Yes, we did it! We engineered a glucose detection system that expresses mCherry according to glucose in the environment. With various experiments we detected fluorescence in Pseudomonas putida. Therefore our system is the first step towards a long-term treatment for diabetic patients. Encapsulation of this measurement technique allows implantation of our device into the human body. It is a long way to go to once and for all treat diabetes. But we just made the first step toward a novel treatment method based on synthetic biology, that improves the quality of life of so many patients.
Transformation of our devices into the Pseudomonas putida plasmid
Since we are working with P. putida we could not use the linearized plasmid backbones for transformation that are provided by iGEM. Instead we used a plasmid that was already transformed into P. putida at an earlier time. We cut this plasmid with the blunt end enzyme Hpa1 and ligated our device by means of blunt end ligation into P. putida.
We did this with all four devices, therefore there is the possibility that they are ligated in different directions. However, this should not make a difference in the expression pattern of mCherry.
Plate reader measurement in M9 medium with different concentrations of glucose
Three biological replicates of each transformed device were inoculated in M9 minimal medium with 5 %, 10 % and 20 % of glucose, respectively. After 16 hours incubation 100 µl of the cultures were transferred to a black 96 well plate with transparent bottom. The plate reader measured the fluorescence (excitation: 584 nm, emission: 620 nm) as well as the optical density at 600 nm. Each sample was measured three times (technical replicates).
Figure 1 displays the average of the technical and biological replicates of each transformed device. The error bars show the standard deviation. The dashed line shows the linear trend line.
The edd device has the strongest signal in media with 5 % and 10 % glucose, compared to the other devices. Second highest signal is shown by the kgu device. However, the standard deviation of these samples is higher compared to the edd, zwf and gad device. The increase of fluorescence with increasing glucose concentration is most distinct in the zwf device. The fluorescence level in gad increases slightly, but not significant between 5 % and 10 % glucose in the medium. Nonetheless, the fluorescence at 20 % glucose in the medium is remarkably higher.
Hence, the trend clearly shows an increase in fluorescence with increasing concentration of glucose in the medium for all devices. This is what we expected since the promoters are not repressed when certain derivatives of glucose bind and thereby change its conformation. Therefore, our devices can be used as detection method for glucose in the medium.
If only the 5 % and 10 % values are regarded, the increase is much slighter. However, at least for the zwf device a significant increase from 5 % to 10 % can be seen.
Plate reader measurement in LB medium with different concentrations of glucose
In M9 minimal medium, glucose is the only carbohydrate source. However, that is not the case in blood where other carbohydrates such as other saccharides occur. Therefore, we tested the functionality of our devices in LB medium, which contains other carbohydrates than glucose.
Three biological replicates of the transformed devices were inoculated in LB medium with 5 %, 10 % and 20 % glucose, respectively. After 12 hours of incubation, 100 µL of the samples were transferred into a black 96 well plate with transparent bottom. The fluorescence as well as the OD600 was measured (Figure 2). Three technical replicates were conducted of each sample.
As with inoculation in M9 medium with different glucose levels, the trend shows an increase in fluorescence with increasing concentration of glucose in the medium. However, the increase is less steep. An exception is the gad device, which displays an exceptionally high level of fluorescence. The reason for that might be, that the values on the diagram are the fluorescence values read by the plate reader divided by OD600. This is done to standardize for the amount of cells in the sample. However, if there are hardly any cells in the sample, the OD600 limits to zero. Thus the quotient of fluorescence and OD600 results in a high value.
Regarding the edd device, the finding that is expresses the highest level of mCherry in the presence of glucose could not be confirmed when incubated in LB medium instead of M9. Instead, it appears to have the least fluorescence. In this setup, gad displays the highest levels of mCherry expression. The devices kgu and zwf display approximately the same values for all concentrations of glucose (see overlapping trendlines).
Interesting are the remarkably small standard deviation for all samples (except the outlier). Further, the absolute values in LB medium are much smaller than in M9. This was expected since the expression of mCherry should not be influenced by other carbohydrates in the medium.
Plate reader measurement over time
One biological replicate of all transformed devices was inoculated into M9 medium with 1 %, 5 %, 10 % and 20 % glucose, respectively. After 6 hours incubation, 100 µl of each were transferred into a black 96 well plate with transparent bottom. The fluorescence and the OD600 was read every 30 minutes for 6 hours.
A. Growth measurement in M9 medium with different concentrations of glucoseFigure 3 shows the optical density as a measure for cell growth over time.
First of all, the samples are measured on the plate without the lid on. Hence the medium can evaporate and therefore the OD600 decrease. That happened with the samples that were inoculated in M9 medium with 20 % glucose. The values for OD600 decreased, because the cells did not grow and the media evaporated. The four all devices inoculated in M9 with 5 % and 10 % glucose, respectively, increased linearly over the whole 6 hours. However, the samples in 5 % glucose grew faster than the ones in 10 %. It is interesting to see the behaviour of the cells inoculated in M9 medium with 1 % glucose. They grow faster than the samples in media with higher concentrations of glucose up to 3 hours, that is 9 hours, considering that they grew for 6 hours before the start of the measurements. From that point on, their growth decreased and the OD600 as well, presumably owing to the evaporation of the media. Obviously, the medium in the samples with 5 % and 10 % glucose evaporated accordingly. Therefore the growth in these media is even higher than the displayed values might suggest.
These results are important for the discussion when the mCherry expression level of the cells should be measured. Further, it gives a picture of the toxicity of high glucose concentration to P. putida.
In general, the growth behavior of all cells in media with different glucose concentrations seems to be similar, independently of the device that was transformed.B. mCherry expression over time
From the same experiment, obtained results show the fluorescence standardized (that is, divided) by OD600 over time.
Generally, the device should display the same level of fluorescence over time to function as an efficient glucose detection system.
While the fluorescence level of edd with 1 % glucose in the media remains approximately the same, the fluorescence level in media with 5 % and 10 % slightly decreases. The cells in media with 20 % glucose display interesting fluctuations in mCherry expression, with increasing and decreasing values of fluorescence every 1-2 hours. This could be due to degradation and anew expression of mCherry.
The fluorescence of kgu cells remains approximately the same over time in all concentrations of glucose in the media. Only for cells inoculated in 20 % glucose a decrease in expression level within the first hours of measurement is visible. Then, the fluorescence seems to increase again. However, as stated earlier, the increasing values could be due to the decreasing OD600.
The same is true for zwf transformed cells. As with kgu cells, the fluorescence seems to increase over time in medium with 20 % glucose. However, the opposite is true for cells inoculated in medium with 10 % glucose. There, the fluorescence is decreasing over time. The values in media with 1 % and 5 % glucose, respectively, are not changing.
Interesting results are visibly for cells with the gad device transformed. As with the other devices, the cells at 20 % glucose show a slight increase in fluorescence level. However, what is different compared to edd, kgu and zwf, the fluorescence of the samples inoculated in lower glucose concentrations increases remarkably. This is an important finding when using the device as a glucose detection sensor.
Considerations towards this experiment are, that even though three technical replicates are measured, it was only conducted with one biological replicate. Therefore it has to be repeated to see if the tendencies are significant or only random fluctuations.
The next steps that we are going to take to improve our system is the exchange of the current untranslated region (UTR) with a UTR that maximized the level of fluorescence. When we did this, we are going to measure lower glucose concentrations, in the range of the blood glucose level of diabetic patients, that is 4 mM to 20 mM (0.072 % to 0.36 %).
Cell Encapsulation Results
A possible future application of our system is to have it implanted in the body, to function as a glucose sensor coupled with insulin secretion. We wanted to develop our system further by encapsulating the engineered P. putida in alginate. By immobilizing the cells in alginate microcapsules, they would be shielded from the immune response at the same time as small molecules like glucose and insulin can diffuse in and out.
We have encapsulated P. putida in alginate microcapsules (approximately 230 μm in diameter). Figure 8 shows one of these capsules containing the bacteria that are expressing the red fluorescent protein mCherry.
Figure 9 shows a 3D visualization of an alginate capsule containing mCherry-expressing P. putida. The 3D model has been constructed from a z-stack of confocal microscopy images.
To assess the stability of the capsules, a protocol for monitoring bacterial leakage from the capsules was developed. The encapsulated cells were stored in different media at both 30 oC and 37 oC (the optimal temperature for P. putida and body temperature, respectively). At 0, 2, 4, 6, 8 and 10 hours of storage, a sample was taken from the tubes containing the capsules. After filtering out the capsules themselves, it was possible to investigate how many bacteria had escaped from the capsules to the medium by doing different measurements of the medium.
Figure 10 shows the results from plating out 100 μl of storage medium, with a comparison between LB and PBS as storage medium. The experiments started with LB as storage medium, where the leakage was apparent after a short amount of time. Coating capsules with poly-L-lysine has been shown to be efficient for stabilizing alginate-encapsulated human cell lines that has been implanted. However, this did not improve the bacterial leakage from the capsules when stored in LB medium. LB is indeed optimal for bacterial growth, and would perhaps not be representative for a future application of our system (like an implant). By switching to PBS medium, a substantial improvement of bacterial leakage from the capsules could be seen.
In addition to plating out, the optical density and fluorescence of the filtered storage medium were measured with a plate reader. Figure 11 shows the third experiment with PBS as storage medium, where the difference between the filtered storage medium and the blank is shown in percent. Although the plating out showed an increase in bacterial leakage after four hours, the plate reader results puts the numbers in perspective. The graphs show that both the OD and the fluorescence do not increase or vary much over time.
To put the number of colonies on the plates further into perspective, the number of encapsulated cells in one storage tube can be considered. The OD of the cell culture in the alginate/cell mixture is approximately 0.05, which in cell concentration would equal around 40,000,000 cells/ml. In one storage tube, there is 0.5 ml of capsules containing around 20,000,000 cells. If plating out 100 μl results in 20 colonies (20 single cells) after four hours of storage, the number of free cells in the whole 30 ml tube would be 6000. This means that only 0.03 % of the cells has leaked out from the alginate capsules!
Although it was not possible to do within the timeframe of this project, it could be possible to reduce bacterial leakage by further tailoring of the capsule properties. This could entail encapsulating the cells in more than one layer of alginate, using a different polymer than alginate, or a combination of the two. The next step could also be to engineer the growth rate genes of P. putida.
Improvement over the state of the art
Our glucose sensor is the only, and first ever, sensor system that directly detects and responds to glucose levels based on synthetic biology methodology. Currently, nurses and medical doctors in hospitals use optical-based sensors for measuring glucose levels in diabetics. Previously, the 2011 iGEM team of Missouri made an indirect glucose detection system, which is based on detecting changes in osmolarity caused by changes in glucose levels. Since our proposed glucose sensor is based on direct sensing of different glucose concentrations, it has the potential for a more precise measurement technique than both the optical sensor and the system of the Missouri iGEM team.
The optical sensing system that is used in hospitals is typically applied twice a day on a patient. This is not frequently enough to get a good handle on glucose-level changes in patients. Our system is designed to continuously measure and report glucose levels. Moreover, upon detection of a critical concentration of glucose, our proposed system will begin to secrete insulin at a dose commensurate with the length of time the supercritical glucose levels are present. Currently at hospitals after detection of glucose with the optical system, the same dosage of insulin will be injected to different diabetic persons with different levels of glucose in their blood.
Consequently, the use of our engineered P. putida strain as a glucose sensor should be considered as a more sensitive method than the existing optical sensor. Additionally, it is not a time consuming procedure, since after its implantation a diabetic person does not need to get a daily care by doctor or nurse. Therefore, by decreasing the workload of medical personnel and physical equipment, the cost of glucose sensing is expected to drop. It is conceptually possible to implant the bacteria since they are encapsulated in alginate, which renders the product undetectable by the immune system of the patient. After implantation, our encapsulated P. putida will last for a long period, so that is clearly a positive point and improvement compared to the optical system which requires frequent access and controls. Finally, a functioning implementation of our proposed system has the potential of significantly suppressing the symptoms of diabetes, thus making it possible to live as a non-afflicted person.
In addition, this would be a more convenient method for diabetics because they can most likely live as a normal person without diabetes.