CRISPR/Cas12a cleavage triggered nanoflower for fluorescence-free and target amplification-free biosensing of ctDNA in the terahertz frequencies

The detection of tumor biomarkers in liquid biopsies requires high sensitivity and low-cost biosensing strategies. However, few traditional techniques can satisfy the requirements of target amplification-free and fluorescence-free at the same time. In this study, we have proposed a novel strategy for ctDNA detection with the combination of terahertz spectroscopy and the CRISPR/Cas12 system. The CRISPR/Cas12a system is activated by the target ctDNA, resulting in a series of reactions leading to the formation of an Au-Fe complex. This complex is easily extracted with magnets and when dropped onto the terahertz metamaterial sensor, it can enhance the frequency shift, providing sensitive and selective sensing of the target ctDNA. Results show that the proposed terahertz biosensor exhibits a relatively low detection limit of 0.8 fM and a good selectivity over interference species. This detection limit is improved by three orders of magnitude compared with traditional biosensing methods using terahertz waves. Furthermore, a ctDNA concentration of 100 fM has been successfully detected in bovine serum (corresponding to 50 fM in the final reaction system) without amplification.


AuNP-DNA1 Preparation
AuNPs(~17 nm) were synthesized following the literature (Yazdani et al. 2021), 0.5 mL of HAuCl 4 (25 mmol/L) was quickly dropped into 48 mL of boiling Milli-Q water in a 100 mL beaker on a hot plate.The boiling solution was vigorously stirred for 10 minutes till the color turned to gold.Then 0.1 g of trisodium citrate dihydrate was dissolved in 10 mL of Milli-Q water and was added dropwise to the boiling water.Continuously stir the boiling mixture for 30 min until its color turns wine crimson.After cooling the mixture to room temperature, the synthesized AuNPs solution (~17 nm) was stored in Milli-Q water at 4 o C refrigerator.
We can easily produce AuNP-DNA1 by quickly chilling the mixture due to the adsorption between AuNP and repeating A base sequence.Briefly, 5 µL of probe DNA1 (500 mmol/L) and 100 µL of AuNPs (~17 nm) were mixed and stored at -20 o C for 4 hours.When the wine-red fluid turned colorless, it meant that the DNA and gold nanoparticles had completely merged.Additionally, the UV-vis spectra of these samples were examined to determine whether DNA was bonded to the surface of gold nanoparticles (as seen in Fig. S2).

DNA band electrophoresis analysis
We have performed agarose gel electrophoresis and polyacrylamide gel electrophoresis (PAGE) for DNA band analysis.The prepared samples were mixed with DNA loading buffer in 5:1, and then added in the lanes of 12% polypropylene gel or 2% agarose gel depending on the size of the DNA sequence.After the gel was electrophoresed for 40 min at 30 mA and dyed in a GelRed for 10 min, the images were obtained under UV light.

SEM characterization of Fe complex
SEM characterization has been performed to check the morphology of flowerlike Au-Fe complexes in Fig. S5.Scanning electron microscopy (SEM) (FEI Scios DualBeam FIB/SEM) was performed at Shenzhen University.We have confirmed that ssDNA is cleavable.Subsequently, we confirmed that the ssDNA linker sequence can be protected and, hence, be effectively used for Au-DNA1 capture in the next steps.

2.Results and discussion
In our study, there are two stem-loop structures in ssDNA, in which the linker region is located in the stem-loop structure at 5'terminal (Fig. S1    To verify whether DNA and Au nanoparticles were successfully combined, UV-vis spectra were taken.Au-DNA1 and Au-DNA2 exhibit a distinct blue shift in Fig. S2, suggesting that the Au-conjugated DNA strands were successfully joined.Furthermore, the hue of the solutions containing Au-DNA1

Au-DNA1 Characterization
and Au-DNA2 turns colorless due to changes in the refractive index, which are influenced by the state of gold aggregation.

Influence of reaction time on Cas12a cleavage efficiency
Research showed that overdosage of Cas12a and reaction time will result in a fully digestion of ssDNA into pieces (less than 10 bp).In our study, 500 ng of Cas12 and more than 1 h of reaction time, both ssDNA and dsDNA were cut into small pieces, as any DNA sequences cannot be observed in lane (Fig. S3 (a)), which is invalid for forming flower-like Au-Fe complexs.Next, we carried out cleavage efficiency assay to optimize the working condition.We optimized working conditions with different dosages of Cas12a (0, 100, 200, 300, and 400 ng) and reactive times (0, 15, 60 min), their PAGE analysis results are shown in Fig. S3, and the transmission spectra are shown in Fig. S4.(blueshift).Note that the frequency shift with 400 ng of Cas12a and that with 60 min reaction time is both 17.7 GHz, which is the smallest signal in the experiments.This indicates that over-dosage and long reaction time can both result in fully cleaved DNA fragments which are too short for linking.PAGE experiments have been carried out to support our hypothesis (Fig. S3 (b)), diffusion bands still can be observed when the Cas12a dosage is setting as 100 ng or less.Therefore, with a dosage of 100 ng and a reaction of 15 min, longsize DNA fragments still exist to function as a linker for subsequent experiments.

Feasibility analysis of the experiment
As shown in Fig. S5 (a), the I-shaped MM was successfully fabricated following the procedure in the literature (Yan et al. 2019).Looking at the sample's morphology in Fig. S5 (c), one might speculate that the frequency shift is caused by the uneven thickness of the sample on the surface of the sensor.Therefore, we have considered the thickness effect in the simulation.As can be seen in Fig. S5 (e), a 7 μm difference in thickness can induce a maximum frequency shift of 50 GHz, and this shift saturates for increasing thicknesses.Considering the size of the nanoparticles (200 nm diameter), the frequency shift (64.9 GHz) is indeed caused by the addition of the DNA-linked Au-Fe complexes.These results further proved the feasibility of our proposed THz biosensing strategy.

Simulations 3.1 Finite element method (FEM) simulation
In this study, the sensing capabilities of asymmetric double-split ring metamaterials were simulated using finite element analysis with Comsol Multiphysics.As shown in Fig. S6, the asymmetric double-split ring metamaterial consists of Au deposited on a Si substrate with a thin oxide layer.Fig. S6(b) depicts the precise configuration parameters, with the corresponding structural parameters of each unit being L=44 µm, d=6 µm, g=4 µm, s=28 µm, and p=50 µm.Si substrate's dielectric constant is adjusted at 11.9 and thickness H is set to 20 m.The dielectric constant of the oxide layer is 3.75 and the thickness is 1 µm.Analyte thickness H is set to 10 µm.200 nm Au was selected for metal deposition, and the metal conductivity is set to 4.09×107 S/m.The boundary conditions are periodic in the x-axis and y-axis, and the plane waves polarised along the y-axis are set to be vertically incident on the unit structure to obtain the asymmetric double-split ring metamaterial resonance frequencies of 0.85 THz and 1.42 THz.We simulated the effect of sample thickness on the metamaterial.The refractive index of the analyte is set to 1.3 to simulate the refractive index of the biological sample.A red shift can be observed when the sample thickness is increased from 1 μm to 11 μm.With the linear increase of analyte thickness, the metamaterial frequency shift saturates at an analyte thickness of 7 µm, and the saturation frequency shift for thickness is 50 GHz.As shown in Fig. S9, we construct a logarithmic function  =  *   ( -) to fit the resonance frequency to the analyte thickness.In the above formula, y is the resonance frequency, and x is the analyte thickness.After fitting, a, b, c and d are 0.003 、1.417、-0.049、1.391,respectively.The simulation results fit well with the nonlinear function (Fig. S7).Finally, we performed a simulation of the refractive index sensitivity of the metamaterial.Since the frequency shift of the metamaterial reaches saturation when the analyte thickness is 7 µm, the thickness of the analyte is set to 8 µm to reach the upper limit of the resonance shift of the metamaterial to the thickness of the analyte.As shown in Fig. S8 (a), when the refractive index of the analyte increases from 1 to 1.5, the resonance frequency of the metamaterial is red-shifted from 0.85 THz to 0.8 THz and 1.42 to 1.335 THz, respectively.The theoretical sensitivities of the metamaterials can be obtained from the sensitivity formula  = ∆ ⁄ ∆ to be 100 GHz/RIU, 170 GHz/ RIU, respectively (∆f is the frequency shift and ∆n is the refractive index change).As shown in Fig. S8 (b).A linear fit of the metamaterial resonant frequency to the refractive index of the analyte was performed, with a linearity of 99.6%.The resonant frequency can be approximated by the equivalent inductance and capacitance in the form of  0 = 1 2  , where C is the equivalent gap capacitance, which is closely related to the refractive index of the gap dielectric material.

Simulation of the sensitivity of the metamaterial
When the analyte with higher refractive index covers the surface of the metamaterial, the equivalent inductance L remains unchanged and the equivalent capacitance C increases.According to the resonance frequency formula, it can be seen that the metamaterial resonance frequency will be redshifted, and with the increase of the analyte refractive index, the metamaterial frequency shift increases.

4.Experiment setup
All experiment were performed in a commercial Terahertz time-domain spectroscopy (TDS) system (TeraPulse 4000, TeraView LTD., Cambridge, UK), which were consisted of a femtosecond laser operating at 780 nm, a pair of low-temperature GaAs photoconductive antennas as terahertz transmitters and detectors, and an optical delay line (Fig. S9).
To avoid the absorption of terahertz by water vapor, the measuring chamber is purged with dry nitrogen during the measurement process to maintain a humidity level of 4 %, and the measurement temperature is maintained at 21 ± 0.1 °C.We also examined the terahertz spectra of various spots on the surface of the metamaterial, which revealed a constant frequency at 1.41 THz.The biological samples were then added to the metamaterial surface and dried in oven at 50 ℃ for 15 minutes.Each sample was performed three times.The error bars of the different ctDNA concentrations were calculated using information from two independent measurements.The transmission spectrums were calculated as following steps.Firstly, multiple reflections of the terahertz pulse at the sample-air interface echoes in the resulting time-domain spectrum, which is filtered by the rectangle function.The frequency waveform is obtained by performing a fast Fourier transform.Finally, the transmission spectra of different samples were obtained and analyzed by using () =   ()/  ().
(a)).The green and blue lines indicate the complementary sequence of ssDNA with Fe-CP1 and Au-DNA1, respectively.The working principle of Cas12a for singlestranded DNA cleavage is illustrated in Fig.S1 (b).The Cas12a protein contains some nuclease domains (REC, RuvC, Nuc) for DNA cleavage.When Cas12a is activated, a positive "pocket" is formed to capture ssDNA and cleavage them into different-sized fragments(Xu et al. 2023;Li et al. 2018).

Fig. S3 .
Fig. S3.DNA band electrophoresis analysis.a) is the agarose gel analysis with overdosage of Cas12a (500 ng) and long reaction time (more than 1 hour).5 µM of crRNA dsDNA and ssDNA; b) is PAGE analysis with 100ng of Cas12a, 5 µM of crRNA and ssDNA, 5 nM of dsDNA.PAGE analysis results were performed to check if cleavage DNA fragments exist in our optimized reaction system.As expected, ssDNA sequences can be observed within a reaction time of 15 minutes, and the size of diffuse bands increases with the increasing dosage of Cas12a, indicating that ssDNA was

Fig. S4 .
Fig. S4.Transmission spectra of optimized reaction system.a) is the transmission spectra of different reaction times (0, 5,15,60min) and their frequency shift values; b) is the transmission spectra of target ctDNA reactive with different concentrations of Cas12a and corresponding frequency shift.The target concentration of ctDNA is 500 pM.

Fig
Fig. S4 (a)-(b) is the reaction time effluence on frequency shift (Cas12a were fix to 100 ng).Results show that the reactive time of 60 min causing a lower frequency shift value (17.7GHz) comparing with the time of 5 and 15min (29.5 GHz).Then we fix the reaction time to 15 min and check the Cas12a dosage effluence on frequency shift.Fig. S4 (c)-(d) clearly shown that the frequency shift increases with increasing concentration of Cas12a from 100-300 ng (redshift can be observed), but decreases when the dosage is up to 400 ng Fig. S5 (b)  and (c) show the micrographs of the I-shaped MM loaded with Fe-SA-CP1 (ctDNA, 0 pM) and flower-like Au-Fe complexes (ctDNA, 250 pM), respectively.The Fe-SA-CP1 loading in Fig.S5(b) presents a relatively flat surface, while flowerlike Au-Fe complexes in Fig.S5 (c) present an irregular pore morphological structure.It may be attributed to the fact that the structure of flower-like Au-Fe complexes prevent the further aggregation of FeNPs compared with that of the Fe-SA-CP1.

Fig
Fig. S5.SEM images of (a) a bare MM, (b) sample loading on I-shaped MM with b) and without c) ctDNA; d) experimental transmission spectra.e) The simulation results of sample thickness on MM.

Fig
Fig. S6.(a) The geometric configuration of MMs-based biosensor, the THz waves normally incident through bare MMs.(b) the top view of the unit cell consisting of MMs.

Fig
Fig. S8.(a) The dependence of transmission on the analyte refractive index increasing from 1.1 to 1.5 under the thickness of 8 µm; (b) The resonance frequency under the different refractive index of analyte extracted from (a).

Fig
Fig. S9.(a) A schematic illustration of the THz-TDS system; (b) Photographs of the measurement chamber, the arrangement of the sample holder inside the chamber, the biosensor fixed on the sample holder, and the microscopic photo of the fabricated metamaterial; (c) Illustration of the sample holder.

Table S1 . DNA sequences
* We design a DNA sequence that is complementary to double-stranded DNA (underline), where the CrRNA structure is provide by GenScript company.