Team:NTNU-Trondheim/Results
Results
Heme approach results
Experiment overview
We performed 16 experiments, which can be divided into three categories based on what variable was explored. This includes varying H2S concentrations, varying protein concentrations and kinetic studies (varying the time) with 2-3 parallels. All experiments involved the observation of the reaction between H2S and N-terminus Cy3-tagged myoglobin, using fluorescence intensity (FI) measurements. Additionally, two emission/excitation scans were performed to find the optimal excitation and emission bands for these measurements. The instrument used for all fluorescence intensity measurements was the Tecan Infinite 200Pro.
Initial results
In the graph below are the results of experiment 1. Here we measured using 5 mg/L of labeled myoglobin-Cy3 reaction mixture, and varying H2S concentrations around the level which is supposed to be unhealthy, not deadly, for fish in RAS. For the measurement we used the optimal excitation/emission wavelengths for our fluorophore, Cy3, which is 550nm and 570nm. As you may see, these results are fairly unpromising, as there is high variance and no discernible trend. Further, we would expect our 0-value to be the highest, as the presence of H2S should lower the fluorescence intensity of the reaction mixture. However, we got odd values for our blank wells (which is just water and buffer), at around 50000 FI, much higher than all the values for the measurement points.
Figure 1: Fluorescence intensity measurements for varying H2S concentrations using 5 mg/L labeled protein reaction mixture. Measurement parameters were 550nm for excitation, 570nm for emission and with 60 gain.
Excitation and emission measurements
Due to the high variance and high FI values for the blank measurements from our initial results, we suspected that the excitation/emission parameters might be too close. The instrument we used for fluorescence measurement uses a 9nm wide band for emission measurement, and a 20nm band for excitation. Since our chosen excitation/emission peaks for the first experiment were 550/570, the strange values might be caused by spectral overlap or poor calibration of the instrument. To investigate further we performed excitation and emission scans for the respective emission and excitation values. The results for these scans can be found in figure 2, for the excitation scan, and figure 3, for the emission scan.
Figure 2: Excitation scan of 5 mg/L reaction mixture (1) and 25 mg/L reaction mixture (2), as well as control solutions containing the corresponding myoglobin concentrations and blank solutions. Measurement parameters were 520-560nm in 2nm increments for excitation (only 520-544 is shown in the figure due overreading), 570nm for emission and with 60 gain.
Figure 3: Emission scan of 5 mg/L reaction mixture (1) and 25 mg/L reaction mixture (2), as well as control solutions containing the corresponding myoglobin concentrations and blank solutions. Measurement parameters were 550nm for excitation, 560-600nm in 2nm increments for emission (only 576-600nm is shown in the figure due to overreading), and with 60 gain.
From these results, we can see that there is a large increase in fluorescence intensity when measuring emission at 570nm and exciting higher than 545nm, and likewise there is an increase when exciting at 550nm and measuring emission lower than 575nm (it is important to note that for the emission values the real area is ±4.5nm, and for excitation it’s ±10nm). For the other values we can see that the reaction mixture measurements have the highest fluorescence intensity, so ideally we want to measure in this area. Using this, we decided to measure with excitation/emission pairs such as 540nm/570nm, 540nm/580nm and 550nm/580nm for the next few experiments. We then saw that 540nm/580nm had the lowest variance of the three pairs, and similar results, so we then decided to use only 540nm/580nm for practical reasons. Due to this, the rest of the results will be from the 540nm/580nm data.
Further results
Following our excitation/emission scans, we performed the measurements for experiment 2 using our new measurement parameters, and got better results. This time, our blank measurements were close to zero, our zero-measurements were the highest, there was lower overall variance and we got somewhat of a better trend, though not perfect. We used a wider range of H2S concentrations for this experiment, which at the higher values has a large molar excess over the 5 mg/L reaction mixture (around 42 times for 25 micromolar of H2S)
Figure 4: Fluorescence intensity measurements for varying H2S concentrations using 5 mg/L bound protein reaction mixture. Measurement parameters were 540nm for excitation, 580nm for emission and with 120 gain.
Using the three parallels from each measurement, we set up a two tailed T-test table to study the statistical significance of the data. This is shown in table 1, and is color coded such that any value below 0.025 is a deep green, above 0.050 yellow and red towards 1. Here we can see that we have good statistical significance for the difference between the zero-measurements and 1uM measurements, as well as between zero and 5. There is however very low significance between 1uM and 5uM, which makes sense since their means are so close. Additionally, there is good significance between 5um and 10um, but in the opposite direction from what we’d prefer (increasing FI by concentration instead of decreasing).
Table 1: Pairwise results of two-tailed
T-test
for each of the
measurements
| T-test | Blank | 0 uM | 1 uM | 5 uM | 10 uM | 25 uM |
| Blank | x | 0.0000 | 0.0014 | 0.0000 | 0.0003 | 0.0007 |
| 0 uM | 0.0000 | x | 0.0240 | 0.0009 | 0.0565 | 0.0864 |
| 1 uM | 0.0014 | 0.0240 | x | 0.5457 | 0.1568 | 0.1885 |
| 5 uM | 0.0000 | 0.0009 | 0.5457 | x | 0.0160 | 0.0401 |
| 10 uM | 0.0003 | 0.0565 | 0.1568 | 0.0160 | x | 0.6086 |
| 25 uM | 0.0007 | 0.0864 | 0.1885 | 0.0401 | 0.6086 | x |
Due to the unwanted upwards trend towards the higher values, we speculated that the large molar excess could be over-saturating the reaction mixture, so in experiment 3 we tried using a higher mixture concentration (25 mg/L) and a narrower/lower range of H2S concentrations (in the ranges that was deadly for fish in RAS). As you can see in figure 5, this did not go well. We got very high variance, no discernible trend and the zero-measurement wasn’t significantly higher than the measurements with H2S.
Figure 5: Fluorescence intensity measurements for varying H2S concentrations using 25 mg/L bound protein reaction mixture. Measurement parameters were 540nm for excitation, 580nm for emission and with 100 gain.
We then lowered the protein concentration back to 5 mg/L, and performed a series of experiments with different H2S ranges, trying to determine a more or less linear range for this protein concentration. In figure 6 and 7 we see the results of experiment 5 and 6, which both had similar trends as in experiment 2, where there is first a dip in fluorescence intensity at lower concentrations then an increase. In both experiment 5 and 6 the measurements for the first H2S value, 0.1uM and 0.001uM respectively, gave a statistical significant change in FI from the zero-measurements.
Figure 6: Fluorescence intensity measurements for varying H2S concentrations using 5 mg/L bound protein reaction mixture. Measurement parameters were 540nm for excitation, 580nm for emission and with 120 gain.
Figure 7: Fluorescence intensity measurements for varying H2S concentrations using 5 mg/L bound protein reaction mixture. Measurement parameters were 540nm for excitation, 580nm for emission and with 120 gain.
Investigations:
Following these experiments, we tried investigating different protein concentrations, as well as the effect of time. In figure 8 we see the results of experiment 8, where the FI of several protein concentrations are shown, measured with three parallels at three time-points. As evident, time doesn’t seem to be a large factor in this experiment. Additionally, we can observe that the larger the protein concentration the higher the variance, particularly past 10 mg/L.
Figure 8: Fluorescence intensity measurements for varying reaction mixture concentrations at three separate timepoints. Measurement parameters were 540nm for excitation, 580nm for emission and with 100 gain.
To investigate the reaction with H2S further, we performed a series of kinetic experiments. These measured the fluorescence intensity of the reaction mixture with and without H2S present over a period of 30 minutes. The results for one of these experiments, experiment 14, is presented in figure 9. Here we see that in the very beginning there is good separation between the H2S FI (blue line) and null FI (orange line), but this shrinks over time, and at about 1200s the H2S FI is in fact larger than the null FI. An important thing to note was that it was unexpected for the null FI to decrease at all, since this should only happen if the reaction mixture comes in contact with H2S. This happened in all the kinetic experiments we performed.
Figure 9: Fluorescence intensity measurements for kinetic study over 30 minutes, using 10 mg/L bound protein reaction mixture and 0.1 uM H2S. Measurement parameters were 540nm for excitation, 580nm for emission and with 100 gain.
Due to the strange results of the previous kinetic experiments, we decided to investigate whether the light from the fluorescence measurements could be degrading either the fluorophore or myoglobin component in the reaction mixture, leading to false readings. In experiment 16, we tried checking for this by performing several 30 minute kinetic measurements, but with varying measurement intervals; 5, 30, 60 and 300 seconds. The results are presented in figure 10, and here it is evident that with the most frequent kinetic intervals of 5 seconds, there was a significant drop in FI compared to less frequent intervals. This is visible when comparing the 5s measurements the red lines) to the grouping of the other intervals’ measurements. However, even for the less frequent intervals the null FI is still dropping over time, and though there is better separation over time this also shrinks to close to the same FI value for many of the lines.
Figure 10: Fluorescence intensity measurements for kinetic study over 30 minutes, using 10 mg/L bound protein reaction mixture and 1 uM H2S. The experiment was run several times using different measurement intervals, and these intervals are A for 5s, B for 30s, C for 300s and D for 60s. Measurement parameters were 540nm for excitation, 580nm for emission and with 100 gain.
Discussion
We managed to get statistically significant differences between null measurements and certain H2S measurements with our reaction mixture early on. Beyond that, however, most experiments in our project involved investigating different factors that affect the variance of our measurements, and potentially the trend between different H2S concentration measurements. Of these, we are fairly certain that light exposure is a big factor in determining FI measurements over time, which makes sense given that the fluorophore we used is light sensitive. Unfortunately, a lot of other important factors were harder for us to test due to our limited resources and instruments.
However, inferring from the results, we speculate that a lot of the variance could be explained by the free diffusion of H2S from the open wells into the air inside the fluorescence measurement instrument and then back into the null wells. Specifically, the decrease in FI of the null measurements in the kinetic experiments seems to indicate this. This speculation is based on H2S being very volatile, and air diffusion within a somewhat enclosed space can distribute H2S in a higher concentration effectively. The diffusion back into the solution in the null well should be less effective, but given the sensor having a low enough limit of detection it would explain the several lines in experiment 16 stabilizing at the same FI value over time. To check if this was the case, we would have needed a fluorometric instrument with closed and separated measurement wells .
There are also some other factors that also could affect the variance. Firstly, H2S can be oxidized very easily, which means that in addition to the H2S lost to diffusion, there might be a substantial decrease in H2S while it’s in solution. To prevent this we would have needed a closed system which we would deoxygenate using a degasser, bubbling argon or nitrogen through. Secondly, the oversaturation of H2S in the reaction mixture with respect to a potentially linear measurement area could explain both variance and the lack of a general trend for our measurements. Myoglobin has a higher chance to irreversibly bind to H2S when it’s present in a relatively high concentration, which we suspect would not produce a signal (decrease in FI), and could explain the increase in FI generally at higher H2S concentrations as seen in experiment 2 and 5. To investigate this meaningfully would require us to lower the variance caused by the prior factors, since as we saw in experiment 3 and 8 the variance seems to increase substantially with increasing the protein concentration.
In order to make sure that H2S was produced in the prepared solutions of Na2S, we used the methylene blue method. Using this method, we observed a colour change to blue, verifying the presence of H2S. We could therefore rule out a lack of H2S production as a source of error in our experiment.
Conclusions
In summary, although we have some results indicating the potential for an H2S sensor with a very low level of detection, the variance and lack of trend in important ranges severely inhibit the sensor performance in other areas. We however note that improvement may only require better suited hardware. As our proposed implementation would be in an LOC, some of the factors affecting variance might be diminished, such as the H2S diffusion between measurement wells.
SQR approach results
On this page the results from the experiments detailed on our experiments page are briefly presented.
Cloning
The five constructs ordered from IDT were cloned into plasmid vectors by Gibson assembly. Next, the plasmids were transformed into supercompetent Escherichia coli cells. Sanger sequencing was used to verify if the cloning had been successful. Four out of five designs came out successful.
The plasmid vector, pENZ004, was PCR amplified with primers containing overhangs corresponding to the ends of the constructs. Since some of the constructs were identical on either side, three plasmid amplifications were made. The PCR gel analysis, shown in figure 1, showed that all three plasmids had been PCR amplified, to the desired length (ca 2.1 kb).
Figure 11: PCR gel analysis of PCR products after pENZ004 is amplified with primers containing overhangs for Gibson assembly. 1kb ruler is used on either side of the plasmids.
Testing
The next step was to test the whole cell sensors. Design 1 was tested with GSSH produced from GSSG and specific concentrations of H2S. Design 2 was tested with H2S directly. Exact concentrations and procedures are found on our experiments page
When exposing the whole cell sensor to GSSH (design 1) and H2S (design 2) the intensity of the fluorescence signal was expected to increase with increasing H2S concentration. Unfortunately, none of the whole cell sensors showed a fluorescence signal when observed under fluorescent light, see figure 2, or when fluorescent was measured on a Tecan plate reader, see figure 3.
Figure 12: Our whole cell sensors after treatment with GSSH and H2S under UV light. On the right, a control containing only LB medium. There was no sign of fluorescence in the samples.
Figure 13: Measured fluorescence using a H2S sensing whole cell sensor plotted against H2S concentration. The system is designed to be sensitive for S2- and includes the reporter gene CreiLOV and a weaker promoter for SqrR. No trend is observed in the data.
As detailed on our engineering page it was discovered that mistakes made during the ordering of our constructs resulted in the CreiLov gene being inserted without a start codon. In addition, the ribosome binding site sequence was inserted twice and inverted, likely disturbing translation of the constructs. It is also a possibility that the cell membrane is impermeable to GSSH due to the size of the molecule.. To improve the approach, an in vitro kit could be utilized. The boundary of the cell membrane would then be eliminated. If time had permitted we would have liked to fix these mistakes and study the approach further. Read more about this on our engineering page
