Optimize protein production and purification
Our sensor is based on utilizing the BlcR transcription factor from Agrobacterium tumefaciens. To create our sensor different experiments with pure BlcR are performed. Because our sensor is based on this transcription factor it was important to optimize the production and purification method. Literature reports indicated that BlcR could be actively produced in E. coli [1][2] . Therefore, we designed a proper plasmid for BlcR production in BL21(DE3) cells (Figure 2).
Figure 2. Production plasmid for BlcR. pET11a plasmid is used as a backbone and the gene coding for BlcR is used as an insert.
The chosen plasmid was a pET11a production plasmid , which could be induced with IPTG for protein production. For easier protein purification with an affinity column, a His-tag was inserted in the N protein terminal, together with a TEV protease site to remove the His-tag if needed. (Figure 3).
Figure 3. BlcR gene upfront TEV site, his-tag, and T7-tag.
For our downstream experiments in the project, we required a highly concentrated pure protein, diluted in a suitable DNA binding buffer with low salt concentration. We went through several protein production and purification cycles to determine the best protein production and purification protocol.
First iteration
Design:
To investigate DNA/BlcR binding affinity, we utilized isothermal titration calorimetry (ITC). However, this method requires a low salt buffer to avoid interference in the DNA/BlcR binding. We found that the iGEM bielefeld team from 2015, which also worked with BlcR, utilized a low-salt storage buffer: 50 mM HEPES, pH 7.2 as the storage buffer of BlcR [2] . We then decided to try this buffer as both a storage and binding buffer. We produced the BlcR in E. coli BL21(DE3) followed by purification with a His-tag affinity column. After His-tag purification, we decided to concentrate the protein solution and dialyse it from the elution buffer (50 mM Tris-HCl, 300 mM NaCl, 300 mM Imidazole, 10% glycerol, pH 7.5) into the storage buffer (50 mM HEPES, pH 7.2) (Figure 4).
Figure 4. First production and purification process scheme of BlcR.
Build:
BlcR was produced in BL21(DE3) cells and purified with His-tag purification columns. 50 mM HEPES, pH 7.2, pH adjusted with NaOH was prepared as a storage buffer for BlcR.
Test:
After the standard protein production and purification, we obtained a stock of 98 µM ( Notebook 1 , 05/07/2022). After the concentration and dialysis steps, we checked the protein concentration again and found a concentration of 1.6 µM ( Notebook 1 , 08/07/2022). The protein concentration decreased by 98%.
Learn:
Our hypothesis for this enormous decrease in concentration is that the protein got stuck in the membrane of the concentrator. We used a concentrator tube with a 30.00 MWCO membrane. Since our protein forms 70 kDa dimers, we hypothesized that the dimers started to form in the solution and they could not pass through the membrane when concentrating the sample.
Second iteration
Design:
To maintain a high protein concentration, we decided to do the protein purification again but this time without concentrating the protein (Figure 5). For atomic force microscopy (AFM) measurements, we needed to dialyze BlcR into 20 mM MOPS.
Figure 5. Second production and purification process scheme of BlcR.
Build:
BlcR was produced in BL21(DE3) cells and purified with His-tag purification. 50 mM HEPES, pH 7.2, pH and 20 mM MOPS, 2,5 mM magnesium, pH 7, and pH adjusted with NaSO4 were prepared as storage buffers for BlcR.
Test:
In this trial, protein production and purification was performed without the concentration step. The protein solution was dialyzed into two different buffers; MOPS and HEPES. After dialysis, we had a lot of protein precipitation (Figure 6). Therefore, we again lost a high amount of the initial protein concentration. After elution, the protein concentration was 86 µM ( Notebook 1 , 25/07/2022) and after dialysis, the protein concentration in the HEPES buffer was only 8 µM and in the MOPS buffer 17 µM ( Notebook 1 , date: 03/08/2022). The concentration of 17 µM of BlcR in the MOPS buffer was good enough for AFM experiments and therefore successful.
Figure 6. Protein solution after dialysis. Protein was dialyzed from the elution buffer (50 mM Tris-HCl, 300 mM imidazole, 300 mM NaCl, pH 7.2) to the storage buffer (50 mM HEPES, pH 7.2).
Learn:
We saw some improvements in the protein concentration without the concentration step in the purification cycle, but we got a lot of precipitation when we performed dialysis. We hypothesized that the protein precipitated due to the absence of salt, which is generally needed for protein stability. The elution buffer contained 300 mM of NaCl, but when changing buffers to a buffer with no salt, the protein started to precipitate. When performing the ITC experiments we found no activity of our protein [TU Delft, results section]. When digging more into literature and having expert advice on protein purification and activity, we decided to drop our efforts to have a no-salt salt concentration in the storage buffer. Since these conditions do not give the ideal environment for good protein activity and stability.
With the MOPS buffer we got a higher protein concentration than with the HEPES buffer, we still wanted to look for a possible storage method in the HEPES buffer since we had limited resources of MOPS and NaSO4 to adjust the pH. MOPS was particularly necessary for electrical experiments, see experimental design , therefore we wanted to save our MOPS and NaSO4 resources for those experiments.
Third iteration
Design:
From what we learned in the previous iteration, we decided to include at least 100 mM NaCl in the storage buffer. The new buffer was then: 50 mM HEPES, 100 mM NaCl, pH 7.2.
Build:
We made a new storage buffer: 50 mM HEPES, 100 mM NaCl, pH 7.2.
Test:
The protein production was performed again following the flow scheme of Figure 5. Once again, a lot of the protein precipitated. However, the concentration seemed to have increased compared to the second production cycle. We checked the concentration after dialysis with a Bradford assay and obtained a protein concentration of 16 uM [ Notebook 1 , 17/08/2022]. The concentration of protein had doubled compared to the HEPES storage buffer with no salt. After dialysis, we ran an SDS PAGE. We noticed that some impurities were present in our protein sample, as some protein bands around 83 kDa were visible (Figure 7).
Figure 7. SDS PAGE to check the purity of our protein of interest. Next to the ladder (L), different concentrations of the protein solution are loaded. 1: 16 uM, 2: 8 uM, 3: 4 uM. Bands between 37 and 49 kDa could correspond to the BlcR monomer and bands between 64 and 82 could correspond to the dimer of BlcR.
Learn:
From this engineering cycle we confirmed that the addition of salt gives more protein stability in solution. Unfortunately, the protein concentration is still low and the SDS PAGE showed some contamination in the protein samples. Since we needed a highly pure, and more stable protein stock, we decided to continue with another iteration for protein purification.
Fourth iteration
Design:
We decided to carry out new production and purification rounds including some more steps to optimize the yield and activity of the protein (Figure 8). Moreover, in this iteration we also included extra purification steps for better protein stock purity and quality. We included the TEV protease digestion to avoid any His-tag interference in protein activity, plus a size exclusion purification step. Finally, we also decided to include glycerol in the storage buffer to give more stability to the protein.
Figure 8. Fourth production and purification scheme for BlcR. This production cycle included also purification with size exclusion and digestion with TEV protease to get rid of the T7- and His Tag.
Build:
In the building phase we prepared a buffer with 50 mM HEPES, 200 mM NaCl, 10 % glycerol, and pH 7.2 for the storage of BlcR. Before size exclusion, the BlcR samples were treated with TEV protease followed by gel filtration to separate the proteins based on their size.
Test:
In the testing phase we checked different elution fractions from the size exclusion on SDS PAGE gel (Figure 8). Two bands were visible around 37 kDa which most likely indicates BlcR with T7-tag, His-tag, and TEV site (Figure 3), and without tags (Figure 9, 3-5). In some fractions contamination of other proteins around 115 kDa was visible (Figure 9, 1-2). The fractions with pure BlcR were combined and concentration was measured with nanodrop. A concentration of 2.9 uM ( Notebook 1 , 29/08/2022) was measured.
Figure 9. SDS PAGE of BlcR samples after size exclusion. Next to the ladder (L), different fractions from the size exclusion are loaded. 1: A9, 2: B5, 3: C1, 4: C2, 5: C4, 6: C8, 7: C12, 8: D1.
Learn:
When we used size exclusion to purify the protein we got higher purity but less concentration. Moreover, we also found that TEV protease was not 100% efficient. After purification we still saw some BlcR + tags present in the eluted sample (Figure 9).
Fifth iteration
Design:
Up until now we had problems in maintaining protein concentration during the purification cycle. In every purification step, we seem to lose a lot of protein. We talked to another expert in the field of protein purification, Theo van Laar, from TU Delft. He suggested eluting with TEV protease instead of imidazole. In theory, only the BlcR with the TEV site would come off the Ni-NTA beads resulting in a pure protein sample. We also decided to change the storage buffer one more time to have a higher salt concentration. The storage buffer was now 50 mM Tris, 300 mM NaCl, 0.5 mM EDTA, and pH 7.5 [1] .
Figure 10. Protein purification cycle with his tag elution by TEV digestion.
Build:
Tris, 300 mM NaCl, 0.5 mM EDTA, pH 7.5 buffer was prepared and His-tag purification was done with TEV digestion as an elution method.
Test:
Since the elution with TEV digestion was not highly successful, only a small portion of BlcR eluted from the column, a very light band is visible around 37 kDa (Figure 11, 1).
Figure 11. SDS PAGE of purification cycle of BlcR. 1: TEV elution 1 2: TEV elution 2, 3: TEV elution 3, 4-5: column wash, 6: column wash, 7: column flow trough, 8: clear lysate.
Build:
We decided to do another elution round with imidazole to get all the BlcR out of the purification column.
Test:
The elution with imidazole and the sample after dialysis into the Tris storage buffer were checked on SDS page gel. The gel showed digested and undigested BlcR around 40 kDa (Figure 12, 4-6). The concentration after elution with imidazole was measured and corresponds to 49.5 µM ( Notebook 1 , 22/09/2022). We did not notice any protein precipitation after dialyzing the protein into the Tris storage buffer. The protein concentration after dialysis was 36.1 µM ( Notebook 1 , 22/09/2022), which only represented a 27% protein loss. We further purified the protein with gel filtration. We only saw contamination of a protein around 115 kDa in the first elution fractions (Figure 13, 1-2). Pure BlcR was present in the other elution fractions (Figure 13, 3-9). We measured a final protein concentration of 13.4 uM ( Notebook 1 , 29/09/2022).
Figure 12. SDS-PAGE gel showing the elution of BlcR. 1: clarified lysate, 2: column flow-through, 3: wash 3, 4-5: imidazole elution, 6: elution after dialysis.
Figure 13. SDS PAGE of different elution fractions after size exclusion. 1 : B1, 2 : B3, 3 : B6, 4 : B7, 6: B11, 7 : C2, 8 : C5, 9 : C7.
Learn:
From this iteration, we learned that TEV digestion for eluting the protein from the column is not very effective in our case. Moreover, we found that we needed at least 300 mM NaCl in the storage buffer to ensure a minimal precipitation and high protein concentration. With this final protocol, we managed to get a relatively high protein concentration with a good purity for the downstream DNA binding experiments. We used the protein sample to successfully determine the binding to engineered oligos, see results [TU Delft, section results , module 3].
Immobilization and electrical measurements
For our final GHB sensor , we used an interdigitated electrode with immobilized DNA (blc operator sequence) to the gold surface. The transcription factor, BlcR, can bind to this blc operator sequence. When GHB is present BlcR will dissociate from the blc operator sequence causing an electrical signal (Figure 14).
Figure 14. Systematic overview of our GHB detecting sensor. A DNA sequence specific to the binding of the transcription factor BlcR is immobilized on an electrode that can measure capacitance differences. In the absence of GHB BlcR is bound to the DNA resulting in a more dense environment between the electrodes. In the presence of GHB BlcR will dissociate from the DNA resulting in a change in environment which can be converted to an electrical signal.
We decided to use atomic force microscopy (AFM) to first verify the DNA and BlcR binding and unbinding from in DNA-coated electrodes. With such a technique, we can scan the gold surface and read the differences in height distribution upon BlcR binding and unbinding. We immobilized ssDNA on the electrode with mercaptohexanol (MCH) to avoid nonspecific binding. We then added the complementary ssDNA to obtain a final dsDNA coated surface ready for BlcR binding and unbinding experiments. In total we want to visualize four different stages; an empty gold surface, gold surface with immobilized DNA, gold surface with immobilized DNA and BlcR, and lastly the gold surface with immobilized DNA and BlcR after addition of SSA (Figure 15). The conditions for the immobilization of the DNA to the gold surface are examined in different iterations of the engineering cycle.
Figure 15. Stages of the gold surface to verify binding and unbinding of BlcR to its DNA sequence. First empty gold plate, second DNA immobilization to the gold surface, third addition of BlcR, fourth the addition of SSA.
First iteration
Design:
For the first ssDNA immobilization trial, we wanted to immobilize the ssDNA onto the gold surface on the electrode. MCH was also immobilized on the electrode to make sure that all of the active gold was covered. We decided to immobilize MCH and ssDNA in a single-step reaction, to reduce incubation time and to have the maximum homogeneity.
Build:
A solution was prepared with final concentrations of 2 µM ssDNA and 1 mM MCH, that was added to the surface.
Test:
While attempting to immobilize the thiol-modified ssDNA and MCH, we immediately noticed that these solutions resulted in a non-soluble mixture, which is not convenient for the surface coating. See Figure 16.
Figure 16. Photograph of two Eppendorfs with a mixture of 2 µM thiol modified ssDNA and 1 mM MCH.
Learn:
From this first binding attempt, we decided to do a sequential ssDNA and MCH binding. The MCH was applied after the thiol-modified ssDNA had first been immobilized on the electrode. Additionally, we found that the electrodes were not sufficiently flat, which led us to use sonication in the subsequent iteration.
Second iteration
Design:
To make sure that both the ssDNA and the MCH can bind to the gold surface we decided to make separate solutions of ssDNA and MCH and first incubate the electrode with the thiol-modified ssDNA, followed by an incubation with MCH. In addition to sequential binding, we also decided to add a sonication step. With such, we expect to see a flatter electrode surface (Figure 17).
Figure 17. Immobilization flowchart. First immobilization of thiolmodified ssDNA, second immoblization of mercaptohexanol, third addition of the second ssDNA.
In the first iteration, the electrodes were not sonicated since it was not expected that this would be necessary. In this iteration round, we decided to sonicate the electrodes to obtain a flatter surface.
Build:
Stock solution of 2 µM ssDNA and 1 mM MCH were prepared separately. And the immobilization of ssDNA and MCH was done in a two-step reaction.
Test:
To verify the binding of the DNA and BlcR, Atomic Force Microscopy (AFM) was used. AFM analyzes the surface of the electrode by scanning it with a cantilever and measuring the force needed to go through the sample.
Figure 18. Atomic Force Microscopy pictures of interdigitated electrodes (IDE). SNL-10B probe (frequency = 4.365 kHz, spring constant = 0.101 N/m). a) Empty IDE in MOPS buffer scale up to 35 nm. b) IDE with immobilized dsDNA scale up to 240 nm. See Notebook 4 (27/07/2022 and 03/08/2022).
A couple of things could immediately be noticed. First, an empty interdigitated electrodes (IDE) was scanned. However, it could be seen that the surface was not flat, instead it had peaks up to 35 nm (Figure 18a). It would be too rough to notice protein binding or dissociation. Second of all, when 2 µM DNA (Figure 18b) was immobilized on the empty electrode, heights of up to 240 nm were measured (Figure 18b).
Learn:
We hypothesize that the differences in height during the measurements were clumps formed by oxidized DNA through thiol-thiol interactions. To avoid this in our next attempt, we decided to introduce a reduction step in our protocol. We also noticed that the clean electrode was not as flat as we wanted it to be. We then decided to introduce an extra cleaning step for having a flatter surface before continuing with the binding experiments.
Third iteration
Design:
To avoid oxidized-DNA clumps in the gold plate, we decided to introduce a reduction step before DNA surface binding. We treated the ssDNA with DTT for 15 minutes. Furthermore, before binding experiments, we decided to add an extra step of sonication to better clean the gold surface. This time we also decided to sonicate the electrodes as it could help in visualizing the DNA better.
Build:
Before immobilizing the ssDNA to the gold surface the IDE was sonicated. 100 µM stock of ssDNA was reduced with 10 mM DTT before immobilization to the gold surface. The DTT was then extracted with ethyl acetate.
Test:
Again, AFM was used to analyze the gold surface. We saw that extra sonication step helped with obtaining a flatter surface. (Figure 19a). Here, the highest point on the surface was 30.4 nM compared to 35 nm without sonication (Figure 17a). Moreover, we found that the DTT DNA treatment also helped. This time we did not observe the giant clumps from the last iteration. The highest point measured with reduced DNA is 63.8 nm (Figure 19b) compared to the 240 nm (Figure 18b) that we saw before.
Figure 19. Atomic Force Microscopy pictures of Interdigitated electrodes (IDE). SNL-10B probe (frequency = 4.365 kHz, spring constant = 0.101 N/m). a) Sonicated IDE in MOPS buffer. Scale up to 30.4 nm. b) IDE with 2 µM reduced thiol-modified DNA in MOPS buffer. Scale up to 63.8 nm. See Notebook 4 (03/08/2022 and 10/08/2022)
Learn:
We concluded that the additional sonification and DTT reduction steps showed better surfaces when scanned using the AFM. We then decided to continue with this protocol for surface preparation and move on to BlcR binding and unbinding measurements.
Fourth iteration
Design:
We incubated the electrode with immobilized dsDNA with BlcR 2.45 µM for one hour. After this hour, we washed the electrode with a MOPS Mg buffer. The electrode was then analyzed with AFM in a MOPS buffer bath.
Build:
BlcR was purified and dialyzed into 20 mM MOPS, 2.5 mM Mg, pH 7. A stock solution of 10 µM BlcR was made and incubated on the IDE with DNA.
Test:
Using AFM, no clear difference in height could be seen when BlcR was added to the electrode with immobilized DNA. Only one big blot of 296 nm (Figure 20).
Figure 20. Atomic Force Microscopy pictures of Interdigitated electrodes (IDE). SNL-10B probe (frequency = 4.365 kHz, spring constant = 0.101 N/m). IDE with DNA and BlcR in a 20 mM MOPS buffer. Scale up to 296 nm. See Notebook 4 (04/08/2022)
Learn:
We came to the conclusion that BlcR could unbind the DNA with the post-binding washing step that we performed in this iteration. We then decided to avoid this washing step, and take the AFM measurements in the incubation bath with BlcR.
Fifth iteration
Design:
Before going ahead with more BlcR binding experiments we decided to test atomically flat gold plates instead of interdigitated electrodes on AFM. Even though the electrodes were flatter after sonication (up to 30 nm) (Figure 19a), we thought that 30 nm height could hinder the BlcR binding and unbinding signal. Moreover, we also decided to take the AFM image in the same BlcR binding buffer. With this, we avoid washing away BlcR after binding onto the DNA.
Build & Test:
After analyzing the new flat gold plate with AFM, we saw that the surface had a maximum height of 14.7 nm (Figure 21). This is significantly better than the sonicated IDEs (30 nm) (Figure 19a).
Figure 21. QI image taken with the AFM of an empty, sonicated gold plate in MOPS buffer. Scale up to 14.7 nm. See Notebook 4 (10/08/2022)
Learn:
Using atomically flat gold plates gives the flattest results.
Sixth iteration
Design:
Using the previously tested gold plates, we decided to test BlcR binding and unbinding. We measured the DNA-coated surface height with AFM upon (i) addition of BlcR (binding), and (ii) BlcR + SSA (unbinding).
Build:
1 µM of DNA was immobilized on the gold surface. The gold surface with DNA was incubated in a 10 uM BlcR solution. The plate was analyzed with AFM. 15 µM SSA was then added to the solution and the plate was analyzed with AFM once again.
Test:
Results show that the DNA adheres to the gold plate effectively (Figure 22b). Moreover, upon BlcR addition, the number of dots on the surface increases significantly, indicating an efficient DNA-protein binding (Figure 22c) . Finally, when SSA is added to the DNA + BlcR surface, we noticed a significant reduction in the number of AFM detected dots (a decrease of 46 % of peaks after addition of SSA) (Figure 22d). This could indicate that the BlcR-DNA unbinding took place effectively.
Figure 22. QI image of (a) an empty gold plate, (b) a gold plate immobilized with 1 µM thiol modified DNA, (c) a gold plate immobilized with 1 µM thiol modified DNA, with 3 µM BlcR and (d) a gold plate immobilized with 1 µM thiol modified DNA, with 3 µM BlcR, with 15 µM SSA. All images are scaled to a maximum value of 35 nm. See Notebook 4 (10/08/2022 and 09/09/2022).
Learn:
Throughout the iterations of the engineering cycle we developed the best protocol to use AFM as a visualization technique for protein-DNA binding and unbinding. We changed the protocol several times to include sonication, chemical reduction, or even switching to atomically flat gold plates instead of the IDEs. All these choices helped to get the final result where we could BlcR associating and dissociating from the immobilized DNA strands (Figure 22).
References
- Pan, Y., Fiscus, V., Meng, W., Zheng, Z., Zhang, L. H., Fuqua, C., & Chen, L. (2011). The Agrobacterium tumefaciens Transcription Factor BlcR Is Regulated via Oligomerization. Journal of Biological Chemistry, 286(23), 20431–20440. https://doi.org/10.1074/jbc.m110.196154
- Gräwe, A.( March 6, 2019). A paper-based, cell-free biosensor system for the detection of heavy metals and date rape drugs. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0210940
- Wenjing Xi and Amanda J. Haes. (February 5, 2019) Elucidation of HEPES Affinity to and Structure on Gold Nanostars. Journal of the American Chemical Society 2019 141 (9), 4034-4042. https://doi.org/10.1021/jacs.8b13211
- Karthika Sankar, R. Baer, Chloé Grazon, Robert C. Sabatelle, Sébastien Lecommandoux, Catherine M. Klapperich, James E. Galagan, and Mark W. Grinstaff (April 12, 2022). An Allosteric Transcription Factor DNA-Binding Electrochemical Biosensor for Progesterone. ACS Sensors 2022 7 (4), 1132-1137. https://doi.org/10.1021/acssensors.2c00133
- Sankar, K., Baer, R., Grazon, C., Sabatelle, R. C., Lecommandoux, S., Klapperich, C. M., Galagan, J. E. & Grinstaff, M. W. (2022, 12 april). An Allosteric Transcription Factor DNA-Binding Electrochemical Biosensor for Progesterone. ACS Sensors, 7(4), 1132–1137. https://doi.org/10.1021/acssensors.2c00133
- Chen, Z., Liu, X., Liu, D., Li, F., Wang, L. & Liu, S. (July 9, 2020). Ultrasensitive Electrochemical DNA Biosensor Fabrication by Coupling an Integral Multifunctional Zirconia-Reduced Graphene Oxide-Thionine Nanocomposite and Exonuclease I-Assisted Cleavage. Frontiers in Chemistry, 8. https://doi.org/10.3389/fchem.2020.00521