Building on insights from Iteration 1 of the fusion Cas engineering cycle and findings from Kweon et al. (2017), fusion guide RNAs (fgRNAs) were designed by combining the sgRNA from SpCas9 with the crRNA from Cas12a. The goal of the first iteration is to replicate the results presented in the literature and demonstrate the editing activity of both proteins using fgRNA. The genes VEGFA and FANCF were selected as targets for Cas12a and Cas9, each target was tested with each Cas protein. Editing efficiency will be analyzed with the T7 Endonuclease I (T7EI) assay. Controls will include crRNAs and sgRNAs as positive controls, and non-targeting guides as negative controls. An entry vector was designed to facilitate efficient cloning of different spacer sequences. The entry vector includes a U6 promoter, the Cas12a scaffold, a bacterial promoter driving ccdB expression, and the SpCas9 scaffold. Successful spacer integration leads to the removal of the ccdB gene, allowing bacterial growth to be used as an indicator for cloning success.
Kweon, J., Jang, A.-H., Kim, D.-e., Yang, J. W., Yoon, M., Rim Shin, H., Kim, J.-S., and Kim, Y. (2017). Fusion guide RNAs for orthogonal gene manipulation with Cas9 and Cpf1. Nature Communications 8, doi:10.1038/s41467-017-01650-w.
An existing vector containing the U6 promoter and the MbCas12a scaffold was selected as the basis for entry vector cloning. The vector and a ccdB-SpCas9 scaffold construct were PCR amplified and fitting overhangs for SapI were introduced. Golden gate assembly with Esp3I was used to create the final plasmid. The transformation was carried out in the ccdB-resistant XL1 Blue E. Coli strain. Desired spacer sequences can be ordered as oligos, annealed, and cloned in via golden gate with SapI.
Figure 1: Concept of the entry vector. The entry vector would have a ccdB in between the two scaffolds. SapI sites can be utilized for Golden Gate assembly cloning in an oligo anneal with fitting overhangs. On the right site the assembled gene fragment is shown.
The T7EI assay showed more bands than expected for both tested loci, potentially due to nonspecific primer binding. However, clear editing activity was observed (highlighted with an arrow).
Figure 2: First test of fgRNA in a T7EI assay. T7EI samples were run on a 2% agarose gel in 1x TBE buffer dyed with gel red. The top gel was used for the FANCF locus and the bottom gel for VEGFA locus targeting. Editing is indicated by an extra band compared to the negative control indicated by the arrow.
Overall low editing activity was observed. Additionally, the PCR conditions require further optimization to improve specificity and reduce nonspecific amplification. Together these results were not sufficient to unambiguously prove editing activity.
To minimize possible off-target amplification, different annealing temperatures were tested ranging from 60 °C up to 72 °C. In parallel, a new set of primers was designed to use for the assay.
New primers for VEGFA and FANCF were designed using the NCBI primer BLAST tool to ensure high target fidelity.
The old primers, tested across different annealing temperatures, did not show a notable improvement compared to the previous test. However, the new primers performed better, revealing editing for all samples tested. No significant difference in editing efficiency was observed between AsCas12a and MbCas12a. The fgRNA constructs showed overall lower editing activity compared to sgRNA controls.
A
B
C
Figure 3: Primer evaluation for further T7EI testing. A and B, different primer pairs for VEGFA and FANCF tested at different temperatures run on a 2 % agarose gel in 1x TBE. Gel image was taken before the T7 endonuclease I digestion step. N, P and S mark the negative control, positive control and a sample, respectively. Contrast was increased to check for byproducts. C, after T7EI digestion, samples were run on a 2% agarose gel in 1x TBE buffer dyed with gel red (gel image not shown). Editing rate for the T7EI assay, calculated with ImageJ from the gel image after T7EI digestion, is plotted. The schematic at the top shows the composition of the fgRNA. Below each spacer is the targeted gene. The symbols below show which parts are in each sample.
The newly designed primers produced improved results, allowing for a more accurate assessment of editing activity. No notable difference in editing efficiency between AsCas12a and MbCas12a was observed. Given the insights gained from the fusion Cas engineering cycle, we chose to only continue with MbCas12a for future experiments.
Building upon the findings from the fusion Cas engineering cycle, we aimed to combine the fgRNA with the fusion Cas protein to assess their combined editing activity. For this purpose, we adapted the previously optimized T7 Endonuclease I (T7EI) assay, using the established primer pair.
Since no prior T7EI assay had been conducted with the fusion Cas, we carefully adjusted the transfection ratios. As the fusion Cas combines the two previously tested Cas proteins into one construct, we therefore used the cumulative transfection amounts of the individual Cas proteins.
The assay shows no editing for the non-targeting negative control, and strong editing for the single Cas protein positive control. The co-transfected single Cas proteins display comparable editing activity with the positive control. The fusion Cas proteins exhibited editing activity in combination with fgRNA, albeit lower than compared to sgRNAs. Notably, better editing efficiency could be shown for VEGFA compared to FANCF.
A
B
C
Figure 4: Editing rates for fusion guide RNAs with fusion Cas proteins. A - C, after T7EI digestion samples were run on a 2% agarose gel in 1x TBE buffer dyed with gel red (gel image not shown). Editing rate for the T7EI assay, calculated with ImageJ from the gel image after T7EI digestion, is plotted. The schematic at the top shows the composition of the fgRNA. Below each spacer is the targeted gene. The symbols below show which parts are in each sample. Cas proteins linked by a dash ("–") were fused to each other. Replicates are marked as individual dots.
The T7EI assay confirmed fusion Cas activity for both targets in combination with fgRNA, demonstrating the feasibility of using fusion Cas proteins and fgRNAs in our system. The results from this and previous iterations, along with the fusion Cas engineering cycle, establish a strong foundation for developing the Cas staples. However, simultaneous binding of both Cas proteins to the fgRNA can not be shown with this type of assay. To address this, alternative experiments need to be conducted.
A proximity assay based on CRISPR activation (CRISPRa) was developed to assess simultaneous binding of dSpCas9 and dMbCas12a. The fgRNA was designed with a tetO target site for dSpCas9 and a non-targeting spacer for dMbCas12a, which was fused to the transcriptional activator VP64. In the assay, a minimal promoter and firefly luciferase gene were positioned adjacent to the tetO binding site. This design ensures that luciferase expression is only induced when both dSpCas9 and dMbCas12a bind to the fgRNA simultaneously. A non-targeting spacer and dSpCas9-VP64 were included as negative and positive controls, respectively.
A
B
Figure 5: Design to test simultaneous binding of Cas9 and Cas12a to the same fgRNA. A shows our model for a successful sample. For the sample, the fgRNA consists of a non-targeting spacer for Cas12a and a TRE targeting spacer for dCas9 resulting in dCas9 binding. dCas-RNA complex formation of the dCas12a with the fgRNA already complexed by the dCas9 would result in activation of the Luciferase gene. In B we also depict the positive control, dCas9 being fused to VP64, and the negative control, both spacers of the fgRNA being non-targeting.
The VP64 gene was PCR amplified and inserted into an already existing plasmid containing dMbCas12a. The final plasmid was then assembled with golden gate assembly. Multiple fgRNA constructs with varying linker lengths were designed and cloned in as annealed oligos in the previously described entry vector.
The positive and negative controls produced the expected outcomes. However, no binding of dSpCas9 or recruitment of dMbCas12-VP64 could be demonstrated across all linker types tested. Data normalization using Renilla luciferase was not possible due to a planning error made during the assay preparation.
Figure 6: Testing simultaneous complexing of the fgRNA with Cas9 and Cas12a. Firefly luminescence intensity measured 48 h after transfection. On the x-axis the negative control (empty spacer), positive control (sgRNA, no linker, 5 nt linker, 10 nt linker) and the sample ( no linker, 5 nt linker, 10 nt linker). The symbols above show the composition of each sample. Data is depicted as the mean +/- SD (n=3). No statistical analysis was performed.
Despite the lack of data normalization, incorrect evaluation is unlikely, as both controls performed as expected. Further literature research was conducted to explore potential reasons for the assay outcome and no published data supporting the functionality of a dMbCas12a-VP64 construct could be found. Additionally, incorrect transfection ratios may have prevented efficient Cas protein binding to the fgRNA, potentially contributing to the lack of observed activity. Based on these findings, the assay design was critically re-evaluated and revised.
Based on previous learnings, the assay was re-designed. dSpCas9 was fused to VP64 and dMbCas12a was used with a targeting spacer to position dSpCas9 near the minimal promoter and firefly luciferase.
Figure 7: Improved design to test simultaneous binding of Cas9 and Cas12a to the same fgRNA. Composition changed for the following iteration of the assay compared to figure 5. For the sample the fgRNA consists of a non-targeting spacer for dCas9 and a TRE targeting spacer for dCas12a resulting in dCas12a binding. dCas-RNA complex formation of the Cas9 with the fgRNA already complexed by the dCas9 would result in activation of the Luciferase gene by the VP64 fused to the dCas9.
Since both dCas12a and dCas9-VP64 plasmids were already available, no additional cloning was required for these components. The modified fgRNAs, containing a dMbCas12a spacer targeting tetO and a non-targeting dCas9 spacer, were cloned into the entry vector via a Golden Gate reaction. Different fgRNA concentrations were transfected to explore varying fgRNA to protein ratios.
Similar to the previous outcome, the controls worked as expected. No change in firefly expression could be shown with all tested fgRNA constructs.
Figure 8: Testing simultaneous complexing of the fgRNA with an improved design. Firefly luminescence intensity measured 48 h after transfection. Normalized against renilla luminescence. On the x-axis the negative control (empty spacer), positive control (sgRNA, CG-linker) and the sample (GC-linker bottom strand, CG-linker top strand, GC-linker bottom strand 1 ng, CG-linker top strand 1 ng). The symbols above show the composition of each sample. Data is depicted as the mean +/- SD (n=3). No statistical analysis was performed.
The adjustments made in this iteration did not lead to improved assay outcomes. Simultaneous binding of both Cas proteins with the fgRNA could still not be confirmed. Future experiments may require the development of new assay systems and further optimization such as adjustments to the linker lengths.
Our idea was to utilize catalytically inactive Cas (dCas) proteins to bring two DNA strands into close proximity. We decided to use dSpCas9 and different dCas12a orthologs (MbSas12a, AsCas12a, and LbCas12a), for further testing. As a start, we focused on well established editing assays, assuming that editing capabilities are positively correlated with DNA binding capabilities. Editing capabilities were tested with a T7 Endonuclease I (T7EI) assay.
For the T7EI, we transfected a SpCas9 targeting CCR5 and each Cas12a ortholog targeting Runx1, Dnmt1 and FANCF. The Cas proteins were constitutively expressed by a CMV promoter and the sgRNA/crRNA were expressed by a U6 promoter. Transfection ratios were chosen as established. Cells were lysed 72 hours post-transfection, and regions of approximately 1000 base pairs around the cut sites were amplified by PCR for analysis. T7EI recognizes and cleaves mismatches generated by the DNA repair machinery, resulting in two DNA fragments of roughly equal size. These fragments were visualized on an agarose gel, indicating whether the target sites were cleaved. Quantification of editing efficiency was performed using the ImageJ software for image analysis.
The results of the T7EI assay are depicted in the figure below. SpCas9 worked, as expected, the best. The Cas12a orthologs displayed lower editing capabilities overall, but demonstrated detectable activity at the RUNX1 locus. Among the Cas12a variants, AsCas12a showed the strongest activity, followed by MbCas12a and LbCas12a. Due to a PCR error, no band was visible for the negative control of RUNX1. However, since all negative controls originated from the same lysate, it is valid to compare this result with the other RUNX1 controls. Unfortunately, editing at the FANCF locus could not be confirmed due to cloning errors during the preparation of constructs.
Figure 1: Preliminary T7 Endonuclease I testing of Cas12a orthologs. T7EI samples were run on a 2% agarose gel in 1x TBE buffer dyed with gel red. SpCas9 targeting CCR5 functions as a benchmark for proper editing. For the Cas12a orthologs, LbCas12a, AsCas12a and MbCas12a, the three loci RUNX1, DNMT1 and FANCF were targeted. Editing is indicated by an extra band compared to the negative control.
Also due to a cloning mistake, we were only able to test editing for the FANCF locus with the LbCas12a and the AsCas12a, but not with the MbCas12a. As one can see, the editing for Dnmt1 and FANCF was non-existent, suggesting that either the chosen site for editing was inaccessible or the spacer being expressed in the crRNA was not able to properly bind to the Cas12a.
Following the T7EI assay results, we identified AsCas12a and MbCas12a as the most promising candidates for further development, while LbCas12a was excluded due to its poor performance. To improve the accuracy and speed of our proximity measurements, we decided to switch to an alternative assay, with faster turn-around time and better quantification. Future iterations will focus on developing new assays and successfully implementing the fusion Cas.
After extensive literature research, we identified this as a promising assay for quantifying editing efficiency. This assay measures the luminescence produced by Firefly luciferase, which decreases in proportion to editing efficiency at the target site. To account for variations in cell count and transfection efficiency, the luminescence is normalized to Renilla luciferase, which acts as an internal control. We hope this assay provides a straightforward and efficient approach for measuring editing activity in a high-throughput 96-well plate format.
We tested SpCas9, AsCas12a, and MbCas12a using sgRNA/crRNA constructs targeting the Firefly luciferase gene. To optimize the system, two variations of the reporter construct were designed: one where both luciferase genes were located on the same plasmid, and an alternative version with the Firefly and Renilla luciferase genes on separate plasmids.
The results from the Dual-Luciferase assay were consistent with the general trend observed in the previous iteration, with SpCas9 showing the highest editing efficiency. However, this assay provided qualitative resolution in distinguishing between the Cas12a orthologs. Contrary to the previous assumption, MbCas12a exhibited higher editing efficiency than AsCas12a. The one-plasmid system, with both luciferases on the same construct, showed significantly reduced editing efficiency compared to the two plasmid system (p=0.0002) but still provided a reliable distinction for the SpCas9. Strongest difference between the two reporter systems was visible when comparing SpCas9 editing. Both showed very little editing for AsCas12a but no significant differences. For MbCas12a, we can see better editing than AsCas12a with both reporters.
Figure 2: Assessment of a suitable reporter system for future dual luciferase assays. Firefly luminescence intensity measured 48 h after transfection. Normalized against renilla luminescence. On the x-axis, the samples Cas9, AsCas12a and MbCas12a are grouped by assay tested. Data is depicted as the fold change calculated from the non-targeting negative control over the targeting sample displayed as the mean +/- SD (n=3). Statistical analysis was performed using 2way ANOVA with Tukey’s multiple comparisons test, only significant differences within a group are shown. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.
The Dual-Luciferase assay successfully confirmed higher editing activity for SpCas9. Interestingly, this iteration revealed that MbCas12a outperformed AsCas12a, a finding that contradicts the previous iteration’s results. Because of this discrepancy, we continued testing both Cas12a orthologs. The assay provided an easy way to efficiently test out different constructs and transfection concentration, while still providing useful and quantitative data. Based on the overall stronger editing, the two plasmid system will be adopted for future assays.
Building on our promising results and further literature research (Kweon et al. (2017)), we fused together MbCas12a to SpCas9 and AsCas12a to SpCas9. These new Cas staple could provide one way to create functional staple, capable of bringing two DNA loci into close proximity. To ensure efficient nuclear delivery and expression, the Cas staples were modified with N-terminal (SV40 NLS) and C-terminal (nucleoplasmin NLS) nuclear localization signals.
Kweon, J., Jang, A.-H., Kim, D.-e., Yang, J. W., Yoon, M., Rim Shin, H., Kim, J.-S., and Kim, Y. (2017). Fusion guide RNAs for orthogonal gene manipulation with Cas9 and Cpf1. Nature Communications 8, doi:10.1038/s41467-017-01650-w.
The constructs were created by PCR amplifying the Vector of the previously tested SpCas9 construct. The PCR product included the plasmid backbone, CMV promoter, and both NLS sequences. The insert, containing the other Cas proteins, was also amplified from the previously tested plasmid. Golden Gate assembly was used to construct the final plasmids, utilizing overhangs with enzyme recognition sites introduced during PCR. The assembled plasmids were verified through sequencing and subsequently transfected into HEK293T cells for functional testing using the Dual-Luciferase Assay.
The Dual-Luciferase Assay was used to compare the editing efficiency of the fused Cas proteins with each other and their co-transfected single counterparts. In each construct, only one Cas protein targeted the Firefly luciferase gene, allowing for a direct comparison of editing efficiency. For the single co-transfected Cas proteins, the SpCas9 showed, as expected, the highest editing efficiency, shown by low relative luminescence units. Contradicting the T7EI, but being in-line with the previous dual luciferase, MbCas12a showed better editing efficiency compared to AsCas12a (p=0.005). The Fusion Cas proteins exhibited less editing efficiency compared to the single counterparts. Of the MbCas12a-SpCas9 fusion, SpCas9 showed higher efficiency. For the Cas12s in the fusion Cas, both proteins exhibit low editing activity, with no significant difference.
Figure 3: Comparison of AsCas12a and MbCas12a with a dual luciferase assay. Firefly luminescence intensity measured 48 h after transfection. Normalized against renilla luminescence. On the x-axis, the samples Cas9 + AsCas12a , Cas9 + MbCas12a, AsCas12a and MbCas12a are grouped by fused proteins. Co-transfected contains single Cas proteins, in contrast to the fusion Cas having the same cas proteins covalently bound to each other. Data is depicted as the mean +/- SD (n=3). Statistical analysis was performed using 2way ANOVA with Tukey’s multiple comparisons test. For better clarity, only significant differences within a group between the same Cas proteins are shown. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001
This iteration confirmed that our selected Cas fusions show editing. Based on the qualitative comparison to the single Cas proteins, improvements can still be made. Based on the results, we chose to omit the AsCas12a-SpCas9 fusion construct and only focus on MbCas12a-SpCas9 for further development. This decision was done on the basis of future plans, further comparing single co-transfected Cas proteins with fusion Cas proteins. We believe both to be suitable for stapling but further tests are needed beforehand.
After successfully showing editing activity of each part of the Cas staple, we aimed to test simultaneous editing activity. We adapted the previously established Dual-Luciferase assay, where MbCas12a continued to target the Firefly luciferase gene, and SpCas9 was now designed to target the N-terminus of Renilla luciferase. This modification enabled simultaneous monitoring of editing at both targets. For normalization, GFP was constitutively expressed as a new baseline. However, to ensure that GFP fluorescence would not interfere with the luminescence readout, a parallel test without GFP was conducted to serve as a control.
To introduce the SpCas9 target site, a well-characterized protospacer was cloned at the N-terminus of the Renilla luciferase gene. Since RenLuc would no longer be used as a baseline for normalization, the transfection ratio required adjustment. We matched the RenLuc plasmid amount to that used for Firefly luciferase (FFLuc) in previous experiments and similarly matched the GFP plasmid amount, which now served as the new baseline. Corresponding adjustments were made to the Cas protein and guide RNA amounts. For the test without GFP, additional staple plasmid was transfected, matching the amount of missing GFP.
The EGFP as baseline led to great variability between measurements. Sadly the negative controls for the co-transfected single Cas proteins did not yield the expected results. Therefore no significance could be calculated. However the fusion Cas negative controls worked great. Overall the results indicate significant editing for MbCas12a in the fusion Cas (p=0.0073) and for the SpCas9 in the fusion Cas in the single sample (p < 0.0001) and both editing (p < 0.0001). Unfortunately the MbCas12a had no significant editing in both editing samples.
Figure 4: Double cut dual luciferase assay with EGFP baseline for simultaneous editing assessment. On the left Firefly luminescence intensity is plotted and on the right renilla luminescence intensity is plotted, both measured 48 h after transfection. Normalized against EGFP fluorescence. On the x-axis the samples non targeting (nt), SpCas9, MbCas12a and Cas9 and Cas12a editing at the same time (both) , are grouped by single and fused proteins. Co-transformed contains single Cas proteins, in contrast to the fusion Cas having the same cas proteins covalently bound to each other. Firefly editing is showing Cas12a editing while lower renilla counts are caused by Cas9 editing. Data is depicted as the mean +/- SD (n=3). Statistical analysis was performed using 2way ANOVA with Sidaks’s multiple comparisons test. For better clarity, only significant differences within a group between the negative control and editing Cas proteins are shown. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001
Retransfecting the same assay lacking GFP showed similar results as the previous test. The control of the co-transfected Cas proteins’, once again did not work as expected. Although some data indicated successful editing, it lacked consistency. In contrast, the fusion Cas construct exhibited reliable controls and demonstrated highly significant editing at both targets, both individually and simultaneously (for all p < 0.0001), reinforcing the effectiveness of the Cas fusion strategy in this assay. Taking a qualitative look at the co-transfected single Cas proteins is strongly suggesting high editing rates for both MbCas12a and SpCas9 in the single and simultaneous samples.
Figure 5: Double cut dual luciferase assay without EGFP baseline for simultaneous editing assessment. On the left Firefly luminescence intensity is plotted and on the right renilla luminescence intensity is plotted, both measured 48 h after transfection. Normalized against EGFP fluorescence. On the x-axis the samples non targeting (nt), SpCas9, MbCas12a and Cas9 and Cas12a editing at the same time (both) , are grouped by single and fused proteins. Co-transformed contains single Cas proteins, in contrast to the fusion Cas having the same cas proteins covalently bound to each other. Firefly editing is showing Cas12a editing while lower renilla counts are caused by Cas9 editing. Data is depicted as the mean +/- SD (n=3). Statistical analysis was performed using 2way ANOVA with Sidaks’s multiple comparisons test. For better clarity, only significant differences within a group between the negative control and editing Cas proteins are shown. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001
Although the assay did not yield the expected results, we could still draw valuable conclusions out of it. Most importantly the Cas proteins can be fused together while retaining functionality, though a quantitative comparison to the individual Cas proteins is not yet feasible without proper controls. Interestingly the EGFP was not the sole problem, co-transfection of multiple Cas proteins appears to prove challenging with a proper negative control. Another observation we thought important, was reduced growth rate in cells harboring the fusion Cas protein. This could be due to the large construct size which may stress the cells. It is apparent when comparing the relative luminescence units (RLU) between groups, for the with fusion Cas transfected cells to have lower RLU in general because of metabolic burden expressing this large protein fusion.
To mitigate the apparent problems and further improve editing, we sought to create a stable cell line for continuous fusion Cas expression. To ensure successful genomic integration and stable expression, the open reading frame was expanded with a P2A peptide and puromycin resistance gene (PuroR) downstream of the fusion Cas protein. The P2A peptide facilitates ribosomal skipping, allowing both the Cas fusion protein and PuroR to be expressed from the same mRNA while being translated as separate proteins.
The plasmid harboring the fusion Cas protein and a second plasmid containing the P2A-Puromycin resistance sequence were PCR-amplified and fitting overhangs for subsequent Golden Gate assembly were introduced. After verifying the correct assembly via sequencing, the final plasmid construct was transfected into HEK293T cells. Selection pressure was applied by addition of puromycin. As a negative control the fusion Cas plasmid lacking the selection gene was transfected.
The creation of the stable cell line proceeded as expected. After initial growth in 6-well plates, the cells were successfully transferred to T75 flasks for continued expansion. Due to a noticeably reduced growth rate, a weekly splitting schedule was implemented.
The stable cell line was used to repeat both the standard dual luciferase assay (iteration 3) and the modified version for simultaneous editing (iteration 4). While the data exhibited high variability, some trends could still be discerned for the editing to work in the co-transfected Cas12a (p=0.0018). As for the stable cell line small changes in luminescence intensity with no significance still allow a qualitative assessment of the data. Overall the luciferase expression, and changes thereof, are smaller than compared to the co-transfected samples.
Figure 6: Single cut dual luciferase assay repeating preliminary tests. Firefly luminescence intensity measured 48 h after transfection. Normalized against renilla luminescence. On the x-axis the samples nt, Cas9 and Cas12a are grouped by single and fused proteins. Co-transformed contains single Cas proteins, in contrast to the fusion Cas having the same Cas proteins covalently bound to each other. Data is depicted as the mean +/- SD (n=3). Statistical analysis was performed using 2way ANOVA with Tukey’s multiple comparisons test. Only significant differences within a group are shown. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001
The modified luciferase assay, designed to measure simultaneous cutting efficiency, provided more consistent results. The co-transfection of SpCas9 (p < 0.0001) and MbCas12a (p < 0.0001) demonstrated highly significant editing for both Cas proteins, both individually and simultaneously (for both Cas proteins p < 0.0001). The stable cell lines did not show significant editing activity, with only minor changes in expression levels allowing just vague qualitative interpretation.
Figure 7: Double cut dual luciferase assay testing fusion Cas in stable cell line. On the left Firefly luminescence intensity is plotted and on the right renilla luminescence intensity is plotted, both measured 48 h after transfection. Normalized against EGFP fluorescence. On the x-axis the samples non targeting (nt), SpCas9, MbCas12a and Cas9 and Cas12a editing at the same time (both) , are grouped by single and fused proteins. Co-transformed contains single Cas proteins, in contrast to the fusion Cas having the same cas proteins covalently bound to each other. Firefly editing is showing Cas12a editing while lower renilla counts are caused by Cas9 editing. Data is depicted as the mean +/- SD (n=3). Statistical analysis was performed using 2way ANOVA with Tukey’s multiple comparisons test. For better clarity, only significant differences within a group between the negative control and editing Cas proteins are shown. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001
Further optimization is required to determine the ideal ratios for the fusion Cas constructs. Integration of the fusion Cas proteins into the genome resulted in lower expression levels, which may not be sufficient to achieve significant editing activity. This iteration all the negative controls for the co-transfected Cas proteins worked beautifully, finally allowing for proper significance testing. These can now function as a proper comparison for the fusion Cas.
To further optimize the dual luciferase assay careful consideration went into the transfection plan and plasmid ratios needed. Based on literature information we factored in differences in maturation time and degradation. With these adjustments we hope to improve the clarity of the assays output.
To ensure a balanced production of Cas protein and guide RNA, we calculated the necessary gene copy numbers to achieve a 1:1 ratio of protein to gRNA. One gene copy was equated to one plasmid, with nanogram amounts of plasmid DNA representing the respective copy numbers for transfection. Due to differing plasmid sizes, we adjusted the nanogram quantities to maintain proportional representation and scaled them down to the maximum possible nanogram amount for transfection. Since the fusion Cas was stably expressed in the generated cell line, we also had to scale the transfected amount based on the cell line generation. Notably, the transfected plasmid per cell in the 6-well plates was half of what we would typically use in a 96-well plate, so all subsequent calculations were adjusted accordingly to maintain consistent expression across experiments.
The data for the co-transfected Cas proteins showed efficient editing for both Cas9 (p=0.0003) and Cas12a (p=0.0005) to almost complete depletion. The stable cell line did not show such promising results, with only minor changes for both cas enzymes, with strong variation within replicates.
Figure 8: Single cut dual luciferase assay improvement of preliminary tests. Firefly luminescence intensity measured 48 h after transfection. Normalized against renilla luminescence. On the x-axis the samples nt, Cas9 and Cas12a are grouped by single and fused proteins. Co-transformed contains single Cas proteins, in contrast to the fusion Cas having the same Cas proteins covalently bound to each other. Data is depicted as the mean +/- SD (n=3). Statistical analysis was performed using 2way ANOVA with Tukey’s multiple comparisons test. Only significant differences within a group are shown. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001
The modified dual luciferase assay revealed similar outcomes. The co-transfection showed strong editing efficiency for both Cas12a and Cas9, individually (Cas12a p=0.0044; Cas9 p < 0.0001) and simultaneously (Cas12a p=0.0048; Cas9 p < 0.0001). The stable cell line did not show significant editing activity for MbCas12a separately and simultaneous, but significant editing for SpCas9 (seperate p = 0.0049; simultaneous p = 0.002) is shown. Again trends are visible for MbCas12a which will be further discussed in the learning from this iteration.
Figure 9: Double cut dual luciferase assay further testing fusion Cas in stable cell line. On the left Firefly luminescence intensity is plotted and on the right renilla luminescence intensity is plotted, both measured 48 h after transfection. Normalized against EGFP fluorescence. On the x-axis the samples non targeting (nt), SpCas9, MbCas12a and Cas9 and Cas12a editing at the same time (both) , are grouped by single and fused proteins. Co-transformed contains single Cas proteins, in contrast to the fusion Cas having the same cas proteins covalently bound to each other. Firefly editing is showing Cas12a editing while lower renilla counts are caused by Cas9 editing. Data is depicted as the mean +/- SD (n=3). Statistical analysis was performed using 2way ANOVA with Tukey’s multiple comparisons test. For better clarity, only significant differences within a group between the negative control and editing Cas proteins are shown. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001
The Fusion Cas stable cell line did not function as expected. We suspect that prolonged exposure to selection pressure resulted in the loss of integrated gene copies due to genomic repair mechanisms. This could explain the lower-than-expected expression and editing efficiency observed in the stable cell line. Consequently, the calculated plasmid ratios for the Fusion Cas system need further tweaking. Assessing the data from figure 8 this becomes apparent taking a qualitative approach for data evaluation. Nonetheless this line of experiments gave us an in-depth understanding on different assay systems and the Cas proteins accepting fusion proteins. We used these learnings and combined them with our fusion gRNA learnings to create functional protein staples with catalytically inactive (dCas) versions.
Building on previous experience of engineering fusion gRNAs (fgRNAs) and fusion Cas systems, the aim of this engineering cycle was to design protein staples capable of bringing specific DNA sequences into close proximity. As the literature on this topic is sparse, all reasonable construct combinations of sgRNA, fgRNA, single Cas and fusion Cas were tested. The first assay we designed to demonstrate such proximity was based on a fluorophore quencher pair. In this system, a fluorophore and a quencher are covalently bound at opposite ends of a double-stranded DNA (dsDNA) sequence. When the Cas staple brings the two ends of the DNA strand together, the quencher should be in close proximity to the fluorophore and thus reduce fluorescence intensity measurably. Careful consideration was given to the design of a suitable fluorophore-quencher pair. Some considerations included thermal stability to allow incorporation into larger DNA fragments by PCR amplification. The quenching strength and availability as ssDNA modifications were also considered. The fluorophore quencher pair of choice was 6-FAM and Iowa Black QG.
Figure 1: Design and theory behind fluorophore quencher assay.Depicted on the left is the dsDNA fragment amplified by PCR with primers containing the fluorophore 6-FAM and the quencher Iowa Black FQ. Introducing the Cas staple, fgRNA with single dCas proteins or sgRNAs with fusion dCas or fgRNA with fusion dCas, as shown on the right, to this dsDNA piece would, as we theorize, bring the fluorophore and quencher into close proximity resulting in a reduction of fluorescence.
A catalytically inactive version of the fusion Cas construct was created through site-directed mutagenesis PCR, followed by Gibson assembly. The DNA fragment used in the assay was derived from a plasmid containing previously validated protospacers. This fragment was amplified by PCR using primers modified with a fluorophore (6-FAM) or a quencher (Iowa Black QG) attached to the 5’ OH group. Special care was taken during the assembly process to maintain mild reaction conditions and avoid exposure to light to preserve the integrity of the fluorophore-quencher pair.
HEK293T cells were co-transfected with various Cas and gRNA constructs along with the linear dsDNA fragment. After 8 hours, the media was replaced to remove any untransfected fragments, and fluorescence intensity was measured every 4 hours. At the 24-hour mark, measurements were taken every 12 hours, continuing until the 48-hour mark. A steady decline in fluorescence intensity was observed across all samples, indicating no evidence of successful DNA stapling.
Figure 2: Fluorophore quencher assay for proving proximity induction of DNA by Cas staples. 6-FAM fluorescence measured in time intervals after a media change 8h after transfection. After 48h the measurement was stopped. As controls on dsDNA fragment without a quencher is used, while the negative control consists of single Cas proteins with sgRNAs. Samples range from the parts of the Cas staple missing(fgRNA, Cas9 + fgRNA, Cas12a + fgRNA) to all parts of the Cas staple present ( 0nt fgRNA, 20nt fgRNA, 30nt fgRNA, 40nt fgRNA). Data is depicted as the mean (n=3). Statistical analysis was not performed.
The assay proved hard to optimize and therefore was not pursued further. Alternative assays were taken into account, with the focus to find a suitable system closer to naturally occurring mechanisms.
One promising application of our tool could be enhancer hijacking. To simulate this, we modified the already existing VP64 activated dual luciferase assay. Instead of directly fusing the VP64 to one of the Cas proteins, a VP64-Gal4 fusion protein was designed. Two plasmids will be stapled together, one of the plasmids has the dCasd12a target site next to the Gal4 target site (UAS), and another plasmid with the dCas9 target site next to the minimal promoter and firefly luciferase gene.
We hypothesized that a linker between the spacer sequences would be essential for simultaneous binding of both Cas proteins to the fgRNA. AlphaFold3 predictions suggested a minimum linker length of 20 nt to facilitate this interaction. Consequently, fgRNAs with linker lengths of 20 nt, 30 nt, and 40 nt were designed for testing in the wet lab.
The amounts transfected were calculated as already mentioned in the fusion Cas engineering cycle, while employing an extra factor having fgRNA and the Cas proteins be transfected in a 1:2 ratio.
Figure 3: Design and theory behind simulated enhancer hijacking assay. Depicted on the left is the reporter plasmid and the Gal4 binding plasmid stapled together by Cas staples. On the right a closer look at a Cas staple stapling the two plasmids together with the Gal4-VP64 binding and acting as a transactivator driving luciferase expression forward is shown.
The Gal4 binding system was constructed by modifying an existing plasmid containing 5 UAS sites. Oct1 binding sites were introduced by annealing three oligo pairs, which were subsequently ligated into the digested vector. New fgRNAs targeting the Oct1 and tetO sites were synthesized as DNA oligos. Following annealing, these oligos were cloned into the entry vector using Golden Gate assembly to create the desired constructs for further testing. As described in the fusion Cas engineering cycle, special care was taken to calculate transfected plasmid amounts. This included adjusting for expression levels and decay rates, ensuring the intended 1:2 ratio of fgRNA to Cas proteins.
Firefly luciferase luminescence intensities were normalized to Renilla luciferase to account for variations in cell count. For the fgRNAs with single Cas proteins, the 0 nt linker exhibited no difference in fluorescence compared to the control. A clear trend emerged, showing increasing luminescence with longer linker lengths, with significant differences observed at linker lengths of 30 nt and 40 nt (p = 0.0185 and p = 0.0011, respectively). The fusion Cas constructs showed no change in luminescence for the sgRNA and 0 nt linker. Interestingly, fgRNAs with 20 nt and 30 nt linkers outperformed those with a 40 nt linker, differing from the results seen with the single Cas proteins.
A
B
Figure 4: Dual luciferase assay on simulated enhancer hijacking. Firefly luminescence intensity measured 48 h after transfection. Normalized against renilla luminescence. A shows stapling with single Cas proteins and B with fusion Cas. On the x-axis in red the negative control is followed by the samples consisting of fgRNAs with the different linker lengths 0 nt, 20 nt, 30 nt and 40 nt. Whereas for the fusion Cas one extra sample with the sgRNAs was measured. Data is depicted as the mean +/- SD (n=3).
This assay successfully demonstrated that a fgRNA can be bound by two dCas proteins simultaneously and staple together two separate DNA fragments, resulting in transactivation for both fgRNA-single dCas and fgRNA-fusion dCas constructs. For the fgRNA-single dCas system, a steady increase in luminescence was observed with increasing linker lengths, suggesting improved Cas binding. For future experiments, longer linker lengths may be tested to further optimize the construct. The fgRNA-fusion dCas system, however, exhibited overall lower activity compared to single Cas constructs. The significant drop in activity at the 40 nt linker length indicates potential structural constraints in fgRNA binding caused by the peptide linker. In future iterations, altering peptide linkers could further enhance activity and improve system performance. The experiment will be repeated in biological replicates to acquire more significant data.
We designed a fluorescence-based readout system to test if different peptide linkers could be cleaved by human cathepsin B in vivo. This would allow for the selective activation of our PICasSO system in tumor microenvironments overexpressing cathepsin B.
Our fluorescence readout system was comprised of a plasmid encoding mCherry with a Gal4 upstream activating sequence (UAS) and a fusion protein that consists of the DNA-binding domain (DBD) of Gal4, one of the five different peptide linkers (GFLG, FFRG, FRRL, VA, FK), and the transactivator domain VP64. Binding of Gal4-DBD-Linker-VP64 to the UAS would induce overexpression of mCherry. Consequently, cathepsin B cleavage of the peptide linker would reduce the expression of mCherry.
We transfected our genetic constructs into HEK293T cells. The fluorescence intensity of mCherry was measured 48 hours after transfection. Our initial tests did not result in the unambiguous identification of a cathepsin B-cleavable peptide linker (see Fig. 1). For all linkers, we did not observe a large decrease in fluorescence intensity between the negative control and test conditions. In some conditions, the fluorescence intensity even increased between the negative control and test conditions.
Figure 1: Fluorescence Readout After 48 Hours for Five Different Peptide Linkers and Three Different Conditions. The fluorescence intensity for mCherry was measured for five different linkers. The negative control was not transfected with the plasmid encoding cathepsin B. The fluorescence intensity of the negative control was set to one. Two different test conditions were investigated, in which either 30 ng or 60 ng of the plasmid encoding cathepsin B were transfected.
Since cathepsin B is a lysosomal protease that is normally only active in the lysosome and the extracellular environment but not in the cytosol, we decided to change the native amino acid sequence of cathepsin B. Upon further investigation of the lysosomal maturation process of cathepsin B, we chose to express a truncated and mutated version of cathepsin B.
We truncated the native cathepsin B amino acid sequence N-terminally by the first twenty amino acids. This would delete a signal peptide that normally facilitates translation of cathepsin B in the rough endoplasmic reticulum ensuring translation of cathepsin B in the cytosol. Additionally, we introduced three point mutations (D22A, H110A, R116A) in the amino acid sequence of cathepsin B. This has been shown to increase the activity of cathepsin B at higher pH values by disrupting the interactions of an occluding loop with the substrate binding pocket of cathepsin B.
We used overlap extension PCR to clone our truncated and mutated version of cathepsin B based on our original nucleotide sequence of cathepsin B. Additionally, we used the same plasmid backbone to overexpress our new construct in HEK293T cells.
We performed the same fluorescence readout assay in HEK293T cells that we also used for full-length cathepsin B. The fluorescence intensity of mCherry was measured 48 hours after transfection. However, we observed no decrease in fluorescence between our negative control and test conditions, indicating that Gal4-DBD-Linker-VP64 was not cleaved (see Fig. 2). Additionally, we performed a western blot, where the bands for the truncated and mutated version of cathepsin B were barely visible or absent altogether, indicating lower protein expression compared to the wild type (see Fig. 3). Another key insight from this experiment was that this version of cathepsin B was not activated by cleavage in the cell, as no additional protein bands were observed.
Figure 2: Fluorescence Readout for the Truncated and Mutated Version of Cathepsin B.
Figure 3: Western Blot of Two Versions of Cathepsin B With and Without Doxorubicin. From left to right: protein ladder, wild-type (wt) cathepsin B with (+) and without (-) doxorubicin, truncated and mutated version of cathepsin B with (+) and without (-) doxorubicin.
Since our truncated and mutated version of cathepsin B did not seem to be active in the cytosol, we decided to go back to wild-type cathepsin B. Therefore, we focused on improving the activity of wild-type cathepsin B in the cytosol.
Instead of changing the amino acid sequence of cathepsin B, we decided to induce lysosomal escape of cathepsin B by treating the transfected cells with the cytostaticum doxorubicin.
After consulting the literature, we decided to treat the HEK293T cells with 500 nM doxorubicin 24 hours after transfection.
We used the same fluorescence readout assay that we had used in our initial testing (see Fig. 4). 24 hours after transfection, the cells were treated with 500 nM doxorubicin. The fluorescence intensity of mCherry was measured 48 hours after transfection. We observed a significant decrease in fluorescence for the GFLG linker between the negative control and the two test conditions.This indicates successful cathepsin B cleavage of the linker in vivo.
Figure 4: Fluorescence Readout After 48 Hours for Five Different Peptide Linkers and Three Different Conditions. The fluorescence intensity for mCherry was measured for five different linkers. The negative control was not transfected with the plasmid encoding cathepsin B. The fluorescence intensity of the negative control was set to one. Two different test conditions were investigated, in which either 30 ng or 60 ng of the plasmid encoding cathepsin B were transfected.
Through these tests, we identified one peptide linker (GFLG) that was consistently cleaved in the cytosol by cathepsin B. This enables us to functionalize our PICasSO system to be selectively active in microenvironments, such as cancerous tissue, that overexpress cathepsin B.
To create a robust FRET measurement system, we first needed a simple staple to bring two DNA strands into proximity and identify suitable FRET pairs for efficient measurement. Our design process began with extensive literature research to select well-characterized DNA-binding proteins and fluorescent pairs that were proven to work in E. coli without compromising binding efficiency when fused. Among the potential candidates—Oct1-DBD, Gal4-DBD, tetR, and NFKB-DBD—Oct1-DBD was the only naturally monomeric protein. We hypothesized that ths monomeric nature is important to ensure stoichiometric binding at staple target sites.
We selected Oct1-DBD as one DNA-binding protein due to its proven strong expression and binding affinity in E. coli. Our second protein was an engineered version of the well characterized Tetracycline Repressor (tetR) that forms a functional monomer. This scTetR is a fusion of two tetR proteins with a flexible linker which was described to maintain the same DNA-binding affinity and specificity as wild-type tetR.
For the FRET pairs, we established the following key criteria: small size, monomeric, minimal photobleaching, high fluorescence and quantum yield, and emission outside the blue and near-infrared spectra to reduce cell damage and background fluorescence. After reviewing literature and using FPbase to predict FRET efficiency, we chose mNeonGreen and mScarlet-I for their superior performance in FRET applications both reported in literature and calculated with FPbase.
We constructed a two-plasmid system: an expression plasmid and a folding plasmid. The expression plasmid contained the following required proteins: scTetR-Oct1 fusion (staple), scTetR-mScarlet-I (FRET acceptor), and Oct1-mNeonGreen (FRET donor), additionally the vector had 15 repeats of the tet response element (TRE) to ensure efficient tetR binding. The folding plasmid was designed with 15 repeats of the Oct1 binding motif. To ensure compatibility, special consideration went into deciding the origin of replications (ori), we selected a high-copy pMB1 origin for the expression plasmid and a medium-copy p15A origin for the folding plasmid.
To maintain balanced expression and minimize metabolic burden, we designed a polycistronic expression cassette under the T7 promoter. This allowed the transcription of one mRNA encoding all three proteins, each with its own ribosome binding site (BBa_B0030). To facilitate future modifications, we included restriction sites between the ORFs and at strategic locations. Gene synthesis was used to generate the inserts, which were cloned into the backbones using Gibson assembly.
As controls, we constructed a positive control by fusing mScarlet-I and mNeonGreen to observe direct FRET efficiency, and a negative control by using a folding plasmid lacking the Oct1 target sequence.
Fluorescence intensity for mNeonGreen, mScarlet-I, and FRET was measured 18 hours post-incubation with varying IPTG concentrations (0.8 mM to 0.0125 mM).
Figure 1: Fluorescence measurement of first simple staple construct.Fluorescent measurement, normalized to cell count, of mNeonGreen (ex. 490 nm, em. 530 nm), mScarlet-I (ex. 560 nm, em. 600 nm), and FRET (ex. 490 nm, em. 600 nm) in E. coli, 18 h after induction with different IPTG concentrations. Data is presented as mean ± SD.
The positive control exhibited strong fluorescence for both proteins and FRET, indicating that the system was functional in principle. However, the constructs expressing the staple and FRET-pairs didn’t exhibit any fluorescence for all three measurements, indicating an issue with protein expression
The lack of fluorescence in the polycistronic constructs suggested problems in transcription or translation possibly due to the metabolic burden of expressing large, complex constructs. This hypothesis is also supported by the positive control where increasing fluorescence could be measured with decreasing IPTG concentrations, reinforcing the idea that overexpression might be causing stress and inhibiting efficient protein production. Importantly this initial testing showed that the FRET pairs are compatible resulting in measurable fluorescence readout, motivating us to further pursue this line of experiments.
To address the potential issue of overexpression causing metabolic burden in our cells, we decided to test the impact of reducing transcriptional activity using a mutated T7 polymerase with lower activity. This approach aimed to reduce cell stress, but should still produce sufficient protein for fluorescent measurement. Since the mutated T7 was on a plasmid with the same origin of replication as the folding plasmid, we opted to co-express it only with the expression plasmid, using BL21 cells that lacked the genomic T7 copy.
The mutated T7 polymerase was on a plasmid with the same ori as the folding plasmid, making it incompatible for co-expression with the full system. As a result, we co-expressed the mutant T7 polymerase only with the expression plasmid in BL21 cells that lack a genomic T7 copy, focusing solely on evaluating expression levels.
Similar to the first iteration, we measured fluorescence intensity for mNeonGreen, mScarlet-I, and FRET after inducing the constructs with IPTG. Fluorescence was successfully detected for mNeonGreen in both ori variants, indicating stable expression. However, mScarlet-I was still not expressed in the staple constructs, suggesting an issue specific to this protein.
Figure 2: Fluorescent measurement of expression plasmid with mutant T7 polymerase.
Fluorescent measurement, normalized to cell count, of mNeonGreen (ex. 490 nm, em. 530 nm), mScarlet-I (ex. 560
nm, em. 600 nm), and FRET (ex. 490 nm, em. 600 nm) in E. coli, 18 h
after induction with different IPTG concentrations. Data is presented as mean ± SD.
Fluorescence could be detected for mNeonGreen, indicating successful expression, but mScarlet-I still failed to show any signal. This suggests that the problem with scTetR-mScarlet-I is not solely due to the high expression levels or transcriptional activity, as reducing transcription did not resolve the issue. Based on the location of scTetR-mScarlet-I in the polycistronic operon (before Oct1-mNeonGreen), the issue seems to be related to translation or protein folding rather than transcription. Further investigations into protein stability or folding will be required to address this.
In the previous iteration, reducing transcriptional activity through a mutated T7 polymerase did not resolve the issue of low ScTetR-mScarlet-I expression. This led us to hypothesize that the burden of a two plasmid system (one of them being high copy) and expression of multiple proteins, is still too high. Therefore, in this iteration, we aimed to reduce the metabolic load by lowering the copy number of the expression plasmid. To achieve this, we designed two alternative constructs: For the first, the rop gene was introduced into the plasmid backbone. The Rop protein is known to regulate plasmid replication and should reduce the plasmid count to approximately 15-20 copies per cell. For the second modification the pMB1 ori was replaced with the low copy pSC101 ori. This should reduce the copy number to 5-10 copies per cell.
To test the impact of lowering plasmid copy number, we cloned the rop protein into the expression plasmid via restriction and ligation cloning, and similarly replaced the pMB1 origin with pSC101 using PCR amplification and cloning. After transformation into Top10, colonies were picked and plasmid purified. Correct assembly was verified by whole plasmid sequencing.
The co-transformation of the modified expression plasmids with the folding plasmid in BL21 did not work. Next we first transformed the expression plasmid, which worked and could be propagated stably. These cells were then made competent following our protocol and used to transform the folding plasmid. The transformation worked but cells did not show stable growth in selection media. These experiments were repeated multiple times to rule out possible mistakes.
The co-transformation of the modified expression plasmids with the folding plasmid consistently failed, and stable growth in selection media was not achieved. These issues suggest that lowering plasmid copy number, either through the rop gene or the pSC101 origin, may have negatively impacted plasmid stability or protein expression. Despite successful individual transformations, co-expressing both plasmids may have led to incompatible replication dynamics or insufficient protein levels due to reduced copy numbers. Given the repeated failures and time constraints, we discontinued this line of experiments.
The previous iterations showed that expression, especially of ScTetR was not possible. To improve expression and reduce cellular burden, we replaced ScTetR with the standard tetR for both the staple construct and FRET acceptor. This substitution likely induces homo-dimerization for tetR binding, which was deemed acceptable.
The modified operon was assembled by restriction and ligation cloning.
The modified constructs were again tested with different IPTG concentrations. This experiment can be found on our results page. In short this final version was successful to induce proximity between the folding and expression plasmid and measure it through the FRET interaction.
The FRET assay successfully demonstrated induced proximity between the folding and expression plasmid, resulting increase in FRET interaction and fluorescent measurement. Moving forward, refining the measurement setup will be necessary to achieve quantitative assessment of successful stapling and proximity. Moving forward, applying more advanced FRET analysis techniques, such as those described by Hochreiter et al. (2019) and Coullomb et al. (2020), would allow for precise quantification of protein interactions, stoichiometry, and affinity. Future iterations could implement acceptor photobleaching or time-resolved fluorescence measurements to further enhance the sensitivity and accuracy of our system, facilitating real-time analysis of DNA-protein interactions.
Coullomb, A., Bidan, C. M., Qian, C., Wehnekamp, F., Oddou, C., Albigès-Rizo, C., Lamb, D. C. & Dupont, A. (2020). QuanTI-FRET: a framework for quantitative FRET measurements in living cells. Scientific Reports, 10(1). https://doi.org/10.1038/s41598-020-62924-w
Hochreiter, B., Kunze, M., Moser, B. & Schmid, J. A. (2019). Advanced FRET normalization allows quantitative analysis of protein interactions including stoichiometries and relative affinities in living cells. Scientific Reports, 9(1). https://doi.org/10.1038/s41598-019-44650-0
For constructing our pNeae2_7D12 plasmid, we first decided to follow a classical restriction-ligation cloning strategy as was done by Pinero-Lambea et al., (2015). We introduced the 5’ and 3’ overhangs for SfiI and NotI recognition sites into the 7D12-nanobody gene block (DSg3). An in silico simulation of the agarose gel after double restriction by SfiI and NotI indicated that we should observe one band of 5801 bp and a second, smaller band of 31 bp, which would not be visible on a 1% agarose gel.
Piñero-Lambea, Carlos, et al. "Programming controlled adhesion of E. coli to target surfaces, cells, and tumors with synthetic adhesins." ACS synthetic biology 4.4 (2015): 463-473.
We first performed PCR amplification of DSg3. Following the restriction digest of the pNeae2 backbone and DSg3 with SfiI and NotI, we analyzed the digestion products by gel electrophoresis (1% agarose). We observed two bands: one above 5000 bp and another at around 3000 bp. While we hypothesized that the upper band was the expected one, we could not explain the existence of the lower band. As the reaction conditions were not optimal for both restriction enzymes, it is reasonable to assume that the smaller product resulted from undesired star activity of one of the enzymes. We chose to continue the cloning procedure with the larger band, having our expected size. After gel extraction, we went on to ligate the digestion products overnight at 16 ℃ using T4 DNA ligase. The next day, the ligation product was transformed into chemically competent E. coli cells, plated on LB agar supplemented with chloramphenicol, and incubated at 37 ℃.
Figure 1: Gel electrophoresis analysis of double restricted 7D12 nanobody and pNeae2 backbone. The first lane represents the DNA ladder, the second lane represents the 7D12 nanobody gene block restricted with SfiI and NotI. Lanes 3-6 represent four samples of the pNeae2 backbone digested with SfiI and NotI.
Several colonies grew on the selective LB agar plate and we assumed that the cloning was successful. To solidify our assumption, we ran a test digest with EcoRI. In case of the backbone without the nanobody insert, we expected only one cut site, corresponding to only one band of the size 5832 bp, and in case of a successful insertion, we would observe two bands: 4050 bp and 2148 bp. The test digest clearly showed a single band, which meant that the backbone had religated despite the incompatible overhangs.
Figure 1: Gel electrophoresis analysis of the test digest of pNeae2_7D12 plasmid. The digest was performed with the EcoRI restriction enzyme. The first lane represents the DNA ladder, lanes 2-7 represent six samples of the cloned pNeae2_7D12 and the third lane represents the negative control, which was the pNeae2 backbone.
Initially we expected to observe a single band after digestion of the pNeae2 backbone. This expectation laid on the assumption that the digest had been fully completed. If undigested parts of the backbone remained, this would result in a second band. However, the size of this band would be 5832 bp and thus indistinguishable from the expected band of 5801 bp. Consequently, excising only the digested backbone and leaving behind the incompletely digested (e.g. digested by only one of the enzymes) plasmid would be nearly impossible. Therefore, the growth of single colonies on the selective plate after the transformation might have been due to retransformation of the circularized backbone without the needed insert. In conclusion, we decided to change the cloning strategy.
In the initial phase, our objective was to investigate the relationship between general scientific knowledge and attitudes toward genetically modified organisms (GMOs). We hypothesized that individuals with a higher level of scientific knowledge would display a more positive attitude toward GMOs. This hypothesis was grounded in the notion that increased scientific literacy could lead to a better understanding of the safety, utility, and potential risks associated with GMOs. Consequently, our survey aimed to explore how a participant's grasp of basic scientific concepts influenced their perception of GMOs.
We began by constructing a questionnaire including 35 questions , divided into two sections: 20 questions assessing general scientific knowledge and 15 questions concerning attitudes toward GMOs. The questions were designed to capture various levels of understanding and opinion, utilizing a combination of open questions, multiple-choice questions, and Likert-scale questions (ranging from strongly disagree to strongly agree in 5 steps). Specifically, we assessed the participants' understanding of key principles in biology and environmental science, as well as their perceptions of safety, environmental impact, and ethical concerns. The development of our questionnaire was based on a review of relevant literature and best practices from previous studies on public understanding of science and technology
To ensure the robustness of our survey design, we interviewed the Chair of Psychological Methodology and Diagnostics at Ludwig-Maximilians-University Munich, Prof. Dr. Markus Bühner. Together, we reviewed the content and structure of our survey and refined the questions to ensure they were scientifically valid and easy to interpret. Prof. Dr. Bühner's expertise in survey design and methodology provided invaluable feedback, improving the reliability and comparability of responses from our survey.
The feedback from our expert consultation led to several key changes in our design. First, we were advised to eliminate open questions, as they would be difficult to compare and quantify across a large sample group. Open responses often present challenges in standardizing interpretations and may introduce biases in analysis. Additionally, Mr. Bühner suggested limiting the total number of questions to 30, as a more concise survey would likely improve response rates and participant engagement. Finally, he recommended conducting a pre-survey to define and balance the difficulty of the scientific knowledge questions. The target difficulty range of a question should be defined as 20% to 80% correct answers. He also advised us to gather responses from at least 20 participants for this pre-survey to analyze the distribution of difficulty effectively.
In response to the feedback received during the first iteration, we designed a pre-survey focused on evaluating the difficulty of the scientific knowledge questions. The primary goal of this phase was to ensure that the questions effectively distinguished between varying levels of knowledge while maintaining a reasonable level of challenge. Drawing from previous studies and literature, we revised the content of the questions to reflect key scientific principles that were accessible yet discerning. Analyzing the pre-survey responses would allow us to fine-tune the questions and ensure that they would yield a balanced assessment of scientific knowledge.
We designed the pre-survey to include 30 true/false questions targeting general scientific knowledge. The questions were derived from school textbooks to ensure they reflected an appropriate level of complexity. We developed a web-based format for the pre-survey to allow easy access and completion by participants.
The pre-survey was distributed online in April 2024 to a randomly selected group of 20 participants. After collecting the responses, we analyzed them to determine the distribution of correct answers for each question.
The analysis revealed that 15 of the 30 questions met the desired difficulty range, with 20% to 80% of participants answering them correctly and were retained for the final survey. Additionally, several questions were identified as being confusing or ambiguous in their wording, leading to varied interpretations. These were reworded to improve clarity and ensure participants could easily understand and respond accurately. This learning phase underscored the importance of question clarity in surveys assessing knowledge.
Having refined the scientific knowledge questions, we proceeded to design the main survey, which aimed to test the original hypothesis. The survey consists of two sections: the first assessed general scientific knowledge, and the second focused on participants' attitudes toward GMOs. To contextualize these responses, we also included demographic questions related to the participant's age, gender, and geographic location.
Using the online platform LimeSurvey, we developed the main survey, which was distributed via social media channels and personal networks to maximize participation and ensure a diverse sample of respondents
The survey was distributed across Germany from June to September 2024. As our questionnaire was distributed online participants from various regions could engage with the content at their convenience. In total, 823 valid responses were collected, with participants representing nearly all German states except Schleswig-Holstein. The broad geographic representation provided a rich dataset for analysis, allowing us to explore potential regional variations in scientific knowledge and attitudes toward GMOs.
The distribution of responses indicated that our outreach efforts engaged with a diverse audience across Germany. This phase also reinforced the utility of web-based surveys in reaching a wide audience, though it underscored the importance of ensuring representative participation from all regions.
With the data collection complete, we focused on analyzing the relationship between scientific knowledge and attitudes toward GMOs. We employed correlation analysis to examine the association between the Knowledge and Attitude Score.
Participants were classified into three knowledge groups based on their responses to the 15 knowledge questions. A point was awarded for each correct response, with a maximum possible score of 15. Participants with scores between 11 and 15 were classified as having High Knowledge, those scoring between 6 and 10 as Moderate Knowledge, and those scoring between 0 and 5 as Low Knowledge. This classification was based on equally dividing the range of scores to ensure an objective grouping. The attitude toward GMOs was assessed using a Likert scale, with 6 negative and 9 positive statements. For negative questions, disagreement indicated a positive attitude toward GMOs, while for positive questions, agreement was indicative of a positive attitude. The responses were scored accordingly, to create a composite Attitude Score for each participant.
Correlation analyses were conducted to explore the relationship between participants' Knowledge Scores and their attitudes toward GMOs. These analyses allowed us to test whether there was a statistically significant association between the two variables, thereby addressing our initial hypothesis.
The results indicated a negative correlation (r= -0.62, p-value =0.000) between Knowledge Score and Attitude Score, suggesting that individuals with lower scientific knowledge tend to hold more negative attitudes toward GMOs. This finding confirmed our original hypothesis, demonstrating the critical role of scientific education in shaping public opinion on controversial topics like GMOs. The insights gained from this analysis can inform educational initiatives into context, emphasizing the importance of fostering scientific literacy to promote informed decision-making.
