Team:Thessaly/Results
RESULTS.
We conducted experiments to validate our designed primer sets amplification performance both in standard PCR and in the RPA reaction. After some rounds of optimization, we were able to amplify our biomarker (IS6110) in both types of reaction. Subsequently, we tested the functionality of our reporter system, initially using a well-characterized toehold switch from the literature and then with a newly designed toehold.
DNA amplification
We first tested our designed primers in PCR reactions, as this is the standard amplification method and thus more likely to succeed. Our goal was to choose the correct reaction conditions (primer combinations, PCR additives, etc.) to be used subsequently in RPA reactions.
PCR
Testing our overhangs
Our experimental design requires addition of overhanging sequences to the 5’ ends of our primers. We tested whether the addition of these overhangs affects the capacity of our designed primer sets to amplify the chosen biomarker sequence. As depicted in Figure 1, the amplification was successful, both with and without overhangs. Primers with overhangs yielded a 136bp sequence, while primers without overhangs resulted in a 80bp sequence.
Figure 1. PCR amplification of our DNA biomarker both with and without the 5’ universal overhanging sequences we designed.
Primer screening
After proving that primers with 5’ overhangs can amplify our biomarker in a PCR reaction, we needed to validate a number of primer pairs for optimal functionality. We screened a total of 9 primer combinations, which yield different amplified product lengths (136, 306 and 556 bp) and different annealing lengths for the reverse primer (32, 36 and 39 bp annealing to the template). With this reaction we sought to test those different primer combinations and their probing efficiencies. This is crucial, because according to the literature and our project design, the biomarker DNA templates we aim to amplify are generally 150 bp long[1].
As depicted in figures 2A and 2B, all primer combinations tested can amplify the target biomarker. Combinations F1-R1, F1-R2 and F1-R3 give 136bp amplicon, while the F2 series gives the 306bp and the F3 the 556bp, respectively. Based on these results and because a short amplicon was desirable, we chose the first forward primer(F1) depicted above with the second corresponding reverse primer (R2), which according to our in silico analysis is the best of the three reverse primers.
Figure 2. A) Primer screening of 3 primer combinations with the 1st Forward primer, which correspond to ~136bp amplified product lengths and different annealing lengths for the reverse primers (32, 36 and 39 bp annealing). B) Primer screening of 6 primer combinations with the 2nd and 3rd Forward primer, which correspond to ~306 and 556bp amplified product lengths and different annealing lengths for the reverse primers (32, 36 and 39 bp annealing).
Sensitivity test and incorporation of PCR additives
Having chosen the best primer pair to be used in our proof of concept reactions, we needed to further validate its suitability and sensitivity of detection at a range of template mimicking biologically relevant levels of the target biomarker. To this end, reactions took place using two different template concentrations, resembling to 1 and 10 copies per reaction, respectively. In Figure 3A, we can clearly see that the selected primers can produce a specific band of substantial intensity, even when 1 copy of the biomarker sequence is present. Concurrently, we checked our reaction’s performance after the addition of a PCR additive, DMSO, as it is known to enhance the efficiency of amplification reactions (Fig. 3B). Indeed, we were able to correctly amplify our biomarker in both reactions, while proving that we can amplify as much as a single copy of our biomarker with our designed primers.
Figure 3. A) PCR amplification of our DNA biomarker IS6110 in the lowest concentrations of our DNA template. B) PCR amplification of our DNA biomarker IS6110 after the addition of DMSO at a final concentration of 5%.
RPA
Having a validated primer pair and identified a range of suitable reaction conditions by PCR, we continued with our main objective. That is, testing whether RPA is capable of giving similar results to PCR and of detecting the target biomarker in conditions similar to those that might be found in biological samples.
We examined a range of different reaction conditions, including those tested previously by PCR reactions in order to overcome technical issues with RPA, including non-specific amplification in our reactions. An example of our initial RPA reactions is shown in Figure 4. Even though RPA is optimized to work for amplicons up to 500bp, we were unsuccessful when trying to amplify the 500bp amplicon primer set. Moreover, when no template was added, non-specific amplification occurred. We hypothesize that this is caused due to the formation of secondary structures of the primers, since their length lies within the 60-80bp range.
Figure 4. RPA test reactions for some primer combinations.
Testing different reaction conditions
Even though RPA is a great method to amplify and detect DNA in a very fast and relatively cheap way, according to the literature[2], its molecular mechanism comes with some intrinsic shortcomings that give rise to potential problems. More specifically, because an RPA reaction is carried out well below the minimum annealing temperature of all dsDNA molecules, this results in frequent primer-dimer formation and presence of several RPA artifacts, along with the desired amplified sequence. Thus, we concluded that it is possible that our chosen primers, due to their length and perhaps additional physicochemical characteristics attributed to their specific sequence, form primer-dimer formations. This in turn, leads to one or more non-specific products being visualized both in test reactions and in negative no-template control reactions. Wanting to further test our reactions under different conditions and check if this will help alleviate the possible secondary structures that may form in our reaction, we tested different reaction times and different temperatures for the PRA with our chosen primers.
Although still we could see an amplified product in the NTCs, we validated that the RPA reaction can amplify our biomarker even in a 5-minutes reaction (Fig. 5A). The standard RPA protocol includes a 20 minute reaction, while there are reported successful incubation times of 3 minutes. We tested a range of incubation times within these limits. As seen in Figure 5A, amplification can be witnessed until the 5min lane on the agarose gel. Moreover, as seen in Figure 5B the reaction shows higher efficiency when the incubation temperature is at 42˚C. This is logical, since the kinetics of the reaction are proportional to temperature.
Figure 5. A) RPA in different reaction times. B) RPA in different reaction temperatures.
Chosen conditions
In order to overcome these issues, we continued our tests so that we can achieve the optimal conditions for our reaction, without any non-specific amplification. Our next step was to combine the 5-minute reaction time and the 42˚C reaction temperature, based on the rationale that the shorter reaction time in combination with a higher temperature could alleviate the possible secondary structures of our reaction. This took into account our in silico analysis, which predicted that possible secondary structures should not appear above 42˚C. At the same reactions we also tested serial dilutions of our template DNA, to establish the sensitivity of the reaction.
This combination of changes to our approach resulted in the successful amplification of our target biomarker without any visible non-specific products in the negative control reaction (Fig. 6). We were also able to establish that the RPA reaction is able to amplify as little as a single copy of our biomarker with our designed primers, just like our PCR-based results, suggesting that our method is very promising in terms of sensitivity.
Figure 6. RPA reaction for the detection of IS6110 with serial dilutions of our DNA template, in a 5 min and 42 degrees Celsius reaction.
After successfully amplifying our target DNA through an RPA reaction, we needed to make sure that our successful result is reproducible. We performed another amplification reaction with the same conditions (Figure 7). This time we used a denser agarose gel (4%) with a low melt agarose (3:1), so we would be able to visualize any small differences in the sized of our amplified products.
As shown in Figure 7, we managed to correctly amplify our biomarker. Evidently, the 4% agarose gel allowed the fine separation of 2 different bands, which are around 10bp apart. We confirmed that the upper band is the specific product by PCR. It is thus clear that the detection of IS6110 is successful, even if 1 template copy is used per reaction. Also, according to existing literature it is documented that (RPA) is prompt to give non-specific products in low template concentrations[2]. This is confirmed, since higher-copy reactions tend to diminish the intensity of the non-specific band. We hypothesize that when there is a high copy-number of DNA template in the reaction, less primers are available to form structures that allow non-specific amplification.
Figure 7. Amplification reaction with serial dilutions of our DNA template
Testing the universality
To test the universality of our design we tested 3 new sets of primers, in which we only changed the annealing sequence of the primer, to be able to aim for the chosen DNA sequence of the HBV virus, while retaining the appropriate 5’ overhangs necessary to trigger our downstream detection module. We started testing our primer sets first through PCR, using all 3 sets, while using as a template the lowest concentrations we had created. The conditions were the same as those used for our proof-of-concept reactions, while we used DMSO at a final concentration of 5% in all of our PCR reactions.
Figure 8. Primer screening for HBV.
After successfully amplifying our second biomarker through PCR (Fig. 8), we established that the best designed primer set was the first one, thus it was used on our next amplification reaction which was an RPA reaction with the same conditions tested for our proof of concept reactions with our first biomarker (IS6110).
Figure 9. RPA reaction for the detection of HBV in the lowest concentration of our template DNA.
According to the results depicted in Figure 8 and 9, we concluded that we were able to amplify our chosen biomarker for the HBV virus, both with a PCR and an RPA reaction. We also attested that the amplification of this template DNA was more successful than that of the IS6110 , which further enhances our hypothesis that the high GC% content of the IS6110 gene is one of the main factors causing non-specific amplification. Still, RPA artifacts are present in the HBV reactions as well. However, since RPA artifacts are common and acknowledged in the literature and result in different lengths than the specific sequence, we hypothesize that they will not be an issue in downstream detection. This can be confirmed by sequencing the artifacts.
Simulating realistic conditions
After a series of exhaustive testing to optimize the amplification reaction, we still had not answered a vital question: Could we detect the biomarker in realistic conditions? To address that, we wanted to simulate the random fragmentation of the MTB genome when found in urine, so a fragmentation protocol was conducted. To ensure the non-specific manner and to simulate the conditions of the human body, a non-specific human endonuclease, DNAse I, was used. After a small series of calibrations to the DNAse I reaction, we managed to fragment the biomarker properly, as seen in Fig. 10.
Figure 10. DNase I fragmentation of IS6110 biomarker for MTB detection. 3min and 6min correspond to the incubation time of the template with the enzyme at 37℃.
The next step was to attempt an amplification reaction on the randomly fragmented template. We successfully amplified the desired sequence from different fragmentation conditions, as depicted in Fig. 12.This achievement holds great value, as we prove that our system would potentially work in clinical samples, where DNA is fragmented through a similar mechanism.
Figure 11. Amplification of randomly fragmented IS6110 biomarker successfully using PCR.
Outlook
Although we were able to amplify target biomarkers for two different diseases using PCR and more importantly RPA, some issues still remain with regard to RPA specificity and fidelity. Some future plans for us to test, include be the incorporation of primers with SAMRS nucleotides at the 3’ end. According to the literature[3], this is a promising strategy to eliminate non-specific amplification. This plan was initially part of our experimental design, however, the cost for the purchase of these primer sets was too high for an iGEM team. However, our results show that in principle we can amplify target DNA at very low concentration and from shredded low MW templated DNA.
In Vitro protein synthesis assay
In parallel with optimizing RPA-based target amplification, we began working towards the in vitro protein synthesis assay necessary for our detection system to function. For our experiments, we used the commercial PURExpress kit from New England Biolabs (NEB) , which is highly purified and reliable for a proof of concept experiment. The highlights of these experimental results are described below.
Cloning Of The Genetic Material
The first preliminary step towards testing our constructs, was to clone them into appropriate vectors for propagation. To do this, we cloned all necessary constructs in either pSB1K3, pSB1C3 of pSB1A3. As seen in Fig. 12, 13 most of the constructs are 900-1100bp, and thus appear around the 1000bp ladder marker. The trigger construct was the smallest and appears between the 100 and 200bp marker. The 2000bp band is the cut plasmid after the diagnostic digestion reaction. In figure 13, the image showing the cloning of β-lactamase represents a colony PCR reaction.
Figure 12. 32B trigger cloning. C=Cut plasmid, U=Uncut plasmid.
Figure 13. Cloning of 32BeGFP, eGFP, 32B β-Lactamase and β-Lactamase.
Calibrating the cell-free protein expression system
We calibrated our cell free system using of the genetic constructs described above. The aim was to initially examine the functionality of the cell-free system using non-regulated constructs and compare them with the results of the toehold-switch regulated constructs.
To do so, we incorporated two reporter proteins, eGFP and β-lactamase. β-lactamase is our reporter of choice for our current project, while eGFP is a common and well-characterized protein for easy visualization and quantification of performance that can be employed in other systems using our developed parts.
The first assay conducted was with eGFP as the reporter. To enable translation of the protein from the toehold regulated construct, different amounts of trigger sequence were added in the reaction. a reaction without a trigger sequence was also included as a negative control and a measure of background un-induced promoter activity. To reduce the cost of the reaction, we lowered the reaction volume from 25 to 7 μL.
After a 3-hour incubation at 37℃ in a thermocycler, protein expression was measured using a plate-reader. To measure eGFP, excitation at 488nm was used, while detection was at 515nm, the protein’s maximum fluorescence emission point. The results of the assay can be seen at the graph below:
Figure 14. eGFP fluorescence after in vitro protein expression
As expected, the non-regulated eGFP emitted the highest fluorescence, providing a positive control and confirming the protocol we used was functional. The toehold-regulated construct produced robust signal in decreasing concentrations of trigger (75nM, 15nM, 7nM). It is interesting that the signal remained the same even with a 10-fold decrease in trigger concentration, while the no trigger control had an almost 2-fold lower fluorescence. This indicates that there was strict regulation in the eGFP construct and higher concentrations of trigger sequences are needed for maximum activation.
Testing the system with β-lactamase
After ensuring the functionality of the system, we proceeded to test the β-lactamase constructs.
Following the same outline as before, we tested both a non-regulated construct, as well as a range of trigger concentrations and a background negative control. Since β-lactamase is an enzyme, it requires an enzymatic assay for the protein expression to be measured. To do this, after the 3-hour incubation in the cell-free system, we added a chromogenic substrate of β-lactamase, nitrocefin, and performed an additional enzymatic assay in a plate reader, at 37℃. Nitrocefin is yellow (with a maximum absorbance at 380nm) when not hydrolyzed, but upon hydrolysis by β-lactamase it converts to a red color (with a shift in maximum absorbance at 480nm). This provides a visual measure of β-lactamase activity that can also be quantified.
The results of the enzymatic assay are depicted in Fig. 15:
Figure 15. Enzymatic assay of β-lactamase with nitrocefin as its substrate, when expressed from a non-regulated and a toehold regulated construct in a cell-free system for 3 hours. Error bars correspond to standard deviation of n=2 replicates. Blank was subtracted.
The positive control Lactamase ensures that the reaction was functional. Unlike eGFP, the 75nM trigger reaction produced a similar signal to the positive control reaction, which is promising. However, when lowering the trigger concentration to 7nM, the signal was almost halved. This is interesting, since both eGFP and β-lactamase were regulated by the same toehold switch. This might imply that the reporter gene and its nature (e.g. enzyme) can affect the produced signal. The no trigger control was substantially lower this time, showing an almost 4-fold decrease in comparison with the 75nM trigger reaction.
The next goal was to assess the performance of the enzymatic assay, when lowering the trigger concentrations and the incubation time. First, we decreased the trigger concentration even further, up to 0.3nM. When incubating for 3 hours the PURExpress reaction, even the 0.3nM reaction produced ~2-fold more signal, compared to the no trigger control (Fig. 16).
Figure 16. In vitro transcription/translation of toehold-regulated β-lactamase with low concentrations of trigger sequence and quantification through a chromogenic enzymatic assay. Error bars correspond to standard deviation of n=2 replicates. Blank was subtracted.
This is promising, since the system can produce detectable signal with very low concentrations of trigger, thus increasing the sensitivity of our diagnostic test.
Since PURExpress’ incubation time is the most time-consuming part of our project, we intended to lower the incubation time and assess the performance of the cell-free system and its sensitivity. We performed 1-hour in vitro protein synthesis, with all the other conditions remaining the same. Again, a gradient of trigger concentrations was included to measure the system’s sensitivity. The results are shown in the figure (Fig.17):
Figure 17. 1-hour in vitro transcription/translation of toehold-regulated β-lactamase and quantification through a chromogenic enzymatic assay. Blank was subtracted.
To our surprise, the 75 nM trigger performed similarly to the 3-hour assay, while the 50 nM produced very similar signal. This might imply that those trigger concentrations are more than enough for reliable protein expression, since a 3-fold decrease in incubation time did not affect the signal, which reached a level comparable to the positive control. Also, both 15nM and 7nM trigger concentrations produced similar results to the 3-hour assay. However, the 3nM trigger concentration did not produce a strong signal, since it was not distinguishable from the no trigger control.
This implies that 7nM is the minimum trigger concentration per reaction, in 1-hour in vitro protein synthesis. This translates to ~500 picograms of trigger concentration DNA per reaction. In addition, a DNA amplification with PCR or RPA, produces 50-100nM of the engineered trigger sequence when 10-100 copies of initial DNA template is present respectively, as measured by our amplification team. Hence, it is safe to assume that a 1-hour in vitro protein synthesis reaction would produce a robust signal and provide reliable detection of the initial DNA template in a biological sample.
Using RPA-amplified sequence as the trigger
The goal of our project is to amplify – and at the same time engineer – a DNA sequence which acts as a trigger sequence for our toehold switch. Since both the amplification and in vitro protein synthesis steps were achieved, we sought to integrate them into one final experiment. In this setup, instead of expressing the trigger from a plasmid, we implemented the trigger sequence in the form of amplification product. In this way, we ultimately simulate the functionality of the system since the trigger sequence is linear, as it occurs after the amplification reaction.
In the final experiment, we followed the same working protocol in a 7μL reaction for in vitro protein synthesis of β-lactamase. The only key difference here, is that during the enzymatic assay we increased substrate concentration. This was done due to technical reasons, as in some previous experiments with low reaction volume (~50ul) the measurements from the plate-reader were sometimes inconsistent. To overcome this issue, we decided to increase the volume of the enzymatic assay to 100 ul, which resulted in much more consistent readings. Also, because of technical difficulties on purifying the amplicon from PCR/RPA in sufficient amounts, we added a synthesized sequence as a proof of concept, which is identical to the expected sequence from the amplification reaction.
After a 3-hour incubation in PURExpress, β-lactamase expression was measured in a 30-minute enzymatic assay. The results are depicted in Figure 18 below.
Figure 18. Measurement of toehold-regulated β-lactamase when the trigger is the previous step’s amplified sequence or in a plasmid vector. Error bars represent the standard deviation of n=2 replicates. Blank was substracted.
As seen in the graph, the produced signal from both trigger sources is closely similar, while a 5-fold increase is observed when compared to the no trigger control. We assume that the slight difference in concentration between linear and circular trigger is insignificant, since as shown before the system is saturated and fully activated when more than 50nM of trigger sequence are added.
After proving that both detection modules can work together, resembling a highly integrated system, we wanted to address one more thing. The toehold switch we had used so far was an established functional switch, while its trigger sequence derived from a Zika Virus strain (Toehold & trigger 32B,[4]). To avoid cross-talk and false positive results in the event of testing someone having Zika, we sought to address this issue by creating synthetic universal toehold switches.
Universal toehold switches
A key aspect of our project is to provide a tool able to detect the DNA of any organism easily and reliably. To this end, we in silico designed a pool of 15 universal toehold switches from hyperthermophile bacteria. Due to budget limitations, only the two most promising switches were ordered and tested according to the in silico analysis, Toehold 13 and Toehold 14. Both toeholds had similar thermodynamics and thus we expected similar performance in terms of protein expression regulation.
To assess the new toeholds’ performance, we performed a series of in vitro protein synthesis reactions, following the same protocol and conditions as before. The major issue with toehold switches is their background signal (translation in absence of trigger sequence), so we tested only one trigger concentration.
Initially we tested Toehold 14, which was less promising than Toehold 13 according to the in silico analysis. Our results confirmed the in silico prediction, as Toehold 14 was unable to control translation, compared to controls (Fig. 19).
Figure 19. Bars showing the performance of toehold 14 to hydrolyze nitrocefin.
As seen in Figure 19, the signal produced by Toehold 14 is very similar to that of the positive control. However, when no trigger sequence is added, the signal produced is the same as the 100nM trigger condition. This can be interpreted as the toehold 14 ‘s inability to control translation. That means that the toehold switch has an open conformation and thus an accessible RBS, even when no trigger is added.
Since testing toehold 14 resulted in a failure, we turned our interest to Toehold 13. Following the same conditions as before, we assessed the performance of toehold 13 in the same β-lactamase enzymatic assay. The results this time were more encouraging, as depicted in Figure 20.
Figure 20. Performance of toehold 13 in a β-lactamase enzymatic assay. Error bars represent standard deviation of n = 2 replicates. Blank was subtracted.
The expression of β-lactamase under the regulation of Toehold 13 was noticeably low in the absence of the trigger sequence, while the 100nM trigger reaction easily reached levels of signal comparable to the positive control. However, when compared to the literature 32B toehold[4], the background signal of toehold 13 was ~3-fold higher. This is sensible, since a more thorough in silico and in vitro screening was conducted in the paper, resulting in around 10 times more toehold switches that were tested.
Even if not perfectly functional, toehold 13 looks promising as a universal toehold switch. Furthermore, it is amenable to optimization both in silico and in vitro, something that we did not manage to do due to time limitations. We aspire that we or future teams will take the foundation we have provided and optimize this promising part into something high performing.
Future vision
Besides the above mentioned results and the goals we have already achieved, there are some aspects of our project we could explore more in the future, to improve it and eventually make it as efficient as possible.
SAMRS
Some future plans for us to test would be the redesign of our primer sets using SAMRS nucleotides, so that the RPA reaction would get no non-specific products amplified along with the specific one. Following reported literature on RPA-SAMRS primers, we have already re-designed our primer sets accordingly. This way, we hope to reduce the primer-dimer formation that we encountered in our non-SAMRS primers. This plan was initially part of our experimental design, however, the cost for the purchase of these primer sets was too high for an iGEM team.
Redesign our universal toehold switches
As we mentioned above, we have designeda pool of universal toehold switches from hyperthermophile bacteria in silico, but due to budget limitations, only the two most promising switches were ordered and tested, according to the in silico analysis. An easily achievable future plan for us, would be the screening of a larger pool of toehold switches (similar to the design principles of the ones we tested) and the assessment of their efficiency until we find the one that can provide us with the unique element of universality as well as with the highest possible efficiency.
Test the positive control
The nature of our design is such, that a positive control is needed for the correct functionality of our diagnostic test. This is why we chose the cox3 gene deriving from the human mitochondrial DNA, for a positive control, which is also included in our product design and will be incorporated into our final product. A future plan would be to actually redesign our primers to be able to target this gene and test if we can detect it and amplify it.
Synthetic urine
During the iGEM competition, we discovered that safety is of utmost importance. Due to this fact we did not use any urine samples to test the efficiency of our design. A future plan for us would be to spike synthetic urine with the fragmented biomarker we chose to target (IS6110 or HBV) and check if we would be able to detect it after a pre-treatment step of the sample. This way, we will be able to simulate, as much as possible, realistic conditions for our diagnostic test.
Cell lysate
In order to minimize the cost of our diagnostic test, we aim to create cell-lysates for the in vitro transcription/translation system. Cell-free expression systems can be prepared using either a cell extract, or a combination of purified recombinant proteins, and be done at 37 – 42℃ optimally. As it would be a very important upgrade of our design, it would be crucial for us in the future, to prepare our own cell-free expression system (cell lysate) and test its efficiency.
References
1.Labugger I, Heyckendorf J, Dees S, Häussinger E, Herzmann C, Kohl T A, Richter E, Milla E R, Lange C. Detection of transrenal DNA for the diagnosis of pulmonary tuberculosis and treatment monitoring, Infection (2016). doi:10.1007/s15010-016-0955-2
2. I.M. Lobato, C.K. O’Sullivan, Recombinase Polymerase Amplification: Basics, applications and recent advances, Trends in Analytical Chemistry (2017)
3.Sharma, N., Hoshika, S., Hutter, D., Bradley, K. M., & Benner, S. A. (2014). Recombinase-Based Isothermal Amplification of Nucleic Acids with Self-Avoiding Molecular Recognition Systems (SAMRS). ChemBioChem, 15(15),2268–2274. doi: 10.1002/cbic.201402250
4. Pardee Keith, Alexander A Green, Melissa K Takahashi, David H O Connor, Lee Gehrke, James J Collins (2016). “Rapid , Low-Cost Detection of Zika Virus Using Programmable Biomolecular Components ” Cell, 1–12. https://doi.org/10.1016/j.cell.2016.04.059.
