In our system, the brightness and darkness of the bacterial colony represents the binary states (1
and 0), analogous to a traditional computer. Consequently, fluorescence intensity, as our primary
output, must be accurately quantified.
In the part of calculation on the solid medium, colonies on the plate receive a certain concentration
of AHL and express sfGFP in response. We used a microscope to capture fluorescence images under
488nm excitation light. Then we used ImageJ to define the boundary of the each round colony and to
quantify the fluorescence based on the gray values of the green channel. The results were then used
to plot the AHL dose-response curve or classified the output as 1 or 0 by comparison with the
control group.
Figure 1. The spotting pattern. The center of
the plate is the place where AHL solution is added. The responses of the surrounding
spiral-distributed single colonies decrease with increasing distance from the center.
In the part of fluorescence detection in liquid culture, we prepared our samples and placed them into
96-well plates. Then, we used a microplate reader to acquire the fluorescence/OD600-time curve by
reading and shaking at fixed intervals. In this case, the examples include plotting the AHL dose
response curve under liquid environment and the degradation curve (characterized by the decreasing
fluorescence) under the treatment of light-controlled AHL degadation enzymes.
Figure 2. Schematic of the light-controlled
degradation enzyme experiment. AHL, reporter bacteria, and the light-controlled degradation
enzyme were added into a single well of a 96-well plate. The mixture was incubated under light
or dark conditions for a defined period, and fluorescence intensity was measured using a plate
reader for comparison.
Part I: Fluorescence Detection on the Solid Medium
There are four major steps: First, inoculating the bacterial solution to specific points on the
plate, according to the experimental design and corresponding pattern (Fig. 1). Second, we used
a
microscope to image each colony at a suitable time. Third, the photo groups were processed with
ImageJ to obtain quantitative data. Finally, the data consistent with the calculator's function
would be further processed and assembled to form the final results.
Step I
Experimental Setup and Inoculation Standardization
Detailed patterns for each inoculation are provided on our Results Page.
To control the irrelevant variables, we used strict and standardized operation protocols.
For each experiment, 12 mL of LB medium was poured into the plate to ensure the uniform
transparency
during imaging.
Because plates stored in the refrigerator were always humid, surface moisture will interfere with
inoculation. We dried every plate in a laboratory oven at 55°C for 2 min 30 s to ensure proper
humidity for inoculation and diffusion.
A consistent volume of 0.3 μL of bacterial solution was inoculated at each spot. Extensive
testing
confirmed that this volume minimized variation in colony size and prevented colonies from
outgrowing
the imaging field.
Finally, 1.0μl AHL was added as required. This volume elicits an optimal response while
minimizing
waste.
Step II
Fluorescence Imaging and Data Acquisition
For imaging, we used the 4× objective of an inverted fluorescence microscope(ICX41). We conducted
a
black balance before placing the colony in the center of the field of view. Exposure time, gain,
and
excitation light intensity were kept constant throughout imaging. All these measures guarantee
that
our data are consistent, informative and suitable for analysis.
Figure 3. Imaging platform used in this
study.
Agar plates were placed on the microscope stage, and each colony was imaged sequentially.
Step III
Image Processing and Colony Quantification
Being limited by the objective size we have, the colonies may partially extend beyond the image
border. However, since we used ImageJ to quantify the average gray value within a defined
rigion(the
colony area), the quantified data remain precise and reliable.
Figure 4. Overview of acquired images.
Colonies
from the same agar plate were stored in a single folder with numeric labels for batch
processing. Image "0" represents the black balance reference.
Because each day's experiment may contain hundreds of images. (e.g., tracking the response curves
of
six concentrations and the blank group at five time points will result in 5 * 7 * 13 = 455
photos.)
Our team member developed a batch program to automatically process high-throughput images based
on
ImageJ’s macrocode.
This program uses the built-in threshold algorithm of ImageJ to determine the colony rigions.
After
systematic evaluation, we characterized the performance of each algorithm and identified the
most
reliable one.
Method
Reliability
Description
Otsu
Very High
Classic and stable, suitable for images with clear foreground/background separation
(bi-modal histograms).
Triangle
High
Particularly suitable when the target is bright and the background is dark, with the
background forming the main peak (very suitable for image of colony ).
Mean
Medium
Based on the overall image gray-level mean, sometimes insensitive to extremely
bright/dark images.
MaxEntropy
Medium
Sometimes works well for fluorescence images, but less stable than Otsu.
Table 1. Our experience with different
processing
algorithms, which provided valuable guidance for optimizing the batch-processing code.
< class="section-content">
This program rapidly and accurately determine colony regions and conduct quantification, thereby
obtaining satisfactory raw data.
Figure 5. Effect of the batch-processing code.
The
algorithm efficiently and accurately delineates colony boundaries (highlighted in red). For
demonstration, we removed the command setBatchMode(true); to display each processing step.
Adding
this command at the beginning suppresses window pop-ups and further improves processing speed.
To contribute to synthetic biology and facilitate the use of our LOGIC toolkit by future users, we
provide the ImageJ macro code for batch colony image analysis. The macro processes numerically
labeled
images within each subfolder under a primary directory and outputs the results as .csv files
(considering that experimental data are often organized in multiple folder levels).
Notably, by simply selecting the desired channel, the code can process not only GFP but also mCherry
and
BFP images.
rootDir = "C:/***/***/***/***/";
setBatchMode(true);
// Get all subfolders
list = getFileList(rootDir);
// Clear the result table
run("Clear Results");
// Traverse each subfolder
for (i = 0; i < list.length; i++) {
if (File.isDirectory(rootDir + list[i])) {
dir = rootDir + list[i];
print("Processing folder: " + dir);
// The images in each folder range from 0.png to x.png
for (n = 0; n <= x; n++) {
filename ="" + n + ".png";
fullpath = dir + "/" + filename;
if (File.exists(fullpath)) {
open(fullpath);
run("Split Channels");
filenameBlue = filename + " (blue)";
filenameRed = filename + " (red)";
filenameGreen = filename + " (green)";
// Close the blue and red channels
if (isOpen(filenameBlue)) {
selectImage(filenameBlue);
close();
}
if (isOpen(filenameRed)) {
selectImage(filenameRed);
close();
}
// It will be processed only if the green channel exists
if (isOpen(filenameGreen)) {
selectImage(filenameGreen);
// Threshold + measurement
setAutoThreshold("Otsu dark");
run("Measure");
// Add folder and image information
row = nResults - 1;
setResult("Folder", row, list[i]);
setResult("Image", row, filename);
close();
} else {
print("Warning: Green channel not found for " + filename);
}
} else {
print("Warning: File not found - " + fullpath);
}
}
}
}
// Save the result to CSV
saveAs("Results", rootDir + "AllResults.csv");
print("Batch processing has been completed and the results have been saved to " + rootDir + "AllResults.csv");
Step IV
Data Analysis and Visualization
The raw data were then selected and assembled to produce clear visualizations. In this process, Data
were
plotted and styled using GraphPad Prism 9 and Adobe Illustrator. For example, the response curves of
different concentrations are plotted together to better illustrate the effects of AHL concentration.
Figure 6. Example dose response curve to C4-HSL.
The
data were processed using GraphPad Prism 9.
Part II: Fluorescence Detection on the Liquid Culture
This part is also divided into four steps:
Step I
Overnight Induced Expression of the Degradation Enzyme.
Our degradation enzyme is expressed in E. coli BL21(DE3) and functions within the
intracellular environment. Therefore, prior to the experiment, we need to induce the engineered
bacteria to express the degradation enzyme. To minimize the formation of inclusion bodies, we
allowed the engineered bacteria to express the enzyme at 25°C for approximately 14 hours. The
specific procedure is as follows:
1. Inoculate the engineered bacterial stock stored at 4°C (e.g., BL21(DE3) strain harboring
the
plasmid pGEX-AiiA) into SB liquid medium at a 1:10 ratio (SB medium: Trptone 32g; Yeast
extract
20g; NaCl 5g; pH~7/L).
2. Incubate in a shaker (37°C, 200 rpm) until the OD600 reaches 0.6–0.8. Then, add IPTG
(final
concentration 0.5 mM) and transfer the culture to the shaker at 25°C, 200 rpm for
approximately
14 hours of expression.
Step II
Use of Engineered Bacteria Containing the Degradation Enzyme to Degrade AHL Molecules in a
Solution
of a Certain Concentration.
For simplicity, we directly used the engineered bacteria expressing the enzyme to degrade AHL
molecules. The reporter bacteria (AHL biosensor) can respond to AHL molecules and express sfGFP,
thereby enabling the detection of the AHL concentration.
Figure 7. Plasmid map of the reporter
bacteria,
using the Rhl biosensor bacteria as an example.
The specific procedure is as follows:
1. Prepare an AHL-2YT medium of a specific concentration by mixing an AHL stock solution
with a
2YT liquid medium.
2. Take 160 μl of the degrader bacterial culture, centrifuge (5000 rpm, 1 min) to collect
the
cells, discard the supernatant, and resuspend the pellet in 160 μl of AHL-2YT medium.
Incubate
the resuspended culture in a shaker (200 rpm, at 25°C or 37°C) for approximately 3 hours.
3. After the degradation period, centrifuge (5000 rpm, 1 min) the mixture; the resulting
supernatant is the solution ready for detection.
Step III
Use of Reporter Bacteria to Detect the Concentration of AHL Molecules in the Degraded AHL
Solution.
Again, for simplicity, we used our constructed reporter bacteria to measure the AHL concentration
in
the test solution. The specific procedure was as follows:
In a 96-well plate, add 50 μl of the test solution, 75 μl of fresh reporter bacterial
culture
(OD600 ≈ 1.0), and 75 μl of fresh 2YT liquid medium to each well, resulting in a total
volume of
200 μl per well. The initial volume of 160 μl for the degradation reaction per sample is
designed to ensure that each group can be measured in three independent replicates.
Place the microplate in a plate reader and initiate measurements. The protocol consists of a
measurement step, followed by shaking for 4 minutes. Since each measurement takes
approximately
2 minutes, the total cycle interval is 6 minutes. Measurement parameters: A600, Fluorescence
(excitation 484 nm, detection 510 nm). The entire measurement process in the plate reader
lasts
approximately 3 hours.
Step IV
Plotting the Response Curve of the Reporter Bacteria to Evaluate Degradation Efficiency.
Finally, we exported the data from the plate reader. After excluding obvious outliers, we
typically
also removed the data from the initial 30 minutes. This is because the reporter bacteria exhibit
certain level of expression leakage, leading to a transient decrease in the relative
fluorescence
units (A.U./OD₆₀₀) immediately after dilution. The initial data were discarded to ensure more
reliable curve fitting. Furthermore, to account for background fluorescence and spectral overlap
between excitation and emission light, we subtracted the initial fluorescence value of each well
from its subsequent readings. The fluorescence value was then normalized to the OD₆₀₀ value to
represent the relative expression level, i.e., the response intensity. Finally, the response
curve
was fitted and plotted using GraphPad Prism 9.
Figure 8. Representative response curve.
Part III: Further Exploration on the Measurement
I. Estimating Actual Protein Expression from
ImageJ
Fluorescence
Although ImageJ provides robust high-throughput quantification of colony fluorescence, its output
is
limited to the 8-bit grayscale range (0–255), obscuring true biological signals at high
fluorescence
levels. To better interpret the experimental data, our modelling team developed a computational
approach that translates ImageJ measurements into biologically meaningful quantities, such as
cell
density and GFP protein expression.
The conversion model integrates multiple factors: the pixel value under red light (primarily
related
to cell density), the pixel value under blue light (transmission plus GFP fluorescence), and
baseline references at the start of the experiment. By mathematically combining these signals,
we
can extract the pure GFP contribution and estimate the underlying protein content in each
colony.
This approach corrects for signal saturation, differential light absorption, and colony growth
dynamics, thereby providing a more accurate representation of gene expression over time.
The code implementing this conversion will be uploaded to our iGEM repository, ensuring
transparency
and reproducibility. By standardizing the conversion of raw fluorescence data into biologically
interpretable metrics, this tool supports future LOGIC users in planning experiments, comparing
results across conditions, and integrating data into predictive models.
Figure 9. Principle of converting ImageJ
fluorescence measurements to biological quantities. Top-left: relationship between red light
pixel value and cell number; top-right: GFP amount over time derived from differential light
components; bottom-left: decomposition of light signals into blue (fluorescence +
transmission)
and red (transmission only) components; bottom-right: measurement principle showing the role
of
the yellow filter and extraction formula. This workflow allows accurate estimation of
colony-level protein expression and cell density from standard imaging data.
II. Predictive Tool Based on Quorum Sensing
Response
Curves
Building on the quantitative fluorescence data from four quorum sensing systems—Las, Rhl, Cin,
and
Tra—our team developed a lightweight predictive software tool to facilitate experimental
planning
and interpretation. The tool has two complementary functions:
1. Forward Prediction: Given a set of input parameters, including AHL
solution
concentration, diffusion distance, and incubation time, the software predicts the expected
fluorescence intensity. This allows users to preview colony responses under specific
experimental conditions without performing extensive preliminary tests.
2. Reverse Design: Usersmay specify a targetfluorescence intensity (0–255)
at a
given time point and a fixed AHL concentration. The software then recommends the
colony-to-AHL
distances most likely to yield the target signal.. This reverse-engineering function
supports
rational experimental design, reducing trial-and-error and enabling more efficient use of
resources.
Figure 10. The predictive tool
integrates
four quorum sensing systems (Las, Rhl, Cin, Tra) and allows users to input parameters
such
as concentration, time, and distance. The software then calculates the expected
fluorescence
intensity based on pre-trained models. In this example, the Las system was selected, and
the
predicted fluorescence value reached 227.00 under the given conditions.Figure 11. In addition to forward
prediction, the tool supports a reverse function. By entering the expected fluorescence
intensity together with concentration and time parameters, the software calculates
recommended colony-to-AHL distances. Here, for the Rhl system with an expected
fluorescence
of 150 at 12 h, the optimal distance is estimated to be 1.8 cm.
The code for this predictive tool will be uploaded to our iGEM repository in compliance with iGEM
standards, ensuring transparency and reproducibility. By integrating multiple quorum sensing
systems
and their dynamic response profiles, this tool extends the practical utility of our LOGIC
toolkit.
Future users can leverage it to design experiments, optimize input patterns, and predict system
behavior, thereby accelerating the design-build-test-learn cycle in synthetic biology.
Notably, the predictive tool is not limited to the four quorum sensing systems we have prepared
for
the iGEM community. Any inducible system with a well-defined input–output response curve— such
as
inducer-to-fluorescence datasets—can be readily adapted into our framework with minimal
modifications. This flexibility makes the tool broadly applicable, turning it into a versatile
“plug-and-play” platform for predictive modeling in synthetic biology.
Tips: How to Set Up the Control Group When Using the LOGIC Toolkit?
Establishing an appropriate control group is essential for ensuring reliable measurements
with
the LOGIC toolkit. Below we provide detailed guidelines:
1. Purpose of the Control Group
The control group serves two key purposes: (i) to correct for background
autofluorescence of
the medium and cells, and (ii) to represent the system's behavior under "zero-input"
conditions.
2. Spotting Pattern vs. Experimental Group
In any spotting pattern (whether provided by us or custom-designed), the only difference
between the control and the experimental group should be the presence or absence of
inducer
(e.g., AHL, IPTG) input. All other variables—including the bacterial strain and the
spotting
volume (e.g., 0.3 μL)—must remain identical.
3. Distance Irrelevance in the Control
In our experiments, the colony-to-AHL distance is a key variable. However, in control
groups
(without AHL), this distance has no biological meaning. Therefore, we typically
calculate
the average fluorescence intensity of all colonies on the plate. For example, in the
spiral
spotting pattern, the twelve colonies arranged radially are averaged to yield a single
baseline fluorescence value.
4. Application in Response Curves
When constructing AHL response curves, the effective fluorescence intensity of each
colony
in the experimental group is obtained by subtracting this baseline average from its
measured
fluorescence.
5. Application in Logic Gates and Cascades
For logic gate or cascade experiments, the averaged baseline fluorescence is used as the
00
input (or 000 input for three-input gates). The fold-change fluorescence is then
calculated
as the ratio between raw experimental fluorescence data and this baseline, without
subtraction.
6. Black Balance as Another "Blank"
Another essential blank control is the black balance performed before imaging. By
selecting
a dark region of the agar plate, we calibrate the microscope to ensure consistent
imaging
conditions and eliminate background noise from the camera and optics.
If you choose to use our batch-processing script, you can photograph this dark region
once
per plate and name it as "0". The quantified result of this "0" image should fall
between
0–3, which confirms that the black balance was correctly set and no abnormal background
signal exists.
This page doesn't work properly in this screen size, you might encounter major bugs that affect your
reading experience. Please try visiting us using a larger screen.
