The goal of CTC-FAST is to detect lung cancer metastasis. We achieve the capture of lung cancer CTCs by specifically targeting the folate receptor on the CTC membrane using folic acid. However, the CCU-TAIWAN team's objective goes beyond detecting lung cancer metastasis; we aim to create a biological machine that can detect metastasis in all types of cancers. To accomplish this, we utilize DNA nanostructures, DNA tetrahedron, to anchor folic acid in our machine. This allows us to easily replace the ligand in the future when we need to detect metastasis in other types of cancers. To create such a biological machine, we overcome several issues and improve our design by multiple engineering cycles as outlined below.
Design:
We designed four separated single-strand DNAs (ssDNAs), which partially complement each other (see design for details). The annealing of these four ssDNA will produce a DNA tetrahedron through base pairing.
Build:
We mixed four ssDNAs strands in TM buffer (50 mM Tris HCl, 10 mM Magnesium Sulfate at pH 7.5), heated the sample to 95°C for 5 mins, and gradually cooled the sample to 4°C over 2.5 hours.
Test:
We analyzed the mixed sample by native PAGE electrophoresis after gradually cooling. The mixture of 2 ssDNAs shows a size of 80 bp, while the mixture of 3 ssDNAs shows a size of 120 bp. Importantly, the mixture of 4 ssDNAs PAGE showed the expected size of 175 bp. However, we also observed side products with larger sizes, which may represent DNA polyhedrons, forming through complementation between tetrahedrons.
Learn:
Due to the unexpected side product of DNA polyhedrons, we decided to change the design of 4 ssDNAs to one single tetrahedral ssDNA.
Design:
We decided to fuse the four separated ssDNA into one continuous ssDNA, namely tetrahedral ssDNA, and amplify it using in vitro rolling circle amplification (RCA). To cleave the long ssDNA into tetrahedral ssDNA, we incorporated cis-auto splicing sequences at the 3’ terminus of the tetrahedral ssDNA sequence.
Build:
The production of tetrahedral ssDNA with highly complementary sequences was rejected by the manufacturer. Therefore, we separated the tetrahedral ssDNA into two fragments (ssDNA-L and ssDNA-R) and added an extra stuffer to reduce the complementarity.
After getting the two parts of tetrahedral ssDNA, we performed the fusion PCR to fuse these two fragments into one tetrahedral ssDNA with a stuffer. To remove the stuffer, we cloned the fused fragment into pET-32a vector, then removed the stuffer by EcoRI. Finally, we excised the tetrahedral ssDNA from pET-32a by XhoI, and self-ligated the excised tetrahedral ssDNA into a circular DNA template for RCA.
Test:
The fusion PCR is successful, we could fuse the two fragment into one ssDNA with stuffer
▲ The electrogram shows the sequencing result of ssDNA-L+R sequence
Learn:
Unfortunately, we are suggested by experts that the high concentration of DNA tetrahedrons generated by RCA also form polyhedrons. Therefore, we decided to apply ssDNA binding protein and in vivo rolling circle replication (RCR) for ssDNA production.
Design:
To enhance DNA tetrahedron production, we applied RCR and expressed the ssDNA binding protein (SSBP) to protect the generated circular tetrahedral ssDNA. The SSBP is also applied to purify the binding ssDNA, as well as avoid unexpected interactions among ssDNAs.
Build:
To build the RCR system, we expressed the replication initiation proteins (RepA) from pC194 in E. coli. The tetrahedral ssDNA with cis-auto splicing sequence was flanked by the start motif (RCORI105) and stop motif (RCORI65) of RepA. The SSBP is co-expressed with Rep A to protect the tetrahedral ssDNA generated by RCR.
Test:
The fusion PCR is successful, we could fuse the two fragment into one ssDNA with stuffer.
▲ The electrogram shows the sequencing result of RcoRI-105 + ssDNA sequence + RcoRI-65
▲ The electrogram shows the Sanger sequencing result of ssDNA Binding protein - RepA
The RepA and SBP protein induction was confirmed by SDS-PAGE and Coomassie blue staining.
▲ RepA and ssDNA binding protein (SBP) expression
Design:
To specifically label lung cancer CTCs, we fused mGreenLantern (mGL) protein with a 12-mer peptide (namely C7) specifically recognizing the membranous FRα.To ensure that C7 recognizes CTCs, we must keep the flexibility of C7 in mGL-C7 fusion protein. Based on previous studies, a linker composed of multiple Alanine residues was chosen. To determine the proper length of the linker, we decided to model the protein structure of mGL-C7 fusion protein with different linker lengths and simulate the docking of these fusion proteins on FRα.
Build:
We downloaded the PDB file of the FRα receptor and generated the mGL-C7 fusion protein with different lengths (0-4) of Alanine linker by I-TASSER (Zhou et al.).
Test:
To determine the proper length of the linker, we use Frodock (Ramírez-Aportela et. al, 2016 ) to perform molecular docking simulation between mGL-C7 with different linker lengths (0-4 Alanine) and the FRα.
Frodock mechanism: The Frodock website combines 3D grid-based potentials with the efficiency of spherical harmonics (SH) approximations for docking model simulation.
Learn:
Through observing the simulated docking results of fusion proteins with different linker lengths and FRα, we found that mGL-4A-C7 did not affect the docking of C7 on the FRα receptor and maintained the optimal distance between mGL and FRα. We then designed the expressing biobrick of mGL-4A-C7 and examined whether mGL-4A-C7 protein is functional.
| Top view | side view | |
|---|---|---|
| mGL-C7 |
|
|
| mGL-1A-C7 |
|
|
| mGL-2A-C7 |
|
|
| mGL-3A-C7 |
|
|
| mGL-4A-C7 |
|
|
Test:
The coding sequence expressing mGL-4A-C7 was cloned into pET15b expressing vector by Gibson Assembly, and the expressed protein was analyzed by WB. After purifying the mGL-4A-C7 by Ni-sepharose beads and FPLC, we treated the SKOV3 cells with mGL-4A-C7 to examine whether the fusion protein was functional.
| control | 50 μg/mL | 100 μg/mL | 200 μg/mL | |
|---|---|---|---|---|
| nucleus |
|
|
|
|
| mGL-4A-C7 |
|
|
|
|
| Merge |
|
|
|
|
X Zhou, W Zheng, Y Li, R Pearce, C Zhang, EW Bell, G Zhang, Y Zhang. I-TASSER-MTD: a deep-learning-based platform for multi-domain protein structure and function prediction. Nature Protocols, 17: 2326-2353 (2022)
E. Ramírez-Aportela, J.R. López-Blanco, and P. Chacón (2016). FRODOCK 2.0: Fast Protein-Protein docking server. Bioinformatics, 32(15), 2386-2388