Team:Freiburg/Project/method

uniBAss - uniCAS Binding Assay

uniBAss - The innovative & easy screen for binding affinity

We developed a novel ELISA-based method to assess the binding efficiency of our proteins: the uniCAS Binding Assay uniBAss. It is a powerful tool for the biochemical characterization of the binding capacity of our dCas9 fusion proteins and our truncated dCas9 versions with high throughput capabilities.

1. First a 96-well ELISA plate is coated with streptavidin. Then biotinylated oligonucleotides containing the DNA sequence complementary to the crRNA incorporated into dCas9 are applied.

2. 42h after transfection with the dCas9-fusion constructs that will be tested and its desired target site on the RNAimer plasmid HEK-293T cells are taken up in dilution buffer and lysed by sonifying. Then the dCas9/crRNA containing cell lysate is ready to be applied to the wells.

3. The dCas9 in the cell lysate can now bind via its incorporated crRNA to the complementary oligo on the bottom of the well.

4. After a washing step the first antibody is added. This mouse anti-HA antibody binds to the HA-tag at the N-terminus of dCas9.

5. When the unbound first antibody is washed away the secondary antibody can be added. The secondary anti-mouse-antibody is coupled to a horseradish peroxidase (HRP) that catalyzes the oxidation of ABTS by hydrogen peroxide resulting in a change of color.

6. The absorption at 405 nm is then measured and represents the amount of dCas9 bound in each well. To substract out differences in dCas9 expression levels the uniBAss data can be normalized to western blot data to reflect binding properties of dCas9 to its target site.

Results

The uniCAS Binding Assay enables us to verify that our dCas9-fusion proteins show the same DNA binding capacity as dCas9 without an effector. Thus uniBAss can be used to quantify the amount of HA-tagged DNA binding dCas9 and therefore helps us to characterize our constructs more closely.

Figure 1: Binding affinity of dCas9 fusion constructs after normalizing the dCas9 amount with the help of a western blot HEK-293T cells were transfected in 6-well plates with dCas9 encoding an EMX1 crRNA on the same plasmid or co-transfected with dCas9 fusion constructs and the EMX1 RNAimer plasmid. Lysates were split and analyzed on uniBAss and western blot 42h after transfection. The uniBAss ABTS readout seen above is normalized to dCas9 expression levels that were western blot quantified using ImageJ. Depicted values therefore represent comparable binding capacities of our dCas9 effector fusion proteins.

Mammalian HEK-293T cells were transfected with dCas9, dCas9-VP16, dCas9-KRAB and dCas9-G9a and the RNAimer plasmid targeting EMX1. 42h post transfection the standard uniBAss protocol was followed. Constructs we analyzed in technical duplicates to assess the accuracy of uniBAss. For every construct the ABTS readout (background level substracted) was normalized to western blot dCas9 expression data obtained from the same cell lysates. Through this normalization we could compensate differences in protein expression between the constructs and thereby compare the actual binding properties. Figure 1 shows the normalized uniBAss data for dCas9, dCas9-VP16, dCas9-KRAB and dCas9-G9a + RNAimer(EMX1) when targeted to biotinylated EMX1 oligos. Transfection controls without dCas9 or the RNAimer plasmid did not show detectable signals in our binding assay.

Development of uniBAss

The need to characterize the behavior of CRISPR/Cas is motivated by several means. Different fusion constructs have been built based on the CRISPR/Cas system such as our dCas9-VP16 or dCas9–KRAB constructs, but it remained unclear if the binding characteristics of dCas9-VP16/KRAB are different from the native dCas9. Simulating complex gene circuits in-silico generates new needs for straight forward characterization of the binding affinity followed by quantification. By this mean information can be gathered about a variety of promoters for different dCas9 fusion constructs. Different assays have been developed to characterize protein-DNA interplay to analyze their respective binding behavior [1]. Examples of these protein characterization assays are the Chromatin Immunoprecipitation (ChIP) and the DNA Electrophoretic Mobility Shift Assay (EMSA). Both systems have individual strength as well as limitations. ChIP can be used to quantify a sample when coupled with qPCR analysis. However ChIP is limited by the absence of high throughput possibilities [2]. EMSA is able to detect low abundance DNA binding proteins from lysate with a high sensitivity but it is difficult to quantify a sample [3]. However, those assays do not address the need to analyze the binding of a protein to its respective DNA sequence and thus perform a quantification with high capabilities. Starting from these needs a new assay for the biochemical characterization of dCas9 had to be established with the requirements of binding capacity test and quantification with high throughput capabilities. Therefore we developed our Enzyme-linked Immunosorbent Assay (ELISA) uniBAss. The assay is based on a streptavidin-biotin interaction [4] with the 5’ biotin tag of the oligonucleotide [5] was used to immobilize the target DNA.

Optimization

To obtain robust conditions for uniBAss, several experiments were performed. Different salt concentrations as well as different oligonucleotide lengths were examined to optimize this assay. Hereafter different dCas9 fusion proteins were tested regarding their DNA binding capacity. In addition Cas9 versions with one nickase domain and without a nickase domain were compared to investigate the influence of the nicking enzyme towards the binding behavior. Subsequently to the optimization process the resulting buffer condition 10 mM Tris, 1 % BSA, 90 mM NaCl, 5 mM MgCl₂ containing EDTA free Protease inhibitor was used to test different dCas9 fusion constructs.

To assess the influence of different salt compositions in the dilution buffer HEK-293T cells were transfected with a nickase Cas9 version containing a crRNA locus for EMX1. 24h after transfection the cell were taken up in standard dilution buffer b₀ (10 mM Tris, 1 % BSA, 10 mM MgCl₂, 10 mM NaCl) and sonified. After that lysate dilutions with different MgCl₂ and NaCl concentrations were prepared in 96-well plates and analyzed via ELISA. Figure 2 and 3 show that the binding behaviour of the dCas9 / crRNA EMX1 strongly depends on the salt concentrations. For the detailed uniBAss protocol click here.

Figure 2: Binding restraint of the nickase Cas9/EMX1 crRNA complex for varying MgCl2 conditions HEK-293T cells transfected with pIG9000 (nickase Cas9 plasmid containing an EMX1 target site) were lysed in dilution buffer (24h post transfection) and subsequently analyzed for binding affinity to EMX1 oligos under varying buffer conditions. Here the MgCl2 concentration was varied and plotted against the ABTS readout of uniBAss. Error bars represent the standard deviation of 3 replicas.

Figure 3: Binding restrain of the Cas9 MX1 crRNA complex for varying NaCl conditions HEK-293T cells transfected with pIG9000 (nickase Cas9 plasmid containing an EMX1 target site) were lysed in dilution buffer (24h post transfection) and subsequently analyzed for binding affinity to EMX1 oligos under varying buffer conditions. Here the NaCl concentration was varied. Error bars represent the standard deviation of 3 replicas.

The experiment (figure 3) showed that the nickase Cas9 with EMX1 crRNA sequence acts differently for different buffer conditions. Therefore an optimization process was performed focusing on two main points. Firstly, different salt concentrations illustrate one parameter for optimization. Secondly, the length of the target oligonucleotides coated onto the ELISA plate was taken into account for characterization.

Different spacer length flanking the site targeted by Cas9 were designed to investigate differences in binding affinity of the Cas9/crRNA complex to its target (see figure 4). Sequences of all oligos can be found in our plasmids and oligo list. The sequences flanking the human EMX1 locus [6] were used to vary the lengths of the oligonucleotides.

Figure 4: Design of the different oligos BTN = biotinylation on 5'end of one of the annealed oligonucleotides. Different spacer length flanking the EMX1 site targeted by Cas9 were designed to analyze differences in binding affinity of the Cas9/crRNA complex dependent on length of the presented double-stranded DNA oligonucleotide.

Figure 5: DNA binding capacity tested with different oligo-lengths Annealed EMX1 oligonucleotide of varying lengths were coated on ELISA plate, transfection and uniBAss were performed as described here. A more robust binding of the nickase Cas9/EMX1 crRNA complex was achieved with longer oligonucleotide versions. Error bars represent the standard deviation of 3 replicas.

Using different oligonucleotide lengths to determine the best size of the coated oligonucleotides revealed that the presence of a longer sequence at the 5’ end with about 11 nucleotides consisting of the crRNA locus is essential to get the nickase Cas9 to bind. The presence of a second long sequence at the 3’ end of the crRNA however is not essential for simple binding assays but enables a more robust binding environment. Biotinylated oligonucleotides for prospective assays were designed analogue to the longest oligonucleotides used in the attempt to offer a point of interaction for fusion proteins.

The second parameter focused on optimizing the buffer conditions. The used buffer is composed of 10 mM Tris, 1 % BSA and varying concentrations of MgCl₂ and NaCl. The salt concentrations were varied against each other generating 16 different buffers.

Factorial design

The second part of the optimization was done using a factorial design. This approach was used to test whether the MgCl₂ concentration is linearly linked to the ABTS conversion. Therefore different concentrations of MgCl₂ – 30 mM, 20 mM, 10 mM and 5 mM – and of NaCl – 150 mM, 120 mM, 90 mM and 60 mM, were varied against each other. Indicating that for high MgCl₂ concentrations the binding of dCas9 is close to zero (Figure 6). Whereas with descending MgCl₂ concentrations more dCas9 is able to bind to the oligonucleotides, indicating an increase in an exponential matter. This is true for all the tested NaCl concentrations however it is to mention that 150 mM NaCl has the lowest bound nickase Cas9.

Figure 6: Factorial design expriment to test the behaviour of the nickase Cas9 towards different salt environments After 24 h transfected HEK-293T cells (pIG9000) were taken up in dilution buffer b0 and the salt concentrations were adjusted individually resulting in the conditions (10 mM Tris, 1 % BSA, NaCl – 150 mM, 120 mM, 90 mM and 60 mM, MgCl2 – 30 mM, 20 mM, 10 mM and 5 mM respectively). The binding behaviour of the Cas9/crRNA towards EMX1 display similar behaviour for descending MgCl2 concentrations. This is true for all four NaCl concentrations however 150 mM NaCl has the smallest yield. Error bars represent the standard deviation of 3 replicas.

Figure 6 shows that the binding behaviour of the nickase Cas9/crRNA towards EMX1 display similar behaviour for descending MgCl₂. This is true for all four NaCl concentrations however 150 mM NaCl has the smallest yield.

Use of factorial design

To test if the experimentally obtained curves can be fitted to an exponential increase within decreasing MgCl₂ concentration a factorial design approach was performed (see table 1). To check if the conditions either behave in a linear or a non-linear way the experimental b0 values were compared to the calculated b0 values resulting from a central point. The results of this analysis are shown in Table 1. It was oberserved that the amount of bound dCas9 increases with a reduction of the MgCl₂ concentration.

Table 1: Factorial design approach to test the behaviour of the Cas9/crRNA towards different salt environments The results are sorted, from the highest difference (top) to the lowest difference (bottom). The lower the difference the more linear is the area surrounding the central point.

The plotting of the data in two dimensions indicated that there is an exponential increase of the ABTS conversion with descending MgCl₂ concentrations. To verify this, a factorial design with different MgCl₂ concentrations was performed within a range of 30 mM and 5 mM MgCl₂. After having the data illustrated in three dimensions (figure 7), it occurred that the decreasing amounts of NaCl also contribute to an increasing yield.

Figure 7: Data from factorial design expriment plotted in a three dimensional matrix To set the points into a common perspective the data from Figure 8 were plotted using MATLAB R2013a.

Subsequently to the optimization process the resulting buffer condition 10 mM Tris, 1 % BSA, 90 mM NaCl, 5 mM MgCl2 containing EDTA free Protease inhibitor was used to test different dCas9 fusion constructs.

Test of a fusion-construct

After having optimized the buffer conditions for the nickase Cas9 our dCas9–fusion proteins were studied aiming on two points. First to gather information about the target site for the CRISPR/Cas system in a straight forward manner and secondly to study which of the high yield buffer conditions displays the most robust readout when used with various dCas9–fusion proteins. The dCas9-VP16 fusion construct for gene activation was tested on different target on the human VEGF (Vascular endothelial growth factor) gene.

Figure 8: 3 different VEGF biotinylated oligos were used as targets for 3 dCas9-VP16 constructs containing the respective crRNA bla

Figure 8 shows the dCas9-VP16 fusion proteins containing different crRNA targeting VEGF loci. This experiment aimed to examine whether there are different affinities between the different crRNAs. It can be seen, that the VEGF loci 1 and 3 behave similar, whereas the locus 2 reveals a higher ABTS readout in uniBAss. This indicates a higher binding affinity of the dCas9-VP16 construct to the VEGF locus 2 oligo.

Quantification of dCas9

To be able to quantify the amount of dCas9 binding to the oligos in the uniBAss ELISA a purified HA-tagged protein (FM) was used. As the ng/µl concentration of this protein is known, a dilution row was performed to get a standard to convert the ABTS readout in ng protein bound. Figure 9 depicts the ng of Cas9 or the different dCas9 constructs calculated via the formula (see figure 9) of the standard dilution row.

Figure 9: Quantification of different Cas9 and dCas9 constructs bla

The uniBAss ELISA performed for figure 9 indicated that the fusion constructs are 3 - fold less expressed than the constructs with the Cas9 or dCas9 only. The comparison between a dCas9 completely lacking the nickase activity and a Cas9 having one active nickase shows no significant difference. However the yield of the dCas9 construct without any nicking activity is slightly higher.

Reason for truncating dCas9

With a size of 160 kDa, the dCas9 protein of the CRISPR/Cas system is the largest molecular tool among Zinc Fingers and Transcription Activator–Like Effectors. Difficulties that occurred during the light induced translocation of a dCas9 fusion construct in the nucleus indicated that the size of dCas9 might be a bottleneck for efficient genome engineering hence a truncation strategy was developed.

How to truncate dCas9

To reduce the size of dCas9 we ran a PCR over the backbone with two primers binding in dCas9, one with an overlap for Gibson Assembly. After the PCR one-fragment gibson cloning was performed. We tried five different strategies where to reduce the size of the dCas9 protein. In the first attempt we deleted 365 bp near the N-terminus of the protein. In a second try we erased 306 bp near the C-terminal part of dCas9. Here we assume the reverse transcription domain of the protein, which will probably be responsible for DNA binding. Therefore this truncation is thought to verify this assumption and can serve as a negative control. For truncation 3, 4, and 5 we just deleted the beginning, the middle and the end of dCas9 more drastically by throwing out about 1000 bp.

Results

The PCR over the pSB1C3 backbone of the dCas9 plasmid worked for all primer pairs. The different lengths of the PCR products, due to differently truncated dCas9, are clearly visible. After Gibson assembly with these PCR products the test digest of the plasmids also shows the expected length. We cut with NotI so that the insert of the plasmid is cut out. Only for truncation 4 we received a shorter fragment than expected. To ensure the expression of the shortened dCas9 version we transfected the midi-preps of them into HEK-293T cells and performed a western blot with the cell lysates. As it can be seen in figure 10 we could proof the expression of the standardized truncations. As expected truncations 1 and 2 are about 100 amino acids shorter resulting in an approximately 11 kDa shift in the western blot compared to the full-lengths dCas9. Truncation 3 and 5 run at around 120 kDa. Here about 300bp are missing, resulting in a size reduction of about 33kDa. Only truncation 4 is shorter than expected. It has about 80 kDa and the sequencing of the midi-prep of T4 revealed a nearly 2000 bp deletion in the dCas9.

Figure 10: Western blot with anti HA-antibody to show the expression of the different truncations. bla

After the proof of expression we tested the binding capacity of the truncated dCas9 versions to DNA with the uniBAss ELISA (see Results of uniBAss). By now we could not show binding of the untruncated dCas9 in pSB1C3 no matter of CMV or SV40 promoter. This is due to a very weak expression of dCas9 in pSB1C3 which results in amounts of dCas9 which are not sufficient for uniBAss detection levels. Because we cannot prove the binding of the full-length dCas9 in the iGEM backbone, we are not able to make a statement about the binding capacity of the truncated versions. At the moment we are trying to increase the amount of dCas9 for uniBAss to be above threshold.

References