Results

Our iterative design process will be further developed by closely integrating computational modeling with laboratory data acquisition. We will generate a series of predictive time-course graphs from our model and systematically compare and contrast these outputs with experimental results, specifically focusing on the kinetics of GFP variants due to their simplicity and well-documented characterization. This baseline will allow us to accurately model various expression rates. By iteratively incorporating new literature values and experimental data (e.g., promoter strengths, RBS efficiencies, and fluorescent protein maturation rates) back into the model, we will refine its parameters to achieve a strong relationship between graphical and numerical predictions. This process will enable us to draw conclusions on the functional trade-offs and optimal use of various promoters in genetic parts.

Figure 1: This figure presents our dynamic overlay comparison of the gene expression kinetics driven by 8 different BBa_J23XX promoters over a 5000 second time course. The graph monitors the accumulation of mRNA (solid lines, left y-axis) alongside the protein (dashed lines, right y-axis) in molecules per cell. The data demonstrates the relative strengths of each promoter and matches assumptions based off of literature regarding Anderson family promoters. J23100 exhibits the strongest expression with a high steady state level for both mRNA (~10.5 molecules/cell) and protein (~12,000 molecules/cell) in the shortest time. Whereas J23113 exhibits the weakest and slowest expression of both mRNA and protein. This data helps in developing synthetic circuits that require finely tuned gene expression to help prevent leaky expression in downstream components or metabolic burden.

Figure 2: This figure provides a detailed side-by-side analysis of the J23100 and J23102 constitutive promoters. The Anderson J23100 promoter demonstrates a strong expression, reaching an mRNA steady state of ~10.5 molecules/cell and a final protein level of ~12,000 molecules/cell. In contrast, the J23102 promoter, a weaker variant, maxes out at a lower steady state of ~9 molecules/cell for mRNA and ~10,000 molecules/cell for protein. Both promoters quickly reach steady state mRNA expression with a slower, sigmoidal accumulation of protein. These dynamics match assumptions based off of literature regarding the dynamics of Anderson family promoters.

Figure 3: This figure represents a kinetic model to predict fluorescent protein maturation under the control of the strong Anderson promoters J23100 and J23102. The model separates total protein into immature protein (recently translated, non-fluorescent) and mature protein (fully folded and fluorescent). For both promoters, the immature protein pool rapidly reaches a pseudo-steady state (~3500 molecules/cell for J23100; ~3000 for J23102), reflecting a balance between translation and maturation rates. The mature protein pool accumulates continuously, illustrating that the observed fluorescence lags behind actual transcription and translation. This simulated data follows what we predict might be observed in a wet lab experiment.

Figure 4: This figure shows how total mRNA levels and ribosome activity evolve over time when the cell’s ribosomes are limited. Initially, the throttle factor (the fraction of free ribosomes available for translation) starts high and drops rapidly — indicating that ribosomes quickly become saturated by mRNA transcripts. As mRNA levels rise, ribosomes are increasingly burdened, causing a global throttling effect where translation slows. Eventually, both the total mRNA and the throttle factor reach steady-state; total mRNA near its equilibrium level, and global throttling near a low value (~0.55–0.6), showing sustained ribosome competition. At KR=100, the system reaches a balance where transcription continues but translation is limited by ribosome availability — a realistic model for resource competition in cells.

Figure 5: This scatter plot is a characterization of the fluorescent reporter's behavior for various Anderson family promoter constructs. The y-axis displays the steady-state mature protein level (a proxy for final brightness, in molecules/cell), while the x-axis quantifies the maturation lag (in seconds), calculated as the difference between the time required to reach 50% of the maximum mature protein (t50_protein ) and the time required to reach 50% of the maximum mRNA (t50_mRNA ). The data provided from literature values will create a clear positive correlation between promoter strength (brightness) and the maturation lag time. This figure highlights the importance of using kinetic models to accurately correct protein maturation time when performing instantaneous measurements of gene expression. The simulated model will eventually be compared to our measured values.

Figure 6: Left: this panel confirms the ranking from the Maturation Lag vs. Brightness plot: CustomStrong ≫ J23100 > J23102 > J23104, and so on. The CustomWeak and even weaker promoters J23113–J23109 produce minimal brightness. Right: The bars look flat—and that can be misleading. Time to reach 80% of each promoter’s own steady state is mostly set by the system’s built-in kinetics (protein maturation and dilution by growth), not by how strong the promoter is, so strong promoters end brighter but don’t reach “80%” much faster. Don’t read the flat bars as evidence of a translation or transcription bottleneck; they reflect those shared time constants. For a speed comparison that does depend on promoter strength, use an absolute threshold (e.g., “time to detectable fluorescence”), which will vary across promoters.

Figure 7: This figure graphs the time (in seconds) required for minimum fluorescence detection (at 1562 mature copies of sfGFP) versus the number of steady-state mature protein copies (those that fluoresce) in the cell. In effect, a negative saturation trend is established between the time required for recordable fluorescence data and the number of sfGFP proteins there are in the cell. While promoters expressing high amounts of steady state proteins such as J23100 are detectable much earlier than others, they, due to high amounts of protein production and folding, burden cell machinery significantly more.

Our simulation framework was developed to explore the kinetics of quorum-sensing induction by systematically modeling the effects of AHL (acyl-homoserine lactone) dosing on gene expression. Using sfGFP and OD600 as a measurable output, we generate predictive time-course plots of inducer dynamics, fluorescence production, and bacterial growth. These simulations allow us to test both single-dose and scheduled-dose scenarios, providing a quantitative baseline for understanding EC50 values, Hill slopes, and the overall robustness of induction systems. By iteratively comparing these computational predictions with experimental data, we refine the model to account for promoter activity, RBS efficiency, and protein maturation rates. This process ensures that the simulation not only captures general induction trends but also becomes a practical design tool for evaluating circuit performance, trade-offs, and optimization strategies in quorum-sensing applications.

Figure 1: This plot shows the single-well AHL potency simulation. The initial AHL concentration was set to 600,000 µM, and a scheduled dose of 1e10 µM was added when the optical density reached 1e-11. The resulting curve demonstrates the sharp increase in AHL potency after dosing and its gradual decline over time.

Figure 2: This figure represents sfGFP fluorescence output as a function of time for the same experimental parameters. Following the initial AHL spike, GFP expression rises sharply and then plateaus at around 31 × 10^6 units. This plateau reflects the system’s steady-state response, where the gene circuit has reached its maximal GFP output given the AHL induction conditions.

Figure 3: This panel tracks simulated bacterial growth as OD600 for the same run (600,000 µM initial AHL; a 1×10¹⁰ µM dose injected when OD600 reaches 1×10⁻¹¹). The curve exhibits:

Together with Figures 1–2, this shows how AHL induction coincides with growth dynamics: dosing occurs at very low density, while sfGFP output rises as the culture enters exponential growth and then plateaus.

Figure 4: This figure shows the EC50 analysis for sfGFP as a function of the initial AHL concentration, run over 100 time steps with no additional doses added. Each point represents the final normalized GFP level for a given starting AHL value. A Hill curve is fitted to the data, shown in black, with AHL concentrations displayed on a log scale (x-axis). The y-axis is normalized to percentage of maximum GFP expression, illustrating the classic sigmoidal dose–response behavior with a sharp transition around the EC50 point.

Figure 5: This figure presents EC50 analysis where the response is measured against scheduled AHL dosing events. The dosing follows the activation rule: [(5e-10, 1e6), (1e-9, 1e6), (5e-9, 1e6), (1e-8, 1e6)], where the first value in each pair is the OD600 threshold that triggers the dose, and the second value is the dose magnitude. The resulting data points are normalized GFP outputs, fitted with a Hill curve (black line). The x-axis is logarithmic (dose trigger thresholds), and the y-axis shows the percentage of maximum GFP, emphasizing the graded activation pattern driven by scheduled dosing events.

Figure 6: This example shows AHL potency in a single well over 600 simulation steps. The trace reflects step-like increases when doses are triggered, followed by plateaus as AHL is consumed or diluted. This mode of visualization provides per-well insight: the same panel can be used to retrieve sfGFP and OD600 dynamics as alternative readouts for the exact same well.