Team:Montpellier/Demonstrate
RESULTS
Our starting hypothesis is that one can add specificity to a protease by fusing it to a target-specific antibody. To test the idea, we designed a simple experimental in vivo assay in E.coli. The KARMA tool for our proof of concept is composed of two parts: the recognition part (the anti-GFP VHH antibody) to maintain an effective activity against the targeted GFP reporter and the scissor part (the TEV protease) to cut the recognition site.
Comparing the activity of KARMA to that of TEV alone
In order to monitor TEV activity in vivo, we designed two strategies but focused on the ssrA removal assay : a sfGFP reporter was designed so that it increases the cell’s fluorescence when cleaved by TEV. This reporter is composed of a sfGFP fused to a proteolysis tag (ssrA), and in between, we placed a TEV cleavage site (TEVcs). For faster reading and comprehension, we will not always show the TEVcs on all the reporters but it is present.
When expressed alone in the cell, there is no TEV activity and the ssrA tag of the reporter is recognized by the ClpX endogenous protease of E. coli. This protease degrades the fluorescent reporter. Therefore the sfGFP’s half-life is very short and we expect very low fluorescence.
We then expressed this reporter along with the TEV protease or KARMA (TEV-VHH). In these cases, there is TEV activity in the cell, and the ssrA tag is cleaved from sfGFP. ClpX can no longer recognise the reporter and sfGFP is not degraded anymore. We should, therefore, observe an increase in fluorescence intensity. By comparing the fluorescence gains caused by TEV alone and KARMA, we can determine which one has a higher cleavage activity.
Determining target specificity
Finally, we performed a specificity assay by replacing the sfGFP in the reporter by mRFP1, that should not be recognised by the VHH. Thus, any increase in red fluorescence with KARMA compared to TEV is a measure of nonspecific cleavage in this case.
We monitored TEV activity by measuring green and red fluorescence signals in both a plate reader.
Here we experimentally show that fusing the TEV protease to a VHH increases the overall degradation rates of the VHH’s target as compared to using the TEV protease alone. We also tried to reduce the protease activity by modifying the temperature to increase the differential effect between TEV and KARMA. Finally, we tested the specificity of our tool by measuring its activity against a non-target protein.
Experimental implementation
Below is a description of how we set up the system experimentally.
The protease generator is under the control of the pTet promoter responding to anhydrotetracycline (AtC). The construct contains three coding sequences in fusion: MBP, TEV, and VHH. The TEV protease that has a N-terminal His tag is fused to a MBP to improve its folding. Both proteins are separated by a TEV cutting site (so MBP would be cleaved out by the TEV moity). TEV is fused to the VHH recognizing GFP by a GGGGS linker.
The reporter is under the control of the Arabinose promoter (pBAD). It’s a sfGFP linked to a ssrA by a TEV cutting site. Both constructs (protease generator + reporter) were transformed into E. coli strain NEB 10β to perform fluorescence measurement.
As a control, we used a construct encoding MBP-TEV with no VHH targeting GFP.
Growth curve
To ensure reproducibility, KARMA production should be induced at a given bacterial density. To do so, we determined the growth curve by measuring the evolution of the OD at 600 nm.
Our PIs advised us to induce the production of KARMA in this strain between 0.3-0.6 at OD600 if measured in a standard cuvette (1cm large). Since we measured the absorbance in a microplate we standardized our values with the pathlength correction. The absorbance values that we will show in our graphs will, therefore, correspond to values measured in a standard cuvette.
We can observe that we reach 0.3 and 0.6 OD at approximately 4h20 and 4h55 respectively.
It means that the optimal time for induction in a plate-reader is between 4h20 and 4h55, so we have approximately 35 minutes to induce KARMA production.
Determination of the optimal concentration of anhydrotetracycline (ATc)
Once we had identified the time window for KARMA induction, we wanted to know the optimal concentration of inducer to use. To induce KARMA expression, we used anhydrotetracycline (aTc), a tetracycline analog that has a much reduced antibiotic activity while maintaining the affinity for the Tet repressor.
We co-transformed cells with KARMA and our reporter, grew them and added increasing concentrations of aTc. We measured the variation in sfGFP fluorescence at 528 nm (Excitation: 485 nm) at 37°C. We expected that the higher KARMA production, the higher fluorescence intensity observed.
In our host laboratory, the optimal aTc concentration used to induce the pTet promoter ranged from 20 to 100 ng/ml of ATc. We decided to test 5 different concentrations: 2.5 ; 5 ; 10 ; 20 and 50 ng/ml of aTc. In this graph, we use NEB10β sfGFP-TEVcs with no proteolysis tag as positive control and NEB10β sfGFP-TEVcs-SSRA as negative control. The first one gives the maximal and the second one the minimal expected fluorescence intensities. The results are presented below.
To simplify the graph, we have only shown the values after induction with the aTc (more precisely after 15 min of induction which corresponds to the first measurement of the plate reader), that is why the curves start from different points. At the beginning (from 0 to 5h) the five concentration of aTc ( 2.5 ; 5 ; 10 ; 20 and 50 ng/ml) seems to have the same effect on KARMA production. After 5 hours, we observed that aTc concentrations of 20 (in yellow) and 50 (in green) ng/ml produce the highest increase in fluorescence intensity.
Determination of the optimal concentration combination of ATc and arabinose
Because our reporter is under the control of the arabinose promoter, we tried multiple combinations of arabinose and aTc to see which concentrations would be the best to express our proteins while avoiding lethality for our bacteria. To do so, we tested arabinose concentrations of 0,1% ; 0,2% ; 0,5% and 1% in combination with the concentrations of aTc 10, 20 and 50 ng/mL. We found that the optimal conditions were 30ng/mL of aTc and 1% of arabinose and used these conditions for our experiments.
Major result #1 : Karma improves TEVcs proteolysis compared to TEV protease alone
We then started to measure the fluorescence of the sfGFP-TEVcs linked or not to the ssrA tag to establish reference GFP intensities corresponding to respectively the minimum and the maximum cleavage efficiency of the :TEVcs. To do so, and for all fluorescence analysis, we used two plate readers from our lab : one for the measurements performed at 30°C and one for the ones performed 37°C. The first temperature is the one at which the TEV protease works best. The second one was chosen to simulate a reduction of the protease efficiency.
We will first show our results then do a statistical analysis of those. Each point on the graphs represent a replicate from experiments, and the bars represent the average of those points.
We have a different number of measure because we lack growth for bacterias in some experiments.
Here, we can see the expression of our different proteins at 30°C. The sfGFP gives a basal fluorescence going at 220 000 GFP/OD on average that goes down to 82 000 when an ssrA is added. So, we have 2.7 times more fluorescence when ClpX can't degrade sfGFP.
When we add a plasmid coding for the TEV protease, the fluorescence measured goes to 144 000, 1.8 fold the value we got from the ssrA version of our reporter. This means that the TEV upgrade the fluorescence to a total of 65% of the sfGFP fluorescence. So the protease recognize the cutting site between the sfGFP and the ssrA. These values will be used as a basal activity from the TEV protease.
When we express KARMA instead of protease alone, we have a fluorescence of 198 000 GFP/OD, 2.4 times more than sfGFP with a proteolysis tag. This value represents 89% of the glowing we got from the sfGFP alone, 25% more efficient than the TEV. So, adding a VHH to a protease does help this one on his efficiency.
When we look at sfGFP fluorescence at 37°C, we can see that adding a ssrA to the reporter gets the value down from 117 000 to 16 000. This difference means that the sfGFP is glowing 7 times more than its ssrA tagged version so, even at 37°C, the ssrA tag is functional. In fact the ssrA tag is more active at 37°C as already described before (see part M0050 page).
If we take a look at the value when the bacteria also synthesize TEV, we observe an increase in fluorescence intensity up to 47 000 GFP/OD, 2.9 times more than sfGFP with a proteolysis tag, 42% of the maximum we could get. That means that the TEV is still working at 37°C.
When the TEV is linked to a VHH, we get to 72 000, 4.5 times the value we had from the reporter with the ssrA tag and 77% of our maximum fluorescence. Compared to the protease alone, KARMA upgrade the recovery of sfGFP by 33% . Even at 37°C, we still have a better proteolysis with KARMA than with the protease alone.
Below is a table summarizing the normalized fluorescence data of all constructions. At each temperature, we used the fluorescence intensity of our positive control sfGFP-TEVcs as 100%.
| 30°C | 37°C | |
| sfGFP-TEVcs-SSRA | 36.6% | 14.7% |
| TEV + sfGFP-TEVcs-SSRA | 64.8% | 41.9% |
| KARMA + sfGFP-TEVcs-SSRA | 89.2% | 77.3% |
| Karma improvement (Karma to TEV ratio using GFP/OD) | 1.4 Fold | 1.6 Fold |
Table 1: summary of the normalized fluorescence intensities in the different conditions.
Both at 30°C and 37°C, fusing VHH targeting GFP to TEV does increase fluorescence intensity. These results suggest that the VHH helps the protease to reach and cleave the target resulting in an improved proteolysis efficiency. Yet, here because the values were measured in two different plate readers, we can't compare directly the efficiency from both temperatures using the raw data. Plus, the difference of fluorescence between the sfGFP-TEVcs and his ssrA version is clearly different at both temperatures: the proteolysis tag reduces by 2.7 times the value of fluorescence at 30°C and 7 times at 37°C. We can hypothesize that other elements are altered when we change the temperature..
Statistical Analysis
To have an idea of the significativity of those results, we compared them with a two-sample t-test or a Wilcoxon test for KARMA vs TEV at 30°C because we couldn’t approximate the values to a normal distribution. In order to compare results from different experiments and conditions as much as we could, all fluorescence intensity measurements were normalized by using our positive control sfGFP-TEVcs as the 100% value. We will only show the p-values from each test we made. To say that test shows a significant difference, they need to be at least lower than 0.05. The p-value is as follow in the table. We compare the combination as written at each temperature. We compared the TEV+sfGFP-SSRA to sfGFP-SSRA to show that the activity of TEV has a clear impact on the measurement and that the difference isn’t due to stochastic events. We also compare KARMA+sfGFP-SSRA to the TEV+sfGFP-SSRA to show the significant difference between both activities.
| KARMA+sfGFP-SSRA | sfGFP-SSRA | |
| TEV+sfGFP-SSRA at 30°C | 0.029 | 0.0073 |
| TEV+sfGFP-SSRA at 37°C | 7.5e-05 | 0.00056 |
Table 2 : p-values for sfGFP constructs calculated
For the comparison of sfGFP-SSRA to TEV+sfGFP-SSRA, we can see that the TEV did have a significant part on the values. When we look at the p-values between KARMA and the protease alone, we can see that both of them are under 0.05. We can conclude from this that we do have a better activity when we add a VHH targeting a place close to the protease targets.
You will find below the plots with the p-values as images for each temperature. The graphs dots are represented in a jitter manner to add noise.
Statistical graph of sfGFP values at 30°C Statistical graph of sfGFP values at 37°C
Major result #2 : KARMA is specific for GFP
After proving that KARMA did help the protease to reach the target of interest, we wanted to show that this improvement in efficiency was specific for the chosen target and not due to an intrinsic improvement in activity conferred by the VHH. For example, isn’t the VHH just making the protease more stable, increasing its concentration inside the cell.
We thus tested KARMA targeting sfGFP using a reporter based on mRFP1, which is not recognized by our VHH.
From those experiments, we expected no significant difference between the fluorescence we would get from the action of the TEV and KARMA. We also made two constructs for the mRFP reporter: one with a TEV recognition site and one with a ssrA tag after the TEVcs.
The measures were also made at 30 and 37 degrees Celsius. Because sfGFP and mRFP are different proteins with different excitation and emission length, we can’t compare both on their Fluorescence/OD value. But we can compare the activities of KARMA vs TEV. Each point on the graph represents a replicate and the bars represent the average of them.
Here, the mRFP-TEVcs represent the maximum of fluorescence we could get and the mRFP1-TEVcs-SSRA the minimum. The first one gives a value of 1 089 RFP/OD while the second produced by 6. This means that the ssrA proteolysis tag also work for the mRFP, reducing by 99% the fluorescence we could get. Actually, the ssrA is very efficient in targeting mRFP to degradation, giving us really low fluorescence intensity values. More generally, the fluorescence value without a proteolysis tag is really low.
Cells expressing TEV exhibited an increase in fluorescence to 51, meaning that the TEV does recognize his target site, producing a 10 fold change compared to the negative control, with a fluorescence of 5% of the maximum. If we use KARMA, the fluorescence will stand at 134 RFP/OD, 10% more than what gave us the TEV.
Just like for sfGFP, we wanted to see if the activity of the protease was lowered on the mRFP at another temperature, to test the reduction of efficiency on a different protein than the target of interest.
The mRFP-TEVcs had a fluorescence of 440 RFP/OD that was reduced by 95% (28 RFU) when we added a ssrA at the end of it. This means that even at 37°C and on the mRFP, the ssrA (linked by a TEVcs) will downgrade the measurement of fluorescence.
When we add the protease alone, the value stays at 27 with a lot of variations, representing only 5% of the maximal fluorescence we could have got. The difference with the values at 30°C are too low to compare between both because of the differences the two plate reader could give.
With KARMA, the values are going at 10% of the mRFP-TEVcs fluorescence which is equal to 47. Just like for the protease alone, we can’t compare between the two temperatures but we can still see, like at 30°C, a higher recovery of the fluorescence with KARMAthan the protease alone by 5%.
Statistical Analysis
Because the values are really close to each other, the statistical analysis is more than important to see if, indeed, KARMA is working more on the mRFP than how worked the protease. To calculate that, we used a Welch two sample t-test (because the variances were unequal) except when we compared the activity of the TEV with the ssrA version of our reporter for which we didn’t have enough experiments nor could approximate to a normal distribution. To compare correctly, we used mRFP-TEVcs as a value to normalize all the other results. Reminder : only the p-value from the test we made are shown, and to conclude in a significant difference they need to be at least lower than 0.05. Here is a table with our p-values. We compare the combination as written at each temperature. We compared the TEV+mRFP-SSRA to mRFP-SSRA to show that the activity of TEV have a clear impact on the measurement and that the difference isn’t due to stochastic events. We also compare KARMA+mRFP-SSRA to the TEV+mRFP-SSRA to show the significant difference between both activities.
| KARMA+mRFP-SSRA | TEV+mRFP-SSRA | |
| mRFP-SSRA at 30°C | 1.6e-06 | |
| mRFP-SSRA at 37°C | 0.093 | |
| TEV+mRFP-SSRA at 30°C | 0.16 | |
| TEV+mRFP-SSRA at 37°C | 0.064 |
Table 3: p-values for mRFP constructs calculated
For the comparison between the values of TEV+mRFP-SSRA and of mRFP-SSRA, we can see that at 30°C the difference is significant. For the difference at 37°C, the p-value is higher than 0.05, meaning we can’t say that the differences are significant. We did in depth study about that and took a look at the power of the test for these values. The power was at 0.42, meaning we had 58% chances to get a p-value above 0.05. This doesn’t mean that the difference is significant or not but only that we would have needed more experiment to have a better understanding of that.
Even though we had 5 to 10% difference of fluorescence between the strains transformed KARMA and the TEV, we can say now that they are not significant. These data suggest that contrary to the sfGFP target, KARMA has a similar activity than TEV alone on the mRFP1-target. However, because the recovery of fluorescence is very low, we would need to optimize and repeat this experiment to strengthen our claims.
You will find below the plots with the p-values as images for each temperature. The graphs dots are represented in a jitter manner to add noise.
Statistical graph of mRFP values at 30°C Statistical graph of mRFP values at 37°C
Conclusion : KARMA principle works
In all, because KARMA worked more than the TEV on the sfGFP, our results suggest that an antibody part fused to a protease will add efficiency to this one by reaching for a target that is close to the enzyme’s recognition site. Also, thanks to the experiments on mRFP, we could say that KARMA doesn’t have more activity on the mRFP than the protease alone. Of course, more experiments would be needed to confirm those results because, with only 3 month of manipulation, we didn’t have enough time to do additional one. Also, we did find some issues in our experimentations. First, the sfGFP has features in the DNA sequence that mRFP doesn’t have (notably insulators), meaning that the first one is highly produced compared to the second one. This prevent us from comparing the % efficiency of both protease type on either reporters In addition the sfGFP reporter seems toxic for the bacteria and alter their growth rate. In fact, the OD measured for the sfGFP is really low with or without a proteolysis tag compared to the mRFP which has normal OD compared to our controls NEB10β. According to previous work[1], a too strong overexpression of ssrA tagged proteins could saturate the degradation system. Also, using another temperature didn’t only change the TEV activity. Indeed, we found that the ssrA system is more efficient at 37, than at 30°C, as described before (BBa_M0050), probably because at this temperature other proteases than ClpX can also recognize and degrade ssrA-tagged proteins. This means that we reduced the TEV activity while adding efficiency to our degradation system.
The last point is that we used a minimum medium for the plate reader experiment so LB doesn’t alter the fluorescence measure. To summarize, here we have a highly produced protein that will overload the degradation system while using all of the low amounts of resources the bacteria has. But when we compare the OD of bacterias transformed with sfGFP-ssrA, sfGFP-ssrA+TEV and sfGFP-ssrA+KARMA, the bacterial growth is increasing in this order. It may be because ClpX has shorter proteins to degrade, recycling the resources faster. To have a better result, the use of rich medium instead of minimum one and mutant protease of TEV to avoid random changes the temperature gave us would be the best options. For the reporter, we would need to reduce the expression of sfGFPs construction and express more the mRFP. Actually, for this one, we could also switch for a different reporter because of its characteristic making it not the best choice. We could either change the type of fluorescence (YFP) or stay with the red yield (a better choice being mCherry [2]).
KARMA is correctly produced in E. coli cells
As a control, we wanted to assess if our proteins were correctly produced in bacteria by doing a Western-Blot. Because TEV is fused with an MBP to have a faster folding, we used an antibody targeting the MBP to see if it is correctly cleaved out of KARMA. KARMA has a His-tag in C-ter (after the VHH) and N-ter (before the TEV), so we detected KARMA using an anti-His antibody. As a negative control, we used a lysate of an E. coli NEB10B that does not contain any plasmid and placed it in the second well of each Western-Blot, placing the ladder in each first well. The size expected for both MBP and KARMA is close to 43 kDa. To check if KARMA was correctly produced and cleaved out of the MBP, we made multiple replicates.
A. Western blot (direct marking) with a mouse anti-his, then revealed with a conjugated HRP anti-mouse antibody. KARMA without MBP has a size of 43.5 kDa
B. Western blot (direct marking) with a mouse anti-MBP then revealed with a conjugated antibody HRP anti-mouse. The MBP alone has a size of 43.4 kDa
From the first Western-Blot, we got all the bands between 37 and 50 kDa, just like we would expect from KARMA to be, at 43.5 kDa. This means that our protein of interest is correctly synthesized by the bacteria.
From the second Western-Blot, we also got bands between 37 and 50 kDa, the size of MBP. This means that MBP is well produced and separated from our protein, else it would have been at 90 kDa.
We can conclude from both experiments that KARMA is correctly produced by the bacteria and that TEV correctly catalyze its own cleavage from the MBP.
[1] Cookson, Natalie A et al. 2011. « Queueing up for Enzymatic Processing: Correlated Signaling through Coupled Degradation ». Molecular Systems Biology 7(1): 561.
[2] Borovjagin, Anton V. et al. 2010. « Noninvasive Monitoring of MRFP1- and MCherry-Labeled Oncolytic Adenoviruses in an Orthotopic Breast Cancer Model by Spectral Imaging ». Molecular Imaging 9(2): 7290.2010.00003.