Measurement | NYCU-Taipei

Overview

Growth phase measurement is not only a fatigued work but also time-consuming. Depending on different species, dozens of hours or a couple of days are required. Even though there are several methods to prevent the hard work, like a probe in the culture tank or a microplate reader with a culture function, it is rather expensive. Therefore, we propose a different method for growth phase measurement, the fluorescent-labeled phase-dependent promoter. An important aspect of this method is to allow the real-time tracking of the growth phase by the naked eye. By combining the fluorescent protein with the phase-specific promoter, the growth phase can be easily determined by the color change of both the colony and liquid.

Fig. 1) The visual effect for each phase indicator in liquid culture

Fig. 2) The visual effect for each phase indicator in solid culture

In our growth phase indicator project, we proposed a mid-log phase specific promoter and a late-log phase specific promoter that are previously unexisting in the iGEM part. Those promoters are verified by combining the low maturation time fluorescent proteins and underwent fluorescent intensity and OD measurement. The stationary phase-specific promoter is also proposed; however, it is limited to a similar condition compared with our experiment since it uses the chromoprotein that has longer maturation to serve as the timer. If the condition is quite different, we suggest using the stationary phase-specific promoter that has already existed as the iGEM part. For example, the stationary phase-specific promoter (osmY promoter) proposed by MIT 2006. Note that the three phase indicator should be used together since the fluorescent protein would not degrade on time.

Fig. 3) The construct for nirBDC-mCerulean

Fig. 4) The construct for glpABC-mCherry

Fig. 5) The construct for hchA-AmilCP

The remote monitor of bacteria growth can also be experienced with the help of our LineBot and monitoring device. With those, the growth status on solid media can be recorded and transferred into LineBot, providing real-time measurement of bacteria growth status.

Fig. 6) The schematic for remote monitor

Results

Growth Status Indicators

We created several constructs by combining potential phase-specific promoters with reporter genes. For selecting the phase-specific promoter, the fluorescent intensity and OD measurement are performed to test the promoter expression. The result suggests that two promoters, the nirBDC promoter and glpABC promoter, have shown mid-log phase-specific expression and late-log phase-specific expression respectively. The result also suggests that the hchA promoter is able to serve as the stationary phase indicator when combined with a reporter gene that has a longer maturation time.

Fig. 7) The fluorescent intensity and OD₆₀₀ - time graph for nirBDC-mCerulean and glpABC-mCherry

Fig. 8) OD₆₀₀ value - time graph for hchA-AmilCP

Fig. 9) The photo of 9hr to 10hr for Fig. 8

Device Process

Take a picture and rotate 60 degrees, take 6 pictures
              Tools used: servo motor DS3320 (servo motor), 3D printing turntable (material:
                  PLA)
Software: Arduino (write motor rotation program), CREO (turntable drawing)
                
Photo saving to computer
              Software: Python-saving to computer to store the experimental datas and also
                  prepare to be process
                  by deep learning model

Run deep learning model
              Model: Yolov5m
                  
                  Yolov5 model is faster and more accurate, the model environment is simpler and newer,
                  there are multiple models V5s, V5m, V5L..., more selective.
                
                  Using 200+ pictures to train the model (with more than 90% of accuracy), basically can detect most
                  colonies that are bigger than Radius of 1 mm. And can separate good colonies and unusable ones
                
                  Other important Python libraries: Cv2 (take photos, record RGB), pytorch (print
                  into yolo model), numpy
                  (convert photos to matrix)
                
                  Output information: Upload time, photo file name, colony coordinates, RGB value,
                  colony type (good
                  colonies and unusable ones)
                
                  To know which phase of colonies are, we have to access theRGB value of colonies, and double check with
                  colony size.
                
Upload to google spreadsheet
              Using software/library: gspread + oauth2client(upload google spreadsheet)
Google has an upload limit of 60 requests per second, so we make a program to sleep every 50 seconds
                  to avoid warning information.

Linebot read PHASE information and send to user
              Get data: phase (inferred from RGB values)

                  Since the mid-exponential phase green color is barely visible, we then choose to separate the color
                  red which indicates the late exponential phase, and blue which indicates the stationary phase. We then
                  took a look at the experimental data, and noticed that with the RGB value B way bigger(more than 30)
                  than G and B, the color is blue. With the color R bigger(more than 20) than G and B, or the yellow
                  color(sometimes the red color would not be so distinct ), the color is red. As a result, we can
                  separate late exponential phase, stationary phase, and part of mid-exponential phase(most of time they
                  are too small to see)
                
Software used: Heroku

            Device Demonstration

Model

We choose “the re-growth time from different phases”, “initial quantity” as two variables for the effect of growth curves. As for the data, we choose the experimental data which was subtracted from blank. First of all, use a logistic function form with three parameters: L goes for the maximum OD value, k stands for the growth rate of mid-log phase, and ti denotes the duration of the lag phase. Secondly, fit data with GRG nonlinear regression in excel using solver. Then, the influence of the variable on the physiologically significant parameters can be obtained.

Discussion

For future improvement, we aim to prevent the interference of lasting fluorescent intensity. We proposed the degradation system that insert TEV protease cleavage site BBa_J18918 into mid-exponential phase fluorescent protein mCerulean and insert HIV-1 protease cleavage site BBa_I712015 into late-exponential phase fluorescent protein mCherry to form a split protein. The TEV protease BBa_K1319008 is combined with late-exponential phase-dependent promoter glpABC. When bacteria enters late-exponential phase, glpABC activates and initiates the expression of TEV protease, which cuts off mCerulean, resulting in time-dependent regulation of protein degradation. The HIV-1 protease BBa_I712667 is combined with a stationary phase-dependent promoter hchA. This allows the fluorescence protein to be quickly degraded in the next phase.

Fig. 10) Schematics of degradation system