Multiple trials went underway for the growth of J2T in knockout MetE media to definitively confirm that the predicted results occured in regards to NCM/J2T growth. As the goal of the experiment was to assure that there was no “leaking” of methionine from the J2T cells when stimulated with methionine production, the following test matrix was performed:

Test Tube # Control Components Results 1 Experimental → M9 Dropout with NCM removed (2 mL) → Added JT2 (25 uL) → KM (2 uL) No growth 2 Positive → M9 Dropout with NCM removed (2 mL) → Stock culture NCM (2 uL) Minor growth* 3 Positive → M9 Dropout with NCM removed (2 mL) → Added JT2 (25 uL) → KM(2 uL) → MetE (100 uL) Growth 4 Negative → M9 Dropout with NCM removed (approx. 2 mL) → KM (2 uL) No growth 5 Negative → Stock M9 Dropout (2 mL) → KM (2 uL) → MetE (100 uL) No growth 6 Negative → Stock M9 Dropout (2 mL) → Added JT2 (25 uL) → KM (2 uL) No growth 7 Positive → Stock M9 Dropout (2 mL) → Added JT2 (25 uL) → KM (2 uL) → MetE (100 uL) Growth

//will figure out how to format this table soon

In this chart, a negative control refers to an experiment where there was an expectation that there should be no bacterial growth as per our hypothesis, and a positive control denotes an experiment where there should have been growth expected.

As can be seen from the chart, the experimental trial, in which the NCM strain was grown and removed from the methionine knockout media, had no growth once J2T was introduced into it. This was a positive result as this suggested that there was little or no “leaking” of methionine produced by the NCM during growth into the media, and thus the J2T did not have any means to grow.

The minor growth of NCM as a positive in control in trial 2 was expected, and was done to confirm that there would be no adverse factors from the removal of NCM that were causing the introduced J2T to not be able to grow, and that the composition of the media was static.

Trial 3 showed with the repliaction of expirimental conditions and the addition of MetE that the reason that the growth was not happening in exsperimental conditions was in fact the lack of methionine in the M9 media. This showed growth as expected.

Trial 4 indicated no growth as expected. This was carried out, with the removal of the NCM and addition of the antibiotic, to confirm that all the bacteria was being taken out and that no other bacterial contamination was what was being seen in the positive control tests that had reported growth.

Trial 5 similarly was carried out with stock M9 to assure no environmental contamination was in the media from the beginning, especially bacteria that required external methionine to grow.

Trial 6 was carried out as a negative control to assure that the stock media itself was not contaminated with any nutrients that could be used by the JT2 to grow. This was important as it is possible that trace amounts could have been consumed in the other trials by the NCM bacteria and thus were not reflected in the experimental trial.

Finally, the 7th trial indicated that the JT2 strain that was in this bacteria was healthy and could grow in the desired conditions within the M9 media, with the methionine and the KM from stock solution.

GFP vs Non-Fluorescent Flow Test Experiment

Our comparison of the flow cytometer against pour plating started with a fluorescent and a non-fluorescent culture of roughly the same OD values. Below our initial ODs and results from both the flow cytometer and pour plating.

These results lead us to believe that the flow cytometer is representative of pour plating, indicating that the flow cytometer is most likely capable of giving accurate readings of a 50/50 population of fluorescent/non-fluorescent cells.

MetE Metabolic Load Experiment

Four sample tubes of JT2 CcaS/R-containing cells in M9 were made, and all were placed in the turbidostat. The samples were made by inoculating the cells off a plate into complete M9 made with casamino acids as the amino acid source.

Two samples were grown under only red light, and two under only green light. The samples were optically insulated from each other. The cells were allowed to grow for 3 hours to acclimatize them to the light conditions they were being kept under. By defining the time of the first OD measurement as time 0, the following trend of OD versus time was obtained (the first time point was excluded because OD was measured incorrectly at that time):

Upon converting this data to instantaneous growth rate over time, it became clear that doubling time was changing gradually but appreciably as OD increased. Therefore, to analyze the data, instantaneous doubling time was plotted against OD. The following graph was attained:

Though it is unfortunate that doubling time increases with OD, the general trend appears to be approximately the same regardless of light conditions. Therefore, it was determined that the metabolic load of MetE expression was negligible if it existed at all. See pages 77 to 79 in the online lab book for additional details.

NOTE: The increase of doubling time with OD was most likely a result of bad stirring, and hence poor aeration, of the samples in the turbidostat. This can be avoided in the future.

“Robot” Experiments

Initial Experiment

Having determined the utility of an automated sampling system, the team first set out to confirm whether or not cells with different levels of GFP expressed could be visually distinguished when exposed to a ~488 nm light source. To this end, a light source with a peak emission at 488 nm and a light filter capable of letting green light but not blue light pass through were borrowed from the Reed lab at the University of Waterloo.

Cells (specifically BW 29655 E. coli) expressing GFP under the CcaR promoter were grown to an OD of ~1 under both red and green light in complete M9. Unfortunately fluorescence measurements were not taken under the flow cytometer, but for the record the cells grown under red light were white while those grown under green light were visibly green. Upon shining blue light on the cells and taking a picture through the filter, the following image was obtained:

The container on the right was the one grown under green light, and it is very clearly distinguishable from the container on the right.

Box Design and Test with Photodiode

A box was designed using AutoDesk Inventor (a 3D computer aided design software) to hold the components of the measurement system (sample tube, LED, and photodiode/camera) in place for each measurement as well as to block out ambient light. Using CAD software also allows us to 3D print the final box design if we so desire. The measurement system involves exciting the GFP with an LED (488 nm) and measuring the emitted fluorescence (510 nm). To measure the emitted light without measuring the excitation light from the LED, the filter lent to us by the Reed lab was used so that only 510 nm light would be absorbed by the photodiode/camera. See the first image below for the initial box design, and the second image below for the principle of operation.

The design was initially tested using a cardboard box with holes cut in it as shown in the CAD drawing. We attempted to measurement the fluorescence of a sample tube of cells in the box with a photodiode. The box prevented ambient light from entering the system and kept the parts in place as desired, however we found we could not get consistent readings when we repeatedly measured the same sample; slight fluctuations in sample and LED position together with possible photodiode malfunction were responsible for these inconsistencies. As a result of this experiment, we decided to try using a camera instead. We also found that our initial placements of the components was not optimal and concluded that we must perform an experiment to optimize the distances and placements of components without the box.

Quantification Test with Camera

After the unclear results of the test with the original box apparatus, it was determined that the test using the camera should be redone in a more quantitative manner. Similarly to the initial experiment, E. coli (this time JT2 containing GFP under the CcaR promoter) were grown up to an OD of ~0.6 in complete M9. Two samples were grown under green light, and two under red light.

The tubes were arranged in a row in front of an iPhone camera, which our light filter in front of it. The tubes were placed in two alternate arrangements, and a movie was taken of the blue LED being passed behind them. See below for one of the movies taken.

As one can see (especially when looking at the bottoms of the vials), the first and third vials from the right are brighter. These were the vials grown under green light, and hence the cells in them had more GFP. Screenshots of the vials were taken as the light was directly behind them, and the images were cropped down to just the vial (see below for an example of a cropped image).

These images were then fed into a Python program that broke them down into their RGB values. The red, green, and blue values of each pixel in each image were taken and averaged over each image. Since some blue light still made it through the filter, average blue values for the images could not be used to distinguish the samples. Additionally, the green values were informative but still relatively similar in each image, mostly because a strong blue light actually emits some green light as well. However, examining the average red values made the cells grown under red and green light easily distinguishable, with the more fluorescent cells being more “red” (as GFP actually emits some red light).

This led the team to conclude that our approach for distinguishing fluorescent and non-fluorescent cells was valid, and the failure of the initial box design was due to poor placement and control of the position of our photodiode, sample vial, and LED. We thus set out to find the optimal arrangement and implement that in future designs.

For the full code of the Python program, see: https://github.com/igem-waterloo/uwaterloo-igem-2018/blob/master/models/image/greenFluorescenceQuantifier.py

For the full experimental protocol, see the online lab book (pages 80 and 81).

Optimization Test with Camera

Following the successful fluorescence-quantization experiment, the team determined that it was necessary to optimize the positions of both the camera and the blue LED relative to the sample. To do this, an experiment was designed according to the following diagram:

The green squares represent positions for the camera, the red squares represent positions for the blue LED, and the white circle represents the sample. Every combination of camera and LED position was tried.

For the samples, new cells were used. DH5α containing a GFP-expressing cassette (registry part I20270) in pSB1C3 was used as a “fluorescent” strain (called strain 346) and empty JT2 was used as a “non-fluorescent” strain. Both cultures were grown up from frozen stock overnight in LB, then the cells were moved to complete M9 for ~3 hours of growth. After the end of this growth, the fluorescence values of the JT2 strain and strain 346 were taken using a flow cytometer in the Ingalls Lab. Under a laser intensity of 10 mW and using a sample size 10,000 cells, the fluorescence in arbitrary units was 2375 for strain 346 and 19 for JT2.

From a previous experiment (see online lab book page 111), we have reason to expect that the OD to cells per mL relationship is similar for empty JT2 and strain 346. After 3 hours of growth in M9, JT2 was at an OD of ~0.15 and 346 was at an OD of ~0.3. Strain 346 was diluted down to an OD of ~0.15 with PBS, and then two 25 mL samples were prepared of each strain. Two ~50/50 mixtures of the strains were also prepared.

After the combination of camera and LED positions were tried and photos of the samples taken through the light filter, several things became apparent. The readings from all camera positions were less consistent between duplicates when the LED was 30 cm away. This was most likely because small changes in sample position or LED orientation resulted in large changes in how the light bounced off the sample tube. Therefore, it is better to have the LED 10 cm away from the sample. When the camera was 5 cm away, it could not focus properly. Therefore, it is better to have the camera 20 cm away. When the LED, sample, and camera are in a line, the light coming from the LED can saturate the image; it goes completely white regardless of the sample type, making it hard to distinguish fluorescent and non-fluorescent cells. Therefore it is best to have the path from the sample to the LED and the path from the sample to the camera be perpendicular.

From the above observations, it became clear that the best setup was to have the LED 10 cm away from the sample, have the camera 20 cm from the sample, and have the LED-sample-camera path to be in an L shape. For that particular set up, the results below were obtained after analysis using the Python program for getting average RGB values.

The fluorescence values are a simple average between the values for the duplicates. As one can see, strain 346 is clearly more fluorescent than JT2, and the mixture is of an intermediate fluorescence. This confirms that the system can be used to measure fluorescence and distinguish between different mixtures of fluorescent and non-fluorescent cells. The mixture should have been ~50% fluorescent cells, and by linear interpolation from the data obtained one can get that the mixture is 58% fluorescent cells (from the calculation (100%)*(94.7-71.85)/(111.25-71.85)).

Second Design and Future Experiments