Amplification

CPU_CHINA prepared the template of LINC00857. It was selected as cancer biomarker by our team. We synthesized the DNA template through the company and obtained RNA by in vitro transcription as shown in Figure 1. Then we measured RNA concentration by UV spectrophotometry and got the RNA’s concentration of 4792.608ng/μL, which is shown in Figure 2.

Figure 1. RNA in-vitro transcription.

Figure 2. Ultraviolet determination of RNA concentration.

Then we used Reverse Transcription-Recombinase Polymerase Amplification (RT-RPA) to generate a large number of dsDNA, meanwhile amplifying the molecular signal. The RPA primers that we designed for LINC00857 is shown in Figure 3.

Figure 3. The primer sequence used for RT-RPA.

The RT-RPA process lasted for 30min and the results of which was verified by electrophoresis (Figure 4). The electrophoretic pattern of RT-RPA showed that after 30 minutes of amplification, nucleic acids were massively amplified. However, different concentrations of the template did not show significant differences in the amount of amplification products. That's probably because the amplification time was too long, making each concentration group enter a plateau. Therefore, we altered the amplification duration in the later experiment.

Figure 4. Agarose gel electrophoresis result of the amplification products.

Cas protein Cleavage Activity

In the initial design, considering CasΦ has a small size, strong specificity, and low dependence on PAM sequence, we hoped to use CasΦ to connect RT-RPA and DNA nanosponge system to enhance the specificity of our entire system. However, the cleavage activity of CasΦ is lower than that of Cas12a and Cas14a. Therefore, we hope to improve CasΦ's cutting rate and specificity and establish an advanced CRISPR-Cas system for detection.

We then performed directed evolution on CasΦ protein and screened for protein mutants with faster cutting rate and sgRNA with lower off-target rate through rational design. By points mutation of the key amino acids in the charge center in helix α7, we obtained mutant CasΦ nucleases. The positive controls Mut-1 (nCasΦ) and Mut-2 (vCasΦ) were designed by Doudna's team, and our team designed the mutants Mut-3 through Mut-6. The detailed mutation content is shown in Table 1. On the other hand, as reported by Gersbach et al, a secondary hairpin structure onto the spacer region of crRNA can increase the CRISPR-Cas system specificity for target cleavage. To verify whether this strategy is effective for CasΦ, we introduced different lengths of hairpin structures into the crRNA of CasΦ (Table 2.). Figure 5 shows that we successfully expressed and purified the CasΦ protein.

Table 1. Mutation sites of different mutants.

Table 2. The sequence of crRNAs with hairpin structures.

At the same time, we expressed and purified the Cas14a protein. Since Cas14a is combined with MBP to enhance its solubility, we expressed the MBP-Cas14a firstly. Then we used TEV to cut their combination, and further purified Cas14a through AKTA Explorer. Its expression can be verified by SDS-PAGE.Shown in Figure 5&6,the results indicated that MBP-Cas14a protein were expressed successfully.

Figure 5. SDS-PAGE results for expression of MBP-Cas14a.

Figure 6. SDS-PAGE results for purified of Cas14a after TEV digestion.

Figure 7. SDS-PAGE results for expression of CasΦ.

We first performed the FQ reporter cleavage assay with wild-type CasΦ, Cas12a and Cas14a to verify the dsDNA target triggered trans-cleavage activity. The results showed that Cas12a had the best activity, followed by Cas14a and CasΦ.

Figure 8. The time-course fluorescence intensity curves of FQ reporter cleavage by Cas12a, Cas14a, and CasΦ systems.

Then, we expressed and purified the Mut-1 (nCasΦ) and Mut-2 (vCasΦ) designed by Doudna's lab, and the results were in line with what was previously published. And the four mutant proteins (Mut-3 to Mut-6) were constructed by our team. The target-triggered trans-cleavage activity of wild-type and mutants was assessed by the fluorophore quencher (FQ) reporter assays. As shown in Figure 9, in Mut-3 to Mut-6, only Mut-4 (Neg-K) had shown a ssDNA target-triggered trans-cleavage activity. Compared to the wild-type, Mut-4 (Neg-K) exhibits a high ssDNA target-triggered trans-cleavage activity, which is comparable to that of Mut-1 designed by Doudna's group. These results further verify the hypothesis that the helix α7 might regulate the accessibility of the RuvC domain for the association of single-stranded DNA (ssDNA).

Figure 9. The time-course fluorescence intensity curves of FQ reporter cleavage by different Cas-crRNA in the presence of ssDNA targets.

We incorporated various lengths of hairpin structures into the crRNA of Cas to test the viability of the trans-cleavage activity of this method. As shown in Figure 10, the specificity of the Cas system for target recognition may be improved by crRNA with a 4-base pair stem structure (H4) according to our comparison of the specific signal from the complementary target and the non-specific signal from MT13.

Figure 10. The reaction rates of FQ reporter cleavage by Mut-4 (Neg-K) with hairpin structure crRNA.

Finally, we blended various numbers of ssDNA target sequences with mismatch sequences (MT13) to prepare artificial samples containing 50% to 0% DNA mutations, in order to test the performance of our optimized CRISPR-Cas system for detecting DNA mutations from a vast number of background sequences. Mut-4 (Neg-K) and crRNA H4 were the two options we chose to build the mutation detection system. As seen in Figure 11, the Mut-4 (Neg-K)/H4 system performed better at detecting mutations than the traditional wild-type/hairpin-free crRNA system (WT/CrRNA in Figure 11).

Figure 11. The reaction rates of FQ reporter cleavage with samples containing 50% to 0% DNA mutations.

After the modification of CasΦ system, we further compared the dsDNA target triggered trans-cleavage activity of Cas12a, Cas14a, and Neg-K/H4 systems. Unfortunately, despite the significant improvement in ssDNA target triggered trans-cleavage activity and specificity of the Neg-K/H4 system compared to the WT/CrRNA system, the dsDNA target triggered activity of the Neg-K/H4 system is still poor compared to the more mature Cas12a and Cas14a systems (Figure 12). This may be due to the long dsDNA targets are difficult to dissociate from the CasΦ proteins, resulting in weak activity. Therefore, in order to use the best Cas system in our system, we narrowed down the candidates to Cas12a and Cas14a.

Figure 12. The time-course fluorescence intensity curves of FQ reporter cleavage of Cas12a, Cas14a, CasΦ and Neg-K/H4 systems.

The dsDNA target triggered activity of Cas14a was futher investigated by a series of dsDNA targets with different concentrations. As shown in Figure 13, these results verified that Cas14a has sufficient sensitivity to identify targets at various concentrations. Plus, the quantification process of Cas14a finishes in 30min. As a result, Cas14a is capable to complete the precise quantification of our system.

Figure 13. The reaction rates of FQ reporter cleavage by Cas14a in the presence of DNA targets with different concentrations.

We further validated the ability of Cas12a and Cas14a to cleave DNA nanosponge in our system. We found that the ability of Cas12a and Cas14a to cleavage DNA nanosponge and release γ-amylase were similar, which was inconsistent with the relative activity of the two in the previous FQ reporter assay. This may be because the Cas system can only effectively release the enzyme in the shallow layer of DNA nanosponge, while it is hard to reach the deep layer. Therefore, the difference between the two is reduced in the cleavage of nucleic acid nanostructures. Since it has been reported in the literature that Cas14a has stronger specificity for the target sequence and can identify single-base mismatches in the target, which is more in line with the requirements of our system for specific target recognition. Therefore, we decided to choose Cas14a in the final project design.

MBD-SA

To construct the DNA Nanosponge, we creatively fused MBD and SA into MBD-SA fusion protein. MBD-SA will bind to the stem-loop structure formed by the CGAT CCGCGGC TCTCTC GCCGCGG ATCG fragment in the RCA product sequence. MBD-SA integrates the advantages of both proteins to enable the fusion protein to bind non-modified DNA multivalently.

The MBD-SA fusion protein was identified by Western Blot to prove that we had successfully constructed MBD-SA fusion protein with streptavidin activity. As shown in Figure 14, MBD-BCCP as the control group did not show a band when interacting with HRP-biotin, but MBD-SA did when interacting with HRP-biotin.

Figure 14. Western blot identification of streptavidin.

In order to further verify the binding ability of MBD-SA to unmethylated DNA, the binding activity of MBD-SA to methylated DNA and unmethylated DNA was measured by MST (microscale thermophoresis) assay. As shown in Figure 15-19, MBD-SA prepared by us has high DNA-binding activity, which indicates that it can better help the assembly of DNA nanostructure.

Figure 15. Affinity of MBD and U-DNA (Kd=13.341μM).

Figure 16. Affinity of MBD and M-DNA (Kd=1.4406μM).

Figure 17. Affinity of MBD-SA and U-DNA (Kd=6.9985μM).

Figure 18. Affinity of MBD-SA and M-DNA (Kd=0.19045μM).

In Table 3, the Kd value of MBD-SA to M-DNA (0.19045) was much lower than that of MBD to M-DNA (1.4406), which indicated that the affinity of MBD-SA was higher than that of MBD. On the other hand, MBD-SA also showed a high affinity for unmethylated DNA, enabling us to directly use RCA products for DNA nanosponge assembly. Table 3. Results of micro-thermophoresis.

In conclusion, the MBD-SA fusion protein has good biological activity and can be used to construct DNA nanosponge.

DNA Nanosponge

In order to construct a three-dimensional network formed by DNA to encapsulateγ-amylase which enables actived Cas14a to cleavage, we designed a DNA nanosponge. The DNA nanosponge is constructed by RCA (Rolling Circle Amplification) products and MBD-SA fusion protein. The MBD-SA fusion protein with 4 MBD structural domains can recognize and bind the repetitive CpG sites in the stem-loop structure in RCA-amplicons. In this way, the RCA-amplicons with stem-loop structure and MBD-SA fusion protein can self-assemble into a three-dimensional network called DNA nanosponge , which we later use to encapsulate γ-amylase (Figure 19).

Figure 19. The Self-assembly of DNA nanosponge.

We conducted agarose gel electrophoresis experiments after processing the amplification products of RCA. As shown in the Figure 20, RCA amplification has a sufficient yield.

Figure 20. Agarose gel electrophoresis of different products of the whole RCA process. Lane 1: Linear DNA; Lane 2 : Cyclized DNA ; Lane 3, Lane4, Lane5: RCA amplicons from different amount of cyclized DNA.

In order to find out the best conditions for preparing DNA nanosponge, we used cyclized DNA with different lengths as raw materials to synthesize different groups of DNA nanosponge. Then we temporarily used DeoxyribonucleaseI to play the role of DNA cleavage, and observed the release of amylase from different groups of DNA nanosponge after degradation. We found that DNA with 60nt of probe formed the best DNA nanosponge (Figure 21).

Figure 21. Glucose release from nanosponges formed by cyclized DNA with different lengths after degradation.

Considering the convenience of storage and useage, we made DNA nanosponge into lyophilized powder (Figure 22).

Figure 22. The images of prepared DNA nanosponge and its lyophilized powder.

We verified the three-dimensional network structure of the DNA nanosponge by Scanning Electron Microscopy (SEM) (Figure 23).

Figure 23. SEM observation of the structure of the DNA nanosponge.

To further verify the DNA nanosponge’s ability to encapsulate amylase and visualize it, we replaced γ-amylase with nanogold of similar size to the enzyme and used Transmission Electron Microscope(TEM) to characterize the DNA nanosponge encapsulating nanogold particles (Figure 24).

Figure 24. TEM observation of the DNA nanosponge encapsulating nanocrystals.

To find out how long it takes from DNA nanosponge degradation to glucose release, we measured the amount of glucose released at different time points. It turns out that the process only takes five minutes, which can help us achieve a rapid detection (Figure 25).

Figure 25. Concentration-Time Curve of Glucose.

Integration

In the end, we combined the parts we have mentioned above for testing, including RT-RPA, Cas14a system, and DNA nanosponge, and determined the glucose concentration after the release of amylase.

We created a target RNA concentration gradient from 10¹ to 10⁴, which covers the concentration that may appear in real-world samples, in order to verify that our system can effectively distinguish targets with varied concentrations and complete quantification in real samples. In the beginning, we examined the experimental setup of RT-RPA amplicons after 10 min, and 200 nM Cas14a. The results revealed that group 10⁴ differed from the other groups in a substantial way, demonstrating the capability of our integrated system to assess the concentration difference. The final glucose content of the other three groups, however, was nearly identical. We thought this was caused by either the high concentration of Cas14a, or the high efficiency of RT-RPA, which led the amplification to a plateau. Because it cannot precisely gauge our goal's concentration, this contradicts our objective.

As a result, we chose to narrow the RT-RPA duration in half to 5 minutes and lower the Cas14a concentration to 100 nM. The results indicate that our system can quantify the target (Figure 26). Because the concentration of the target we input is favorably connected with the final glucose signal when the settings are changed, we can assume that our system has accomplished our primary objective of creating a practical, precise, and sensitive detection platform.

Figure 26. Integration 1|Cas 200 nM RPA 10min.

Figure 27. Integration 2|Cas 100 nM RPA 5min.

Overall, we established a detection platform which can identify the precise concentration of LINC00857 in less than half an hour. We fulfilled our initial configuration.

Future Improvements

Due to the objective constraints, we cannot make our platform perfect in the short term, but our team will further optimize the experimental conditions, adjust the operating hardware, and simulate more realistic experimental samples to make a better platform in the future.