Results

As the foundation of our project, we constructed a variety of AHL-induced biosensors. A series of experiments was conducted to test diffusion response curves and to identify the orthogonal groups between AHL-biosensors pairs.

To validate the feasibility of our diffusion and serial experiments, we used the plasmids LasR-deGFP and LuxR-deGFP from 2024 WHU_China with E. coli BL21(DE3) strain, constructing high-pass engineered bacteria with fluorescence induced by 3-oxo-C12 HSL (LasR-deGFP strain) and 3-oxo-C6 HSL (LuxR-deGFP strain).

After confirming the sequences of the engineered bacteria, we established the solid diffusion response curves for the LasR-GFP and LuxR-GFP reporter bacteria using agar plate colony experiments (Figure 1A) and fluorescence microscopy (Figure 1B). The fluorescence intensity decreases with the increasing diffusion distance. At higher AHL concentrations, more colonies exhibit detectable fluorescence (Figure 1C).

In the agar plate colony experiments, the fluorescence intensity of LasR-GFP reporter decreases significantly with the increase of distance (Figure 1D, Figure 2). Biosensor of 3-oxo-C6 HSL does not show a good decline curve, presumably because proximal high AHL concentrations perturbed engineered-bacterium metabolism or colony growth, whereas distal fluorescence arose from leakage expression.

High-pass Engineered Bacteria

Based on the preliminary experiments, we constructed the biosensor bacteria to detect C4-HSL, 3-oxo-C8 HSL, 3-OH-C14:1 HSL molecules, and further established solid diffusion response curves for these strains, including 3-oxo-C12 HSL biosensor bacteria for extending our selection of orthogonal AHL systems. (Figure 3)

Although the culture time increased, the maximum fluorescence intensity remained unchanged. As time went on, the colonies located at greater input distance showed fluorescence, indicating the diffusion of AHL in solid agar plate.

Band-pass Engineered Bacteria

To implement the half adder, we engineered a band-pass strain expressing sfGFP under IPTG induction, since the IPTG induced strain is more likely to exhibit band-pass characteristics. Following induction with specific concentrations of IPTG and AHL, only the colonies with medium diffusion distance showed high fluorescence intensity, indicating that band-pass engineered bacteria were successfully constructed.

Orthogonality

We evaluated the orthogonality of the five AHL signaling systems by quantifying their crosstalk indices. The crosstalk index was defined as the ratio of the activation strength caused by a non-cognate molecule to that induced by the cognate molecule. As shown in the heatmap (Figure 5), most systems exhibited limited responses to non-cognate signals (off-diagonal elements, low values). However, certain degrees of crosstalk were observed between specific pairs, indicating partial overlap in recognition.

Our biocomputer design requires three orthogonal AHL systems. Based on the experimental results, we identified two candidate triplets: (1) 3-oxo-C6/LuxR, C4/RhlR, and C14:1/CinR; (2) C4/RhlR, 3-oxo-C8/TraR, and C14:1/CinR. Because LuxR-based biosensors suffer from strong crosstalk^[1], we rejected the first set and adopted the second, whose orthogonality had been validated previously in the study.

Considering the widespread use and experimental convenience of 3-oxo-C12/LasR in synthetic biology, we also included it when needed.

Our project design includes an amplifier, which synthesizes AHL after receiving the inducer to extend the diffusion distance. Through agar plate colony experiments, the engineered bacteria as amplifiers have the ability to synthesize AHL molecules, but there is leakage expression. (Figure 6)

Connector Bacteria

In the operation of our biological calculator, the interaction of different QS systems and their sensors is involved. In order to verify the feasibility of series connection, we constructed RhlR-LasI and LasR-RhlI engineered bacteria, and measured the fluorescence intensity of the corresponding downstream biosensors.

When the upstream connector received the inducer, the fluorescence intensity of the biosensor was significantly higher than that without inducer. However, from Figure 7B we can see that there is a leakage expression in the connector within 1-2 cm input distance.

After constructing high-pass and band-pass engineered strain, we explored suitable conditions for them to be used as logic gates. By changing the concentration, diffusion distance and diffusion time of the inducer, we constructed AND gates, OR gates, XOR gates and other logic gates successfully.

High-pass Engineered Bacteria

We applied our high-pass bacterial cultures onto the agar plates following the droplet guidance map shown below for four different biosensors (Figure 9A). To simulate logic gate inputs, AHL droplets were deposited on the left (01), right (10) or both sides (11) of the bacterial droplets.. In this way, a single plate can generate multiple logic gate setups with varying input distances, allowing us to examine their corresponding outputs. Figure 8 shows the ideal fluorescence intensity of colonies programmed to perform AND, OR and XOR logic gates.

As shown below, The 3D plots effectively help us visualize the distribution patterns of AND and OR gates across different systems, providing a basis for determining the formation range of each gate.

Screening revealed reproducible AND and OR gate responses, albeit with variable signal strengths. The following figures highlight the representative conditions under which we observed the relatively pronounced AND and OR gate behaviors in each system.

RhlR TraR CinR LasR

From the results, we observed that OR gates generally produced stronger outputs with clearer signals, whereas AND gates often showed slightly weaker responses, with partial activation at the 01 and 10 input states.

To address this limitation, we plan to introduce a T7 RNA polymerase under the control of AHL-responsive promoters to drive EGFP expression. This design is expected to sharpen the response curve and enhance the overall signal strength.

Band-pass Engineered Bacteria

The band-pass logic gate XOR was inoculated following the same schematic protocol as the high-pass logic gate.

During the construction of the band-pass engineered strain, we initially chose IPTG as the inducer, based on references ^[1].

The band-pass response was more evident to the naked eye than in quantitative measurements. Due to time constraints, we did not perform high-throughput screening, but we speculate that an IPTG induction concentration between 0.1-5 M could potentially achieve an even stronger band-pass response.

Nevertheless, these results demonstrate the feasibility of forming a band-pass circuit and provide strong support for its future application in the AHL system.

Next, we attempted to partially replace the IPTG circuit with the Las circuit and Rhl circuit.

Despite time constraints, we have not yet identified a combinatorial scheme capable of implementing an XOR logic gate. However, we have already obtained some results as evidence that these systems can achieve XOR effect (Figure 4B, Figure 4C).

Carry Input Simulation

We experimentally simulated a critical step in the computational process: the calculation and handling of carry inputs.

Taking the “110” input in the first panel of Figure B as an example, we added 1 mM C4-HSL (as indicated in the figure legend) to the two triangles at the top of the spotting pattern to activate the RhlR-LasI connector located at the square below. The connector then produced C12-oxo-HSL, which diffused to the LasR-sfGFP engineered bacteria located 0.75 cm away from the center (as indicated in the figure legend), representing the first “1” of the 110 input. Simultaneously, 1 mM C12-oxo-HSL (as indicated in the figure legend) was added to one of the two triangles located 2 cm below the center, representing the second “1” of the 110 input. In this way, the fluorescence intensity of the central LasR-sfGFP colony reflects its response to this input pattern.

The schematic below illustrates our design (Figure 12A). Four parameters were varied in the experiment: (i) C4-HSL concentration activating the RhlR-LasI carry connector (square symbol in Figure 12A), (ii) the spacing between connector Ci-1 and the current bit, and (iii-iv) the concentrations and positions of primary inputs Ai and Bi.

Due to experimental constraints, the current bit's unit was implemented using only the high-pass engineered strain LasR-EGFP (central circle). By measuring its fluorescence intensity, we were able to determine its response to the three inputs and thus infer the computed output.

The input combination 000 indicates that the connector is present on the plate but not activated with AHL. 100 corresponds to activating only the connector, while 001 indicates only one of the primary inputs (Ai or Bi) is present. 011 means both A and B are present, and 110 corresponds to the connector plus one of the primary inputs (Ai or Bi).

Each output was named using hexadecimal notation, which can be converted to an eight-bit binary format representing the corresponding input combination.

To reduce the workload of initial screening, the inputs A and B were considered equivalent, and the combinations 010 and 101 were omitted from both the experiments and the truth table.

The experimental results demonstrate that the current bit's computation can produce outputs corresponding to multiple logic gates depending on the input conditions. However, the carry input (Ci-1) does not yet exert the same weight as the primary inputs A and B, preventing the output from perfectly matching the expected computational behavior.

We plan to further optimize factors such as distance, timing, and inducer concentration, aiming to balance the influence of the previous carry input and the two primary inputs. Achieving approximately equal weighting would allow all three inputs to contribute equally to the current bit's computation.

AHL Degradation Enzyme

In this part, we used the constructed biosensors to quantify AHL depletion. Microplate assays confirmed that AiiA degrades all tested AHL variants efficiently. AiiA expression was induced in the degrading bacteria, which were then incubated with AHL in 2YT medium After incubation, the degrading bacteria were removed by centrifugation. Then the detection bacteria and fresh 2YT medium were added to detect the remaining AHL concentration in the 96-well plate.

For detailed methods, please refer to the Engineering and Measurement sections. AiiA's broad substrate spectrum and high degradation efficiency met the requirements of our project. Subsequently, we purified AiiA using a GST-affinity tag and preliminarily validated its activity on solid media (for details, see Supplementary Materials), facilitating further experiments.

In order to further understand AiiA's structural characteristics, we performed molecular dynamics (MD) simulations. The computational methods and parameters are described in the Engineering section and Supplementary Materials.

To identify flexible regions, we employed Root Mean Square Fluctuation (RMSF) analysis. However, not all flexible regions are suitable candidates, as active sites can also exhibit flexibility. Therefore, we examined the AiiA protein structure and its ligand-binding sites to exclude potential functional regions, such as residues 66, 135, and 168.

The SPELL algorithm provided Split Energy and Solvent Accessible Area (SAA) data for each residue. This information helps us to understand the self-assembly activity and hydrophilicity of the split site. Molecular dynamics (MD) simulations were used to assess the structural stability of the split fusion proteins and to provide supplementary evidence for the SPELL results. For instance, the region spanning residues 150-200 was identified as a critical hydrophobic core that contributes to the compactness and stability of the AiiA structure.

Based on these in silico findings, we initially selected three split sites for subsequent experimental validation.

Following the identification of these candidate sites, we constructed the corresponding plasmids and proceeded with functional assays. Subsequently, the degradation efficiency experiment was carried out using the same experimental method as mentioned above.

Initial observations using the split mCherry system indicated that the split AiiA constructs retained partial degradation activity at both 37°C and 25°C. We attribute this activity to the self-assembly of the protein fragments. However, no enhancement in degradation efficiency was observed upon blue light illumination.

As a positive control, we conducted parallel experiments under identical light conditions using BL21(DE3) strain containing the pGEX-AiiA plasmid (This strain expresses the full length AiiA without splitting) . The positive control demonstrated effective degradation, confirming that the blue light illumination itself does not cause significant damage to AiiA activity within the experiment.

The three different split sites yielded similar degradation efficiencies. We hypothesize that this result is due to their similar Split Energy values, which likely led to comparable self-assembly propensities and, consequently, similar functional outputs. This consistency confirms the effectiveness of our initial in silico screening methodology for selecting split sites.

We are currently investigating the underlying reasons for the lack of light-induced enhancement in degradation efficiency.

Protein Degradation Tag

For the light-inducible degradation module of result display proteins, we chose the modular degradation tag LOVdeg and fused it with mCherry. We used solid plate colony experiments to test its efficiency (Figure 12). Ten hours of blue-light exposure reduced mCherry-LOVdeg fluorescence by > 50 %. This indicates that our LOVdeg tag can effectively accelerate the degradation rate of target protein under blue light.