AUG and RBS for translation initiation. When trigger B expressed, the trigger will bind to 3WJ switch RNA. The resulting trigger–switch complex has a stable 3WJ structure that effectively sequesters the RBS and the start codon within the loop and stem of the switch RNA, respectively, and strongly represses translation. When both trigger A and B are expressed, the binding of trigger RNA to the toehold sequence allows the ribosome to bind to the former RBS and break open 3WJ stable hairpin.
Figure 9. Two-input toehold and 3WJ repressor IMPLY gate
The left figure shows that when INPUT A=0, INPUT B=1, the fluorescence of GFP is low. And the fluorescence intensity of GFP was high in the other three groups. This corresponds to the situation described in the truth table on the right. INPUT A=1 means that aTc (0.25 mg/mL) is added, INPUT B=1 means that HSL (0.1 mg/mL) is added. Error bar: SD (n=9).
2.6 XOR Gate
The XOR is inspired by the NIMPLY gate, consisting of a toehold switch and two triggers. The trigger’s core sequence is the same and at the triggers’ both ends there are the nucleotide-binding domains. When input one of these triggers, the switch can turn on. Moreover, when input these two triggers simultaneously, they can pair together and form a ring in the middle. As a result, the switch will still be in the OFF state.
Figure 10. Two-input toehold XOR gate
The left figure shows that when INPUT A and INPUT B are both 0 or 1, the fluorescence of GFP is low. And the fluorescence intensity of GFP was high in the other two groups. This corresponds to the situation described in the truth table on the right. INPUT A=1 means that aTc (0.25 mg/mL) is added, INPUT B=1 means that HSL (0.1 mg/mL) is added. Error bar: SD (n=9).
3. Improvements
Stability, versatility and compatibility are important for biocomputers. We add the hairpin structure at the 5' end of triggers to improve the trigger stability, design the software for calculating the threshold to improve the universality, and use trans-action factors such as σ factor to improve the compatibility. We also designed more complex multiple-input logic gates and swapping gates to show more possibilities of logic gates.
3.1 Stability
The stability of RNA is a key factor affecting the mechanism of its interaction. Therefore, we hope to optimize the ON/OFF ratio of the toehold switch by improving the stability of trigger RNA. After consulting many pieces of literature, we learned that adding hairpin structure at the 5 'end of RNA could prevent the degradation mediated by RNase, thus improving the stability of RNA. Therefore, we tried to add the hairpin structure at the 5' end of triggers. We used NUPACK and RNAfold to analyze the structure of the trigger with the 5’ end hairpin, and then used NUPACK to analyze the combination of triggers with the 5’ end hairpin and switches.
Figure 11. The secondary structure
(A) The secondary structure of 5’ hairpins analyzed using NUPACK. (B) The secondary structure of 5’ hairpins combining with trigger analyzed using NUPACK.
3.2 Versatility
We find a certain difference in the threshold value of different logic gates, which is not conducive to combining different types of logic gates. To improve the generality of these logic gates, it is necessary to find a universal threshold. Therefore, we have developed a SOFTWARE that can calculate the common threshold. Threshold calculation can give the normalized parameters of fluorescence intensity and the calculated threshold, so that the designers can get the common threshold of different logic gates easily and combine different types of logic gates better.
|
|
Figure 12. Demonstration of software
Figure 13. The normalization of date
3.3 Composability
Incorporation of ribo-computing devices into sophisticated layered circuits will require systems that can provide RNAs as output species. Thus, we use σ factors to achieve this function. The σ factors have several highly orthogonal pairs, so that this functionality can be implemented.
3.4 Multiple-input
On the basis of single-input logic gates and two-input logic gates, we also designed the three-input logic gates and four-input logic gates. These enrich the types of our logic gates. See the MODEL (Multiple-input) page for details.
Figure 14. The secondary structure of the three-inputs logic gates
Figure 15. The secondary structure of the four-inputs logic gates
3.5 Swapping Gates
In the IMPLY and NOT swapping gate, we use a trigger control the transformation between two logic gates. We combined the inhibitory hairpin constructed switch with the 3WJ repressor, which achieved the transformation between IMPLY and NOT. If input the trigger A, it is IMPLY gate. If not, it is NOT gate.
|
|
Figure 16. The secondary structure of IMPLY and NOT swapping gate
In the AND and OR swapping gate, we use different triggers to control the transformation. We combined the single hairpin constructed switch and the inhibitory hairpin constructed switch to achieve the transformation between AND gate and OR gate. When input the trigger B, it is OR gate. When trigger A is absent, it behaves like an AND gate.
Figure 17. The secondary structure of AND and OR swapping gate
4. Application
As a basic logic gate element, our project can be applied to many aspects. In order to show the wide application prospect of these logic gates, we used them in the construction of half-adder and subtractor, as well as virus detection.
4.1 Arithmetic Units
We construct the half-adder and the subtractor with the AND, XOR and NIMPLY gates. The half-adder is capable of carrying out simple binary mathematical operations necessary for digital computing. The subtractor can be done by adjusting the concentration of trigger RNAs. In the future, they can be used in the more complicated calculations.
4.1.1 Half-adder
We used the AND gate and XOR gate to make up the half-adder. A half adder is a type of adder, an electronic circuit that performs the addition of numbers. The half adder is able to add two single binary digits and provide the output plus a carrying value. It has two inputs, called A and B, and two outputs S (sum) and C (carry).
The XOR is inspired by the NIMPLY gate, consisting of a toehold switch and two triggers. The trigger’s core sequence is the same and at the triggers’ both ends there are the nucleotide-binding domains. When input one of these triggers, the switch can turn on. Moreover, when input these two triggers simultaneously, they can pair together and form a ring in the middle. As a result, the switch will still be in the OFF state.
AND gate constructed from two input RNAs that bind to yield a complete trigger RNA. When either input RNA is expressed, it is incapable of activating the switch because neither trigger sub-sequence alone can unwind the repressing hairpin. The toehold switch can only be turned on when the two input RNA species hybridize and form a complete trigger sequence.
When there is only one input, the reporter of AND gate is not expressed, and the reporter of XOR gate is expressed. When there are two inputs, the reporter of AND gate is expressed, and the reporter of XOR gate is not expressed.
Figure 18. Half-adder circuit diagram
4.1.2 Subtractor
We used the NIMPLY gate to make up the subtractor. We use inducers to turn on trigger expression, use inducer concentration to represent the input trigger concentration, and use fluorescent protein expression levels to react and output results. When two triggers are expressed at the same time, trigger S1 and trigger S2 will combine to form an RNA double strand that cannot be turned on by RNA Switch. Only when trigger S2 is expressed, the downstream gene expression of RNA Switch will be turned on. This subtracter can simulate S2-S1=Ps, and Ps is the fluorescence intensity of the reporter gene.
Figure 19. Abstractions of subtractor
However, we only find one NIMPLY gate sequences in the literature. In order to achieve the subtractor that can calculate different concentration range, we designed NIMPLY upon that sequence.
The NIMPLY gate consists of a toehold switch and two input triggers that can pair together. The toehold and the triggers’ core sequences we used originated from the previous literature. Then we added the nucleotide-binding domains at both ends of the trigger’s core sequence. This sequence is named “trigger1”. The other one was designed for completely complementary pairing with trigger1 and not pairing with switch sequence. These sequences were designed using NUPACK and were further screened using RNAfold and RNAstructure.
Figure 20. The secondary structure of the pairing two triggers analyzed using NUPACK.
4.2 Virus Detection
Under the influence of COVID-19, our experimental time was more limited and the control of virus usage became stricter. Therefore, we didn’t conduct experimental verification on the application of virus detection, but only demonstrated the application of virus detection with a logic gate in a biological computer through design and modeling simulation.
5. Future
The ultimate goal of combining molecular biology and engineering is to realize the biocomputer. Since an electronic computer is so well-developed according to our research, it seemed unlikely that the biocomputer element we research into will easily surpass the calculation speed of electronic computers. Therefore, biocomputer should demonstrate excellence in another area, such as, high humidity tolerance, high biocompatibility and mass storage capacity. Biocomputer should have special functions which emphasis on biological characteristic and would be complementary with electronic computers eventually. The expertise we obtained has inspired us to consider realizing biological functions with our biocomputer designs.
Our project is not trying to solve a real world problem with practical applications right away, we build elements of a biological computer. Logic gates are the basis of computing. To build a biocomputer, we need to start from the basic parts built by biological elements. We believe that these logic gates have great potential in resistance gene detection, virus detection, and disease detection.
References
[1] Kim, J., Zhou, Y., Carlson, P. D., Teichmann, M., Chaudhary, S., Simmel,F. C., … Green, A. A. (2019). De novo-designed translation-repressing riboregulators for multi-input cellular logic. Nature Chemical Biology. doi:10.1038/s41589-019-0388-1
[2] Green, A. A., Kim, J., Ma, D., Silver, P. A., Collins, J. J., & Yin, P. (2017). Complex cellular logic computation using ribocomputing devices. Nature, 548(7665), 117–121. doi:10.1038/nature23271
[3] Green, A. A., Silver, P. A., Collins, J. J., and Yin, P. (2014) toehold switches: de-novo-designed regulators of gene expression. Cell 159, 925– 939, DOI: 10.1016/j.cell.2014.10.002
[4] Meyer, S., Chappell, J., Sankar, S., Chew, R. and Lucks, J.B. (2016), Improving fold activaton of small transcription activating RNAs (STARs) with rational RNA engineering strategies. Biotechnol. Bioeng., 113: 216-225. doi:10.1002/bit.25693
[5] Song, T., Garg, S., Mokhtar, R., Bui, H., & Reif, J. (2016). Analog Computation by DNA Stran Displacement Circuits. ACS Synthetic Biology, 5(8), 898–912. doi:10.1021/acssynbio.6b00144
[6] Moon, T. S., Lou, C., Tamsir, A., Stanton, B. C., & Voigt, C. A. (2012). Genetic programs constructed from layered logic gates in single cells. Nature, 491(7423), 249–253. doi:10.1038/nature11516
[7] Jongmin Kim, Elisa Franco, RNA nanotechnology in synthetic biology, Current Opinion in Biotechnology, Volume 63, 2020, Pages 135-141, ISSN 0958-1669, doi: 10.1016/j.copbio.2019.12.016.
[8] Carrier, T.A. and Keasling, J.D. (1999), Library of Synthetic 5′ Secondary Structures To Manipulate mRNA Stability in Escherichia coli. Biotechnol Progress, 15: 58-64. doi:10.1021/bp9801143