1. Overview

Figure 1: The Overview Genetic Circuit

We designed a genetic circuit capable of switching between different types of logic gates based on user requirements, essentially creating a programmable logic device (PLD) using genetic circuits in organisms. By utilizing inducible promoters, CRISPRi, and recombinases, we can modify gene pathways according to specific targets, ultimately implementing programmable logic gates.

Our circuit consists of four components:

1. The Tangram section contains different recombinases attached to different inducible promoters.
2. The Register section is where recombinases perform their functions, and where the desired logic gate is generated.
3. The Output section collects information from the Register circuits and delivers the final output of the logic gate.
4. The Patch section is designed to avoid the steric hindrance effect of the dCas9 protein.

2. Logic gates

2.1 16 ⊆ 5*5

Consider a Boolean function of 2 variables, use A and B as input variables, and Z as the output variable. The functions F(0)/F(1)/F(2)/F(3) in Z column can take values of 0 or 1. So the total Boolean function has 2×2×2×2=16 possibilities.

There are two ways to present the 16 possible 2-input 1-output logic gates: logic functions and Boolean algebra. For logic functions, each set of 4 numbers in the Logic functions column below represents F(0), F(1), F(2), and F(3). For example, 1001 represents F(0)=1, F(1)=0, F(2)=0, F(3)=1. For Boolean algebra, A and B represent the inputs, while A' and B' represent the complement of A and B.

Thus, the function has 2×2×2×2=16 possibilities as listed above. However, it is too complex to express the 16 possibilities in a single cell with one gene circuit. So we split these 16 logic gates into two genetic pathways. The first pathway expresses one of these 5 elements: A, A+B, A+B', A'+B, or A'+B'; the second pathway expresses one of these 5 elements: A', A+B, A+B', A'+B, or A'+B'. Taking the intersection of any two elements from these two sets yields all 15 possibilities except 1111.

2.2 Implementation of A and non-A

Next, how to implement A & non-A in gene circuits? We decide to use inducible promoters and CRISPRi. As mentioned in the Description, CRISPRi is an excellent NOT gate for the implementation of A and non-A required in our design.

If taking IPTG as input A, and GFP as output Z, a single inducible promoter equals an A logic gate. Then we design a sgRNA that binds to a constant promoter and interferes with the binding of RNA polymerase to the promoter. The sgRNA is attached to an IPTG-inducible promoter, thus achieving a non-A.

Figure 2: Implementation of A

Figure 3: Implementation of non-A

3. Circuit Design

As stated in the Overview, our genetic circuit consists of 4 components.

3.1 Tangram

Tangram is a dissection puzzle consisting of seven flat polygons that can be put together to form various shapes. We found this concept highly analogous to our design, where recombinases play the role of altering the shape of the genetic circuits, thus enabling the execution of different logic gates.

The Tangram section contains four recombinases: tp901(Guiziou et al., 2019), bxbI(Neil et al., 2019), a118(Roquet et al., 2016), phiC31(Guiziou et al., 2019), each preceded by an inducible promoter. When the corresponding inducer is present, the respective recombinase is expressed. For example, the addition of the inducer Ara leads to the expression of the recombinase tp901.

Figure 4: Tangram Section

3.2 Register A & B

To better illustrate the function of our program, here we use IPTG to denote input1, rha for input2, and GFP to represent output Z. Different inputs and output may be alternated according to specific needs.

As previously stated, we can utilize two elements to create 16 possible logic gates. The first element can be one of the following: A, A+B, A+B', A'+B, or A'+B'; the second element can be: A', A+B, A+B', A'+B, or A'+B'. Our two registers, along with a patch system, enable us to achieve all possible logic gates.

For Register A, the output will be A'+B if no inducers are added. Adding Ara (blue) results in A+B', and adding ATc (red) yields A'. Adding both inducers is also an option, but the order of addition is crucial. If Ara is added first followed by ATc, the output will be A+B. Conversely, if ATc is added before Ara, the output will be A.

Figure 5: How Register A transforms into different logic gates.
Ara is the inducer of recombinase tp901, while ATc is the inducer of recombinase bxbI. Different sizes of triangles represent different pairs of recombinase-recognition sites. Once the inducers are added, the corresponding recombinases will bind to its specific recognition cites, catalyze inversion (when the sites are anti-aligned) or excision (when the sites are aligned).

Register B operates similarly to Register A but uses different inducers and recombinases. While Ara is replaced with DAPG (yellow) and ATc with xylose (purple), the principle of inducing different outputs by sequentially adding inducer remains the same.

In summary, we can use Register A and Register B to produce the majority of the desired logic gates, except A'+B', which is not feasible due to the steric hindrance caused by CRISPRi. To overcome this limitation, we have engineered two separate patches that are designed to express B' exclusively.

Figure 6: How Register B transforms into different logic gates.
DAPG is the inducer of recombinase a118, while Xylose is the inducer of recombinase φC31. Different sizes of triangles represent different pairs of recombinase-recognition sites. Once the inducers are added, the corresponding recombinases will bind to its specific recognition cites, catalyze inversion (when the sites are anti-aligned) or excision (when the sites are aligned).

By combining the CRISPRi system, the recombinase expression system (Tangram section), and the designed sequences above (Register A & B), we can achieve the transformation of logic gates by adding recombinases in different orders (Figure 7-8).

Figure 7: Register A

Figure 8: Register B

3.3 Output

The Output GFP is linked to a lux promoter, which turns on only when AHL & LuxR both exist, utilized by the 2023 UCAS-China team. By using the Lux promoter, we can achieve the intersection of the Register A and Register B pathways(Figure 9).

Figure 9: Output

3.4 Patch

In the design-build-learn-test cycle, we found that in the original pathway, if we inhibit the downstream promoter using CRISPRi, the normal expression of the gene of interest (GOI) cannot be achieved due to the steric hindrance effect of the dCas9 protein, even when the upstream promoter is actively initiating transcription. Therefore, we additionally designed a patch system (Figure 10).

Figure 10: A patch system using Cre/loxP system and fimE inversion system

In the original pathway, to achieve the output of A' + B', the dCas9 protein at B' inhibits the normal expression of A'. Therefore, we separately express B' in the Patch system to realize A' + B'.

We introduced a cumate-inducible cymR promoter to express the Cre recombinase in the pathway; upon adding cumate, the expressed Cre recombinase recognizes the loxP sequence, allowing the LuxI gene sequence to invert and be expressed by the constitutive promoter. This constitutes the Patch portion of Register A. Similarly, in the Register B pathway, we incorporated the PobR promoter used by the 2023 UCAS-China team and the fimE inversion system from E. coli (Ham et al., 2006) to complement the expression of the LuxR gene.

4. Directed evolution

During the experiments, we found that some recombinases exhibited relatively low recombination efficiency. Therefore, we aim to obtain more efficient recombinases through directed evolution.

4.1 Traditional Mutagenesis

Traditional directed evolution generates a random mutant library by mutating target genes through error-prone PCR (epPCR). We designed an expression module (Figure 11) and a selection module (Figure 12). The expression module induces recombinant enzyme expression with IPTG, while the selection module uses a phage lysis gene φX174 E (Din et al., 2016). We used epPCR to amplify the expression module in order to generate a random mutant library for the recombinases.

Figure 11: Expression module of traditional mutagenesis

Figure 12: Selection module of traditional mutagenesis

The mutation and selection modules were then introduced into E. coli, and the screening plan was designed as follows:

Initial screening: The growth of E. coli on the medium indicates whether the φX174 E gene has been successfully inverted. If the recombinases mutant retains biological function, the φX174 E gene will not express, allowing for colony growth on the plate. Conversely, if the mutant loses biological function, the φX174 E gene will express, lysing E. coli, resulting in no observable colonies.

Secondary screening: The brightness of GFP was measured using a microplate reader. The higher the recombinase's recombination efficiency, the greater the proportion of RNA transcribed from the inverted gfp gene to the total RNA, visually represented by increased unit brightness of GFP (assuming transcription levels before and after gene inversion remain constant).

4.2 Epi-hypermutation Architectures of continuous directed evolution

Due to the autonomous reproduction of bacteria, we can construct a persistent high-mutation system within bacteria, allowing for continuous mutations during DNA replication.

MutaT7, initially a fusion of cytosine deaminase (AID) and T7 RNA polymerase (T7RNAP), enables mutations from dC→dT and deoxyguanosine (dG) to deoxyadenosine (dA) (caused by dC→dT mutations on the complementary strand). On this basis, when combined with adenosine deaminase, A→G and T→C mutations can be achieved (A:T→G:C). The enzyme, enhanced for mutagenic activity, is called eMutaT7(Park et al., 2021). In the expression section, the gene for the recombinase is placed between the T7 promoter and T7 terminator, allowing transcription by T7 polymerase while utilizing AID for mutations.

Recombination direction factors (RDF) can control the activity of recombinases to promote specific recombination reaction directions. Research indicates that the Gp44 protein(Mandali et al.,2017), as a recombination direction factor, can enhance the excision or inversion reaction of A118 between attL and attR sites.

Figure 13: Continuous directed evolution system

Thus, we designed a continuous directed evolution system (Figure 13).

• pHyo094 (eMutaT7): The eMutaT7 enzyme is expressed under the control of the PBAD promoter, which is induced by arabinose (Ara).
• Target plasmid: We designed a bistable switch system focused on regulating the expression of the A118 recombinase and the Gp44 protein. The "Turn over" switch is controlled by the lactose operon, while the "Turn back" switch is regulated by the tetR protein.
• Selection plasmid: A constitutive promoter, J23119, is used to express the inverted gfp gene.

We transferred the above three modules into E. coli. First, we added IPTG and Ara to activate the "Turnover" switch, inducing the expression of eMutaT7 and mutating the a118 recombinase sequence. If the expressed A118 mutant retains its biological function as a recombinase, the gfp gene will be flipped and normally expressed. After sampling the strain, we removed IPTG and Ara and added ATc to activate the "Turn back" switch, relieving tetR repression on the PLtetRO promoter, which allows Gp44 protein expression to flip the gfp gene back to its original state. After multiple cycles, we can screen for mutant recombinases.

4.3 Future plan: Focused Mutagenesis

Traditional mutagenesis methods are often limited to small-scale random variations, making significant functional improvements challenging.

Since the structures of certain recombinases, including A118, and their interactions with the recombinase binding site remain unresolved, we took A118 as an example to predict the composite structure of A118 and its attP site using AlphaFold 3. The predicted structure is shown below (Figure 14-15).

Figure 14: Predicted structure of A118 and attP

Figure 15: molecular animations of the predicted structure of A118 and attP

Through our analysis of the A118 recombinase and the attP site structure, we discovered that the A118 recombinase contains two domains rich in positively charged amino acids located in the middle and at the C-terminus, which interact with DNA. Based on literature regarding the recombinase mechanism, the C-terminal domain serves as the DNA-binding domain, specifically recognizing attP. Additionally, the positively charged alpha helix in the middle region likely exhibits cutting activity, facilitating the cutting, converting, and reconnecting of DNA through dimer formation. Consequently, we aimed to conduct directed evolution of this region to enhance A118's activity and select key amino acid residues for NNK mutagenesis (N=A+C+G+T; K=G+T).

We are going to continuously using expression module (Figure 11) and selection module (Figure 12), and incorporating mixed NNK mutation sequence primers into the PCR system to generate a library of various mutants. The mutation and selection modules are then introduced into E. coli, and the screening plan are as shown in Traditional Mutagenesis.

5. Experiment verification

5.1 Orthogonality matrix

Figure 16: An orthogonality verification system

To validate that the six recombinases selected for our project are mutually orthogonal, we designed an orthogonality verification experiment(Figure 16). We created two plasmids: one plasmid expressing a specific recombinase under the T7 promoter and the other plasmid expressing the inverted gfp using the constitutive promoter J23119. Both the ribosome binding site (RBS) and the gfp have recognition sites for a specific recombinase. We varied the recombinase genes and recognition site sequences in the two plasmids and co-transformed them into E. coli, then we have two methods to assess orthogonality. On the DNA level, we can use colony PCR to check whether the segments have been inverted. On the protein level, we can measure the intensity of green fluorescence.

5.2 Length between the promoter and the start codon

The optimal distance between a promoter and the start codon in a gene construct can vary depending on several factors, including the specific promoter used, the context of the gene expression system, and so on. As the distances between promoters and the start codon differ in our Register system, we hope to explore the relationship between the distances and transcription rates.

We want to obtain some experimental data through experiments, construct a model with the modeling team members, and finally use the summarized data to optimize the design of the Register system.

6. References

1. Din, M., Danino, T., Prindle, A. et al. Synchronized cycles of bacterial lysis for in vivo delivery. Nature 536, 81–85 (2016).
2. Park, H., & Kim, S. (2021). Gene-specific mutagenesis enables rapid continuous evolution of enzymes in vivo. Nucleic acids research, 49(6), e32.
3. Mandali, S., Gupta, K., Dawson, A. R., Van Duyne, G. D., & Johnson, R. C. (2017). Control of Recombination Directionality by the Listeria Phage A118 Protein Gp44 and the Coiled-Coil Motif of Its Serine recombinase. Journal of bacteriology, 199(11), e00019-17.
4. Ham, T.S., Lee, S.K., Keasling, J.D. and Arkin, A.P. (2006), A tightly regulated inducible expression system utilizing the fim inversion recombination switch. Biotechnol. Bioeng., 94: 1-4.
5. Guiziou, S., Maranas, C.J., Chu, J.C. et al. (2023). An recombinase toolbox to record gene-expression during plant development. Nat Commun 14, 1844 .
6. Roquet, N., Soleimany, A. P., Ferris, A. C., Aaronson, S., & Lu, T. K. (2016). Synthetic recombinase-based state machines in living cells. Science (New York, N.Y.), 353(6297), aad8559.
7. Guiziou, S., Mayonove, P., & Bonnet, J. (2019). Hierarchical composition of reliable recombinase logic devices. Nature communications, 10(1), 456.