Team:USTC/Proof

Modeling

Chengle Zhang

Yu Xie

Ya Jiang

After the preliminary investigation and the massive work on building our two systems, there comes the most exciting part of our project----the verification part. We conducted several verification experiments to test our hypothesis.

Experiment Design

For the Pro Priontein system we conducted experiments to measure the fluorescence intensity of the split sfGFP in both Escherichia coli(E. coli) and Saccharomyces cerevisiaes(S. cerevisiaes). In E. Coli, the gene for sfGFP 1-10 was expressed in BL21 on a pSB1A3 vector and the gene for sfGFP 1-10 was on a pSB1C3 vector. In S.cerevisiae (S. cerevisiae), sfGFP 1-10 was placed on YeWGAP and sfGFP11 was placed on YeUGAP.(YeWGAP and YeUGAP are two kinds of yeast plasmids which enable the transformant of yeast cells to grow on a medium lacking tryptophan and uracil respectively. )

Firstly, we compared the fluorescence intensity of the E. coli respectively containing sfGFP1-10 and sfGFP11 with that of the E. coli in which sfGFP1-10 and sfGFP11 are coexpressed to see whether or not our spilt sfGFP system can work as an effective indicator in real condition.

What we expected is that the E. coli can produce much brighter fluorescence when the two separate part of sfGFP protein (sfGFP1-10 and sfGFP11) are expressed in the same bacteria.

And then we changed our chassis to S. cerevisiae to verify the function of the kill switch we designed, which works like this: heat shock can cause the aggregation of Sup35 while GdnHCl can disaggregate them. We tried to detect the aggregation level of Sup35 under temperature and GdnHCl concentration gradient by measuring the fluorescence the S. cerevisiae produced. As what we’ve already mentioned before, the aggregation condition of Sup35 is indicated by the fluorescence intensity. The low level of fluorescence intensity can indicate high level of aggregation of Sup35, this is because when sup35 aggregates, sfGFP11, which is fused with sup35, will fail to bind with sfGFP1-10. According to the mechanism of split sfGFP system, it can lead to weaker fluorescent intensity. Briefly speaking, the brighter the fluorescence is, the weaker the proteins' aggregation is.

We conducted experiments to compare the different fluorescence intensity induced by heat shock under temperature gradient. And in further experiment, we added GdnHCl solution to the system and then we measured the fluorescence to see if GdnHCl can disaggregate the protein. So that we could know if our kill switch can works effectively.

The expected results are fluorescence intensity will decrease under heat shock. But if we add GdnHCl solution to the system the fluorescence will turn back due to the curing function of GdnHCl.

Experiment Results

1. The spilt sfGFP experiment in E. coli

To test if the split sfGFP can function as our expectation, we transform the plasmids containing sfGFP1-10 and the plasmids containing sfGFP11 respectively in BL21, cultivating the bacteria at 37°C and shacking at 250 rpm/min overnight. We use fluorescence microscope to observe the bacteria under 100X objective lens. From these fluorescent images, we find that fluorescence of sfGFP1-10 is stronger than the fluorescence of sfGFP11 and both of them are weak. It corresponds with our expectation that either of separate part of sfGFP have part function of complete sfGFP and sfGFP1-10, which is longer, may be brighter after excitation.

However, when the pSB1C3 carrying the part of sfGFP11 and the pSB1A3 carrying the part of sfGFP1-10 ware expressed together in E. coli , it doesn't present stronger fluorescence than either plasmid is expressed in E. Coli respectively. the possible reasons are as follows. Firstly, different metabolic stress of two plamids causes indistinct results. The pSB1C3 carrying sfGFP11, whose gene length is shorter, may have higher expression level than pSB1A3 carry sfGFP1-10. Secondly, there is obvious fluorescence quenching after excitation. Therefore, results of fluorescent images and fluorescent images appear that co-expression of sfGFP1-10 and sfGFP11 has weaker fluorescence intensity. Thirdly, because sfGFP1-10 and sfGFP11 may as well not be expressed in E. Coli at 1:1 ratio, the collision probability is lower than our expectation.

2.The spilt sfGFP experiment in Yeast(S. cerevisiae,W303)

Firstly, we transformed the two kinds of plasmids into the S. cerevisiae W303 to get three types of S. cerevisiae respectively containing sfGFP1-10, PR-sfGFP11 and both. Then we cultivated the cells at 30 ℃ for 24 hours.

(1) Testing the effect of heat shock

We measured fluorescence intensity of two groups of S. cerevisiae cultivated separately at 30,34,38,42 ℃ without GdnHCl. The first group was cultivated for 1 hour and the other one for 2 hours.

Here are original images of the two groups’ fluorescence experiment.

By using Image Processing method described in the appendix, we got these two diagrams.

The data of the points above is showed below.

	Par1	Par2	Par3	Par4	Par5	Average
30°C	169.0165	160.7907	173.7442	177.6913	177.0472	171.658
34°C	167.8331	164.8636	133.7497	153.8358	143.118	152.68
38°C	121.4902	120.289	105.2324	124.6197	132.0618	120.7386
42°C	116.3674	107.4996	112.4971	114.5681	106.1929	111.425

Heat Shock 1h

	Par1	Par2	Par3	Par4	Par5	Average
30°C	153.2803	116.601	131.8142	122.8715	165.9979	138.113
34°C	110.8943	123.2279	106.1407	101.0241	83.48351	83.48351
38°C	116.4582	107.2333	122.945	95.81423	95.33545	107.5572
42°C	118.5441	104.3932	94.09721	94.72964	107.0703	103.7669

Heat Shock 2h

As shown in the Figure 1 (2 hours of heat shock) above, the brightness of the bright dots in the photo, which represents the level of sfGFP 1h after heat shock, decreases almost linearly as the temperature increases. According to the linear fitting, the brightness drops about 5 units for the increase of temperature of 1°C, which is about 3% of the level at 30 degrees Celsius. In the modeling part, we predicted a linear decreasing relationship between temperature and the sfGFP output level. So it coincides with our experimental results.

It is noteworthy that high temperature is likely to affect the binding of sfGFP1-10 and sfGFP11, and destabilize the sfGFP complex. So we must exclude the possibility that the result in the figure above can be only attributed to this factor, instead of the aggregation of Sup35. According to Zhang et al, from 30 degrees to 42 degrees, the fluorescence intensity decreases for about 20%, owing to the effect of high temperature. However, in our experiment, it decreases with 35%, which is obvious more than the effect of temperature only. So there must be another mechanism, which should be that aggregation of Sup35 blocks sfGFP11 and make it impossible or at least harder to bind with sfGFP1-10. What’s more, the difference of the decreasing ratio, 15%, is precisely identical to the predicted ratio in our modeling result. Thus we can safely draw the conclusion that the sfGFP level decreases with increasing temperature, due to or at least partly due to the aggregation of Sup35.

As for the Figure 2 (2 hours of heat shock), it shows that there is a huge decrease of fluorescence intensity when temperature increases from 30 ℃ to 34 ℃ while no obvious changes occur when temperature varies from 34 ℃ to 42 ℃. That is not fully in line with expectation. The possible explanation is that heat shock indeed causes the aggregation of Sup35 and relatively higher temperature can enhance the aggregation effect. But there is another important factor you have to notice, which is that when Sup35 is in prion state (non-prion state of Sup35 nearly don’t form aggregation), it can propagate and transform the normal Sup35 protein into its prion state. And 2 hours is so long for the cells to finish the form changing process and aggregation of all the Sup35, regardless of the little differences in the temperature of heat shock. In conclusion, the latter three points in diagram 2 have similar values because the aggregation of Sup35 had come to saturation. Now you might want to ask why group 1 doesn’t show this contradiction. The reason might be that 1 hour is not sufficient for process of injection and aggregation to finish and during that time temperature has a dominant effect on the aggregation of Sup35.

In a word, from the temperature experiments above, we provide valid evidence which showed that our system could work effectively!

(2) Testing the effect of GdnHCl

We measured fluorescence intensity of S. cerevisiae cultivated at 42 ℃ in 0.25 mM, 0.50 mM, 3 mM, 5 mM GdnHCl for 4 hours.

By using Image Processing method described in the appendix, we got this diagram.

According to statistics analysis, with the increase of concentration of GdnHCl, fluorescence intensity increase in the first stage, and then decrease. In the first stage, fluorescence became brighter because Sup35 disaggregated and our kill switch turned on again. In the latter stage, the fluorescence intensity decreases which was opposite to our modeling results. We assumed that GdnHCl, which possesses high electric charge, may lead to the misfolding of split sfGFP. It is likely to disturb the two fragments assembling and reconstituting. Thus, GdnHCl is a suitable curing reagent for Sup35 in S. cerevisiae within limited concentration range.

Experiment Design

For the Propri-ontein system we measured the fluorescence intensity of the S. cerevisiae containing the plasmid carrying pGAL+GFP without the activation of AD and BD.

For further experiment of the Propri-ontein system, we can compare the fluorescence intensity of the S. cerevisiae containing all of the three plasmids (AD, BD and pGAL+GFP) with that of the S. cerevisiae containing only pGAL+GFP. The ideal results are that the fluorescence produced by the S. cerevisiae containing AD, BD and pGAL+GFP will be much brighter than the S. cerevisiae containing only pGAL+GFP after heat shock. And if we add GdnHCl to the system, the fluorescence will decrease because it can disaggregate the protein so that AD and BD can no longer bind together, which can eliminate the expression of the downstream gene under the control of pGAL.

Experiment Results

The pGAL-GFP experiment in S. cerevisiae

For the pGAL-GFP device, we conducted several tests in S. cerevisiae.

Firstly, we transformed the recombinant plasmid containing pGAL1-GFP into the S. cerevisiae W303. Then we cultivated the cells at 30 ℃ for 24 hours. Next we measured the fluorescence of the cells under the same condition. (We used fluorescence microscope to observe cells under 100X objective lens.)

Here're the results of the fluorescence experiment.

Compared with the results from spilt sfGFP experiment, the fluorescence intensity was extremely low. And that is exactly what we expected. So we proved that the pGAL-GFP device really functions !

As for the other devices in propri-ontein system. We tried to transformed the three plasmid-AD, BD and pGAL-GFP to S. cerevisiae. Unfortunately, we didn’t get the colony on our plates. The possible reason is that transformation of three kinds of plasmid at the same time brings too much pressure to the cell, which can cause the death of cells. It was really a pity that we had to stop the experiment due to the time limitation. But we will finish the verification of the whole system in the near future.

For the original photos we got, we handled all of them in the same way. The image processing method is showed here.

Take one photo for example. This is an original photo of fluorescence experiment(Image 1).

We use the mathematical software Matlab to calculate the brightness of fluorescence according to how bright the pixel is. Then we convert the photo into a 3D plot where XY surface denotes the location of pixel and Z axis represents the brightness, which looks just like this (Image 2)

Then we converted it into a contour line where same value would show in the same curve, which is shown below (Image 3)

In this picture, different color indicates different brightness value. Then we reverse the contour figure to make it coincide with the original figure. (Image 4)

Then we only select the bright part of the figure, and omit other parts

Finally, we calculate the average brightness value of the illuminating area so that the value can represent the quantitative and precise fluorescence intensity of the original photo

All of analysis above is processed with Matlab. You can find the original code by clicking this hyperlink .