Team:Bielefeld-CeBiTec/Results/CFPS
CFPS
Molecular machinery at work
Motivation
Cell free protein synthesis has rarely appeared at iGEM in the recent years. This is opposed to quite many advantages (see here) CFPS has compared to cell-based approaches. We set out to establish CFPS protocols that can easily be performed by iGEM teams. We gathered literature, information and asked experts to be up to date in a research area that evolves at a grate pace.
Preliminary Experiments
The extract of E. coli contains the molecular machinery that is needed for in vitro transcription and translation. For optimal yield of ribosomes, cell harvest has to be performed at mid- to late exponential growth phase. We therefore started to measure growth curves for various strains at 100 mL scale. The strains we investigated possess a genomic coded polymerase of bacteriophage T7. T7 polymerase is a monomer, very specific to its promoter and more stable than bacterial polymerases (Sousa and Mukherjee, 2003), which makes it perfectly suited for in vitro transcription.
So...
we observed that induction of T7-polymerase expression at OD600 = 0.8-1.4 was not critical for growth kinetics. Regarding published data (Kwon and Jewett, 2015), we concluded that a cell harvest at an OD600 = 3-4 would be optimal. At this stage of growth, the cells would have a highly active translation machinery, but they would still be far away from stationary phase.
What to do with the cells?
To generate crude cell extract, the bacteria have to be disrupted. This can be achieved by different methods, for example bead beating (Sun et al. 2013) or homogenizing with high pressure (Yang et al. 2012). However, we decided for sonification. A recent publication from Kwon and Jewett showed that it is fast, cheap, reliable, easy to perfom and that it works with small volumes (Kwon and Jewett, 2015) – which altogether is indicative for the use of this disruption method at iGEM competition.
The parameter with the greatest impact on crude cell extract quality is the sonifiers energy output (expressed in Joule per second) (Kwon and Jewett, 2015). Insufficient energy does not lyse cells efficiently, whereas too much energy inactivates proteins. Although we had access to a sonifier, we faced a problem: The device did not display its energy output. However, we tackled this problem by measuring how the temperature of a water sample changed as a function of sonification time. With an equation that takes all important parameters into account we were able to correlate the sonification time needed for a desired energy output.
Poor cells... any survivors?
We immediately wondered how effective our sonifier was when it comes to the disruption of E. coli. We cultivated, harvested and washed E. coli cells. Then we sonicated them in cycles of 10 seconds. We took aliquots and plated them onto LB plates to get a first idea about efficiency, and we observed a decrease in cell viability with increasing sonification cycles.
In the same manner we took aliquots after the subsequent centrifugation steps according to protocol, and viable cell numbers were even lower. During our project, we continued to investigate how many cells survived the process. In our fully optimized extract, a 100 µL aliquot of flash-freezed crude cell extract contained only 8 colony forming units. Our final biosensor device did not contain any living E. coli at all, like you can see here. Although 8 colony forming units in crude cell extract are still too many in purposes of biosafety issues, we conclude that our sonification is extremely efficient in decreasing viable cell numbers and therefore counteracts a potential energy depletion source in our final application.
Our reporter
Most CFPS approaches are based on T7 polymerase and its cognate promoter (see background page). We used T7 polymerase for nearly all of our cell free expression experiments, which appeared to be a good idea: In tests with endogenous σ70-polymerase and non-T7 promoters we observed very low protein production (for details, see heavy metal result page).
An ideal reporter protein for in vitro protein synthesis is superfolder GFP, abbreviated sfGFP (Lentini et al. 2013). In our first experiments with selfmade E. coli cell extract, we used PT7-sfGFP from the parts registry. In a Fluostar platereader, we measured a 10 fold increased fluorescence when compared to the negative control after 7.5 h. This was the first time we realized that we made it; in vitro sfGFP transcription and translation was possible with our extract!
Template optimization
But we were not yet content with the results, we were sure that a further optimization was possible. By literature screening, we designed a translation enhancing sequence (5'-untranslated region, 5'-UTR) and inserted it into PT7-sfGFP, thereby creating PT7-UTR-sfGFP, BBa_K1758102. Our assumption was that if translation was a bottleneck in our extract, this sequence would improve sfGFP production.
This was the case in vivo: We observed a faster production of sfGFP when the plasmid DNA contained 5'-UTR in front of sfGFP coding sequence. Another reporter protein, mutated red fluorescent protein (mRFP) was also employed. This was done to check if the 5'-UTR was useful for other proteins as well. For mRFP normalization on OD600 we faced the problem that mRFP emits fluorescence at 607 nm (Lentini et al. 2013). However, results for sfGFP and mRFP equally demonstrate the usefulness of 5'-UTR for protein production in general.
This was the case when 5'-UTR was employed in in vitro experiments: We observed a more than 3 fold increase in fluorescence. This clearly showed the importance of this enhancing element in CFPS, and further demonstrated that translation efficiency in vitro is a major issue for protein synthesis.
Next steps
With these results, we had a good positive control plasmid for our following reactions. Nevertheless, we observed batch-to-batch variation in our extracts, a phenomenon also described in the literature (Takahashi et al. 2015). To exclude any batch-to-batch variation in following reactions, we set up a 5 L fermentation and harvested cell pellets to produce by far enough cell extract for the summer (details in the notebook).
We started using a Tecan platereader, constantly heating to 37 °C, to measure kinetics of sfGFP production. In this and every following performance we used at least triplicates for every reaction tested. We normalized on accompanied cell lysate from cells that had produced sfGFP. Measuring nearly every minute, we traced production of sfGFP in real time. The experiment once again showed the benficial effect of 5'-UTR.
Depending on E. coli strain and cell harvest, it can be necessary to perform a so called run-off reaction during cell extract preparation. We determined that 30 min of run-off reaction enable best performance for our extract.
During our project, we constantly optimized our extract. Our success is demonstrated by the following bar graph which shows how the fluorescence signal of our positive control in CFPS reactions increased during summer. Time issues held us back from further optimization, although we are confident that further optimization is still possible.
Furthermore, we tested extract from different E. coli strain cultures. Even when the bacteria carried plasmids, generation of functional extract was possible (see Heavy Metal results for example. However, each time the extract has to undergo some optimization steps. In particular, we observed that different strains' and extracts' performances are highly dependet on glutamate salt concentrations. This phenomenon is often described in the literature (for example in Sun et al. 2013).
Also the addition of certain molecules can be very beneficial for final protein yield. One optimization approach is depicted in the following section.
Optimization: Positive effector RraA
The product of the rraA gene (Regulator of ribonuclease activity A, former menG) has been reported to interact with RNase E from E. coli and to alternate its activity (Lee et al. 2003,Yeom et al. 2008, Gorna et al. 2010). Airen showed that if RraA-protein is added to a cell free protein synthesis reaction, the productivity raises about 30% in his cell free system (Airen, 2011). He postulated that activity of RNase E in the reaction is lowered due to the interaction with RraA-protein, therefore mRNA-levels are stabilized.
To characterize RraA in vivo, we created two E. coli strains. One carried the rne gene that codes for RNase E under control of the inducible T7-promoter, whereas a second strain carried an additional second plasmid with the RraA coding sequence. While there was no observable growth drop when T7 polymerase was induced in the latter strain carrying both plasmids, induction of T7 polymerase in the first-mentioned strain lead to a clear growth inhibition. This difference was apparent although the strain carrying rne-plasmid only grew slower than the double transformed strain.
Therefore we conclude RraA overexpression rescues E. coli by decreasing activity of RNase E.
Purified RraA that contained an N-terminal 6xHis-Tag was added to our CFPS reaction to see if RraA had the same positive effect in our setup.
We could verify Airens observation that RraA is a positve effector in cell free protein synthesis. More precisely, the results of Airens and our experiment are similar: In our reaction the final signal raised about 33.8 ± 4.5 %, and in Airens experiment the signal raised 28.6 ± 3.1 % respectively when compared to a reaction were RraA is not present (Airen, 2011).
The reason for this effect is investigated in detail in Airen, 2011. The shape of the fluorescence signal curve when RraA is present (green dots) indicates that the protein acts as stabilisator in our reaction. It is likely that by reducing RNase E activity, the rate of mRNA and rRNA degradation is slower.
If one is interested in obtaining high amounts of protein via CFPS, the addition of RraA – or similar positive effectors – is definetly recommended.
This experiment demonstrates how versatile CFPS can be. You can simply add an external molecule to the reaction and analyze its impact on protein yield. Although the reason why a substance is beneficial or disadvantageous may remain unclear, the experimenter has free access to the reaction. Less time is needed in this case when compared to in vivo optimization (Sun et al. 2013, Takahashi et al. 2015).
To summarize ...
We optimized our extract by adding RraA to the cell free reaction. The protein is useful in CFPS because it can inhibit the RNase E activity in the extract, which thus leads to a final output signal that is about 33% higher. The coding sequence of RraA and a protein generator to produce His-tagged RraA has been send to the parts registry (Biobricks BBa_K1758120 and BBa_K1758122, respectively). iGEM teams as well as labs that deal with E. coli cell free protein synthesis can employ these biobricks to optimize protein production.
CFPS on Paper
As depicted above, we established a robust cell free protein synthesis system for the production of sfGFP in our lab. The previous reactions were mostly performed in 364 well plates, so to test if our cell extract worked on paper we proceeded with using different papers as base for our reactions.
We started with a simple test: The CFPS reactions were prepared according to protocols at 15 µL scale, and were then applied on small paper discs with a diameter of about 6 mm. Using a 96 well plate, we tried to measure fluorescence, but no signal was seen.
We supposed papers were full of RNases and therefore quite a hell for our RNA in the reaction. By autoclaving the paper discs three times, we hoped to reduce RNase contaminations. When we repeated the experiment with autoclaved paper, results were simply astonishing. Although the standard deviation was quite high, probably because of little differences in the shape of the paper discs, we demonstrated that it is possible to conduct a CFPS reaction on simple autoclaved paper.
The fluorescescence signal depended on paper type. In total, we tested five different paper types on suitability. CFPS worked on every paper tested, however, for subsequent experiments we decided to use the two paper that worked best: 827B from Macherey & Nagel (M&N) and C350L from Munktell. Interestingly, when we used the commercially available S30 T7 cell free expression system at the same scale, the resulting fluorescence signal was far lower. This may be attributed to the fact that the commercial system is not designed to work under these conditions. However, we showed that our cell free expression system based on crude cell extract works very well on paper. Furthermore, it is very amenable to these conditions as well as to other environments (see robustness section).
We wondered if it was possible to lyophilize the paper discs together with the CFPS reaction, like Pardee et al. did. The 15 µL reactions were applied on paper discs, put into microcentrifuge tubes and flash-freezed in liquid nitrogen. Lyophilization was carried out for 24 h. Afterwards, the paper discs were rehydrated with tap water instead of RNase free water.
The results were once again beyond our best expectations. CFPS worked well on paper after lyophilization even though tap water was used for rehydration. Commercial cell extract showed a signal, too, but our crude cell extract performed better on both paper types.
In the preceding experiment, rehydration was performed right after lyophilization. As our aim was the creation of a cell-free biosensor for the detection of variuos substances outside of the lab, the shelf life of the lyophilized paper discs was of outstanding interest. We repeated the reaction und the same conditions, except that after lyophiliztation, the microcentrifuge tubes were stored at 8°C and room temperature. 6 days later, reactions were rehydrated with 15 µL of tap water and fluorescence signals were measured in Tecan plate reader.
We demonstrated that conduction of a CFPS on paper is still possible after 6 days of storage at 8 °C, a fact that is of outstanding interst for our final application. The fluorescence signal was lowered by about 50% compared to the directly started reaction, however, storage conditions can easily be adapted to maintain activity, as has been proposed for example by Pardee et al. 2014.
Addtionally, 6 day storage was performed at room temperature and 60 °C. Reaction started after rehydration when stored at room temperature, which was remarkably. When the tubes containing the paper discs with the reaction components were provisional sealed with adhesive film, a higher fluorescence signal was observed (see nearby graph). Therefore we propose that a professional sealing would be better for storage and prevent the paper from moistening.
Storage at 60 °C was not beneficial or more precisely fatal for extract activity. No fluorescence signal raised after rehydration. This was not unexpected, and would have been rather incredible though.
With a smartphone and our measurement device, we took pictures of the fluorescent paper discs and evaluated the results with our app. Take a look on these pages for more images, details and lots of intersting stuff!
Robustness
Our CFPS system occured to be very robust towards different substances. In particular, 5% of EtOH and 9% lake water instead of RNase free water had little and no effect, respectively, on final fluorescence signal, as was the case for tap water when rehydrating lyophilized extracts (see section CFPS on Paper).
As our cell-free system is designed to work outside the lab, we dealt with the question if it is possible to conduct sfGFP synthesis at temperatures other than 37°C. The optimal temperature for cell-free protein synthesis depends on the protein and the system itself (Spirin and Swartz, 2008).
Our CFPS still worked well at 29 °C and 25 °C, although still 37 °C yielded the highest fluorescence. As we can synthesize sfGFP at temperatures around room temperature, a performance in the field is quite conceivable.
We further investigated the effects of different heavy metals on our CFPS system, results hereof can be found at the Heavy Metal result page.
Low cost energy solution
Laboratory reagents are never for free. But you will always try to get the most out of your money. With this in mind, you may be surprised that the reagents for a small scale CFPS reactions cost less than the reagents needed for a PCR (Pardee et al. 2014). We calculated the cost to perform a 15 µL CFPS reaction one has to spend when following our protocols. To keep it simple, we only took into account all the chemicals needed to perform CFPS.
We analyzed how it would be possible to even further reduce the costs of CFPS reactions. One major cost aspect relies on the energy source which enables ATP regeneration. Phosphoenolpyruvate (PEP) is already cheaper than the often used 3-Phosphoglyerate (3-PGA). However, we tested two easily affordable substances to fuel our reactions: The common polysaccharide maltodextrin and the polyphosphate molecule Hexametaphosphate, which have been shown to work in cell free protein synthesis (Caschera and Noireaux, 2015a).
We showed that these energy sources can substitute PEP when Na-oxalate is removed from the reaction. The observed fluorescence signal was similar in multiple performances. The low cost energy solution is a good alternative for CFPS. A cost estimation revealed: When Hexametaphosphate and maltodextrin substitute for PEP and Na-oxalate, the costs for a 15 µL reaction (reagents only) are reduced by about 20%.
However, the output signal obtained with the low cost energy solution can not compete with the signal from our fully optimized PEP-based system, as you can see in the nearby graph. The advantage is that one surely can optimize the low cost energy solution system as well (Caschera and Noireaux, 2015a), albeit this was not the aim of our project.
References
Górna, Maria W.; Pietras, Zbigniew; Tsai, Yi-Chun; Callaghan, Anastasia J.; Hernández, Helena; Robinson, Carol V.; Luisi, Ben F. (2010): The regulatory protein RraA modulates RNA-binding and helicase activities of the E. coli RNA degradosome. In RNA (New York, N.Y.) 16 (3), pp. 553–562. DOI: 10.1261/rna.1858010.
Lee, Kangseok; Zhan, Xiaoming; Gao, Junjun; Qiu, Ji; Feng, Yanan; Meganathan, R. et al. (2003): RraA: a Protein Inhibitor of RNase E Activity that Globally Modulates RNA Abundance in E. coli. In Cell 114 (5), pp. 623–634. DOI: 10.1016/j.cell.2003.08.003.
Yeom, Ji-Hyun; Go, Hayoung; Shin, Eunkyoung; Kim, Hyun-Lee; Han, Seung Hyun; Moore, Christopher J. et al. (2008): Inhibitory effects of RraA and RraB on RNAse E-related enzymes imply conserved functions in the regulated enzymatic cleavage of RNA. In FEMS microbiology letters 285 (1), pp. 10–15. DOI: 10.1111/j.1574-6968.2008.01205.x.
Find further references here at the CFPS background page!