A dual biomarker detection platform for quantitating circulating tumor DNA (ctDNA).

Circulating tumor DNA (ctDNA), which includes DNA mutations, epigenetic alterations and other forms of tumor-specific abnormalities, is a promising "real-time" biomarker for noninvasive cancer assessment. Tumor DNA is of great value in the process of cancer treatment, including diagnostic and prognostic information before, during treatment and at progression. Here we introduce a peptide nucleic acids probe-gold nanoparticles (PNA-AuNPs) and lead phosphate apoferritin (LPA)-based dual biomarker detection platform, which could be used in a DNA biosensor to quantify ctDNA by detection of tumor-specific mutations and methylation of PIK3CA gene. On the one hand, PNA probe and anti-5-Methylcytosine monoclonal antibody (anti-5-mC) were used to recognize the different parts of ctDNA, forming a sandwich-structure on a screen-printed electrode (SPE) surface. On the other hand, AuNPs and LPA were introduced to construct the biosensor for double signal amplification. Square-wave voltammetry (SWV) was used to monitor the electrochemical signal of lead ions released from apoferritin. The proposed DNA biosensor yielded a linear current response to ctDNA concentrations over a broad range of 50-10000 fM with a detection limit of 10 fM. It also successfully detected ctDNA collected from cancer patient serum. Therefore, we anticipate this new platform opens up an approach to detect and monitor diverse malignancies, facilitating personalized cancer therapy.

Introduction

Cancer is a leading cause of death globally which needs appropriate diagnosis methods. Currently, most therapy decisions are solely based on tissue biopsies, as an invasive test, which can be a burden to the patients. Additionally, most tumors are composed of multiple subclones which are constantly evolving 1. However, tissue biopsies provide just a static snapshot of the tumor at the time of biopsy and do not necessarily represent the entire tumor genome 2. Therefore, it is of great value to develop noninvasive detecting methods that could monitor the real-time dynamic of cancer. “Liquid biopsy” may be an ideal one for patients with cancer, as it could provide valuable information of current tumor dynamics, facilitate monitoring of systemic anticancer therapy, and help in identifying appropriate therapeutic targets 3-5. Early research suggested that circulating tumor cell capture and analysis had potential as a noninvasive marker of tumor diagnosis 6, however clinical development of these technologies have been slow. Several studies have now suggested that highly sensitive assays can detect the concentration of circulating tumor DNA (ctDNA), which is significantly higher in cancer patients compared with healthy individuals, potentially reflecting the biology of a patient's cancer 7-10.

Recent advances in noninvasive diagnostics highlight the clinical potential of ctDNA, but also the challenges involved in analysis of it. Tumor DNA is fragmented to an average length of 50 to 250 base pairs (bp) and is present in only a few thousand amplifiable copies per milliliter of blood, of which only a fraction may be diagnostically relevant 11, 12. Various technology platforms for ctDNA analysis have been developed using different strategies such as dPCR 13, bead emulsion amplification magnetic (BEAMing) 14, tagged-amplicon deep sequencing (TAM-Seq) 15 and CAncer Personalized Profiling by deep Sequencing (CAPP-Seq) 16. Although these technologies have led to significant contributions to ctDNA detection, further application is still hindered by expensiveness, low sensitivity, tedious steps, harmful reagents and sophisticated instruments. Thus, the development of a novel and sensitive platform that enhances ctDNA analysis efficiency would dramatically increase the use of ctDNA in diagnostics, prognostics and targeted therapy. Tumor DNA can be detected by tracking tumor-specific mutations or gene methylation. Mutation at two hot spots of E542K and E545K and methylation of ctDNA of PIK3CA are well-known mutations in many types of cancers including stomach, breast, lung, liver, colon and brain 17-19. Based on above, we developed a new platform for ctDNA analysis on the basis of dual biomarker: genetic mutation and methylation alteration, which could increase the sensitivity and specificity of ctDNA based diagnostics.

In view of small size, low concentration and free floating of ctDNA in the peripheral blood, capturing and enriching ctDNA targets are essential steps to completely eliminate unspecific binding. Peptide nucleic acid (PNA), the artificially synthesized DNA analog with an uncharged peptide backbone 20, has a great potential to be used as a probe for DNA biosensor. The hybrid formed between PNA and its complementary DNA is more stable than the corresponding DNA/DNA hybrid due to the absence of electrostatic repulsion between negative charges along the phosphate backbone existing in the DNA structure 21. This characteristic makes PNA probe more acceptable to use in mutation detection studies because PNA probe makes a large difference in ΔTm (15 oC) between perfect match and single mismatch, including even insertion and deletion. Therefore, PNA molecules could be an ideal probe to target ctDNA accurately, which eliminate nonspecific interference on the sensor platform under thermally controlled conditions.

Current strategies for quantification of ctDNA mutation are localized surface plasmon resonance 22, surface enhanced Raman scattering 23, 24 and digital PCR 13. Electrochemical immunoassays and biosensors have evolved dramatically over the past decades and have been successfully used for in-field biomarker detection and clinical diagnosis 25 due to their rapidity, portability, simplicity and low-cost. In order to meet the increasing demand for ultrasensitive detection, the sensitivities of conventional biosensor need to be further enhanced by using novel signal amplification strategy. One of the most popular amplification is the introduction of nanoparticles, such as gold nanoparticles, carbon nanotubes, graphene oxide, and silica nanoparticles 26-28. Besides, protein cages, such as apoferritin, also have been reported as an immuno-label to enhance detection sensitivity 29. Apoferritin, an iron storage protein with a diameter of 12.5 nm, could act as a nanoreactor to synthesize voltammetrically distinguishable metal phosphates inside 30 for signal amplification of biosensor.

Herein, we demonstrate initial work toward developing an ultrasensitive, highly selective, and disposable ctDNA biosensor based on a dual biomarker platform. In this platform, ctDNA was effectively enriched since PNA and 5-Methylcytosine monoclonal antibody (anti-5-mC) were used to recognize the different biomarker (tumor-specific mutation and epigenetic methylation, respectively) of ctDNA, thereby forming a sandwich-like DNA biosensor, which could analysis ctDNA both in complex biological matrices and cancer patient serum. For ctDNA biosensor fabrication, as shown in Scheme 1, AuNPs were served as nanocarrier to immobilize more PNA probes to form PNA-AuNPs conjugate, which could enrich more ctDNA, thus enhancing the electrochemical detection signal. PNA probes were designed with matching for E542K and E545K perfectly, with one mismatch for normal circulating DNA (ncDNA). Moreover, lead phosphate apoferritin loaded with anti-5-mC (LPA-anti-5-mC) could improve the selectivity of the target ctDNA and amplify the signal in further. In presence of ctDNA, PNA-AuNPs/ctDNA/LPA-anti-5-mC sandwich-format system could be constructed on a disposable screen-printed electrode (SPE) surface (Scheme 1B). Highly specific and sensitive detection of ctDNA was achieved by the electrochemical signal of lead ions released from apoferritin. Conversely, as ncDNA neither matched perfectly with the PNA-AuNPs probe, nor reacted with LPA-anti-5-mC, no sandwich-structure could be formed (Scheme 1C). Combined with more accurate sequencing technologies and specific immune response, as well as double amplification strategies, the proposed DNA biosensor exhibited excellent sensitivity and selectivity, indicating great potential for target analysis and personalized therapy at point of care.

Scheme 1

(A) Schematic diagrams of preparation of PNA-AuNPs conjugates and LPA-anti- 5-mC bioconjugate. (B) and (C) Schematic illustration of DNA biosensor and ctDNA detection.

Experimental

Reagents

The oligonucleotides used in this study had the following sequences: PNA probe for hot-spot E542K: 5′-HS-AGTGATTTTAGAGAG; PNA probe for E542K: 5′-HS-CCTGCTTAGTGATTT; ctDNA: 5'-CAmCGAGATCCTCTCTCTGAAATCACTGAGCAGGAGAAAGATTTTCTATGGAGTCACAGACACTATTGTG and 3'-GTGCTCTAGGAGAGAGACTTTAGTGACTmCGTCCTCTTTCTAAAAGATACCTCAGTGTCTGTGATAACAC were purchased from TSINGKE Biotech. Co., Ltd. (Wuhan, China). Apoferritin, lead nitrate, human serum (H4522), avidin, biotin N-hydroxysulfosuccinimide (biotin-NHS), bovine serum albumin (BSA), phosphate buffered saline (PBS), acetate buffer and Tris-HCl buffer were the products of Sigma-Aldrich. Monoclonal anti-5-Methylcytosine antibody (anti-5-mC) was obtained from Epigentek Group Inc. (Farmingdale, NY). AuNPs were purchased from XFNANO (Nanjing, China). Cancer patient serum was provided by Zhongnan Hospital of Wuhan University (Wuhan, China). Other reagents were commercially available and were of analytical grade. All solutions were prepared using ultrapure water (≥18 MΩ, Millipore).

Apparatus

SWV measurements were carried out on a CHI 660C workstation (CH Instruments Inc., Shanghai, China). SPE consisting of a carbon working electrode, a carbon auxiliary electrode, and an Ag/AgCl reference electrode (Scheme 1) were purchased from DropSens (Asturias, Spain) for electrochemical measurement. A sensor connector was used to connect the SPE to the CHI 660C workstation. TEM micrographs obtained on a Tecnai G2 F20 S-TWIN (FEI, USA). The UV-vis absorption spectra were recorded on a UV-2550 spectrophotometer (Shimadzu, Japan). The EQCM instrument is composed of quartz crystal microbalance (QCM200 Stanford research systems), which was used to measure in situ changes in mass (measured by the frequency shift Δf).

Preparation of PNA-AuNPs conjugates

PNA-AuNPs conjugates were prepared according to the previous report 31, 32 with slight modification. Briefly, AuNPs were first modified with alkanethiol oligonucleotides by adding 10 μL PNA to 200μL AuNPs solution. After being incubated for 24 h at room temperature, the solution was adjusted to the pH value and ionic strength of the PBS (10 mM, pH 7.0) and allowed to stand for 6 h. Then 1 M NaCl was added to the solution to bring the total NaCl concentration of the solution to 0.05 M. The process was repeated 14 h later by adjusting the NaCl concentration to 0.1 M. The solution was stabilized under these conditions for additional 24 h. Excess oligonucleotides were then removed by centrifugation for 30 min at 12,000 rpm. Following removal of the supernatant, the red PNA-AuNPs conjugates were washed three times with 0.1 M NaCl, PBS (pH 7.0) by centrifugation. Finally, the red PNA-AuNPs conjugates were dispersed in 1 mL ddH2O and stored at 4℃. AuNPs and PNA-AuNPs were characterized by TEM and UV-vis.

Preparation of LPA nanoparticles

The preparation of LPA nanoparticles was performed as follow 29: 40 μL apoferritin (50 mg mL-1) was first diluted with 0.1 M Tris-HCl to a volume of 2.0 mL. Secondly, 250 μL lead nitrate (6 mM) was slowly introduced into the above apoferritin solution, and the mixture was continuously stirred for 1 h to allow the lead ions to diffuse into the cavity of apoferritin. Then 250 μL PBS (pH 7.0) was added drop-by-drop and the mixture was stirred for another hour to form a phosphate metal core inside the apoferritin. The mixture was centrifuged for 5 min at 5,000 rpm. The supernatant was washed three times with Tris-HCl using an Amicon filter (MWCO = 25,000). Finally, the prepared LPA nanoparticles were reconstituted into a 1.0 mL solution and stored at 4℃.

Preparation of LPA-anti-5-mC bioconjugate

The LPA-anti-5-mC was prepared using an avidin as a bridge to link biotin modified anti-5-mC and LPA 33. Herein, avidin was used to connect biotin-LPA and biotin-anti-5-mC since avidin possesses 4 active sites which will react with one biotin residue and operate as acceptor for another biotin-labeled protein. Briefly, 800 μL of the above prepared LPA nanoparticles were mixed with 200 μL biotin-NHS (0.1 mg mL-1) and stirred for 3 h at room temperature. The mixture was dialyzed against Tris-HCl overnight to remove excess biotin-NHS. The resulting biotin-LPA was reconstituted into a 1.0 mL solution containing 0.5% BSA and stored at 4 oC. The same process was performed to prepare the biotin-anti-5-mC. Subsequently, 1 mL avidin (0.06 mg mL-1) was mixed with biotin-LPA and reacted for 2 h to form avidin-biotin-LPA conjugate. Then an amount of biotin-anti-5-mC was added at an equal molar equivalent to the avidin-biotin-LPA solution and the reaction proceeded for another 2 h. Excess agents were removed using an Amico filter (MWCO = 300,000). The resulting anti-5-mC-biotin-avidin-biotin-LPA (i.e. LPA-anti-5-mC) was dispersed in Tris-HCl buffer containing 1% BSA and kept at 4 oC for the following experiment.

Construction of DNA biosensor

Before the experiments, the SPE was washed with distilled water and dried with nitrogen. PNA-AuNPs solution containing 0.05% (w/v) chitosan was dropped onto the working electrode surface of the SPE. After washing with distilled water and dried, 100 μL ctDNA or ncDNA at the desired concentration was then introduced onto the PNA-AuNPs modified SPE and retained in an incubator at 62 °C for 30 min 10. After washing by PBS at 62 °C, ncDNA was washed away due to different melting temperature. Then SPE was incubated with LPA-anti-5-mC bioconjugate for another 30 min, followed by rinsing with Tris-HCl buffer thoroughly to remove the physical adsorption of LPA-anti-5-mC on the SPE. An aliquot of HCl was dropped on the working electrode of SPE to dissolve the captured LPA. The detection solution containing 0.5 μg mL-1 Bi in acetate buffer (0.2 M pH 4.5) was added on the above PNA-AuNPs/LPA-anti-5-mC/ctDNA SPE for 5 min to release lead ions from the apoferritin. This is followed by SWV detection of the released lead ions at an in situ plated bismuth film electrode.

Serum sampling and detection

Serum samples were collected from ten cancer patients and four of them randomly chosen for the experiments. For detecting human concentration of ctDNA, 100 μL of serum was added onto the PNA-AuNPs modified SPE and incubated for 30 min. Then SPE was incubated with LPA-anti-5-mC bioconjugate for another 30 min, followed by dropping with HCl. The results were quantified by SWV detection with electrochemical workstation. Serum sampling was done according to ethics principles set out by the biosafety committee at Zhongnan Hospital of Wuhan University and consent forms were signed by all donors.

Results and Discussion

Characterization of the AuNPs, PNA-AuNPs conjugates and LPA nanoparticles

AuNPs and PNA-AuNPs conjugates were characterized by UV-vis spectrophotometer. Fig. 1A shows that the plasmonic peak of AuNPs surface modified with PNA probe has a red shift about 4 nm. This plasmonic damping was consistent with the previous reports 22. Moreover, transmission electron microscopy (TEM) was used to observe the AuNPs and PNA-AuNPs (inset of Fig. 1A). The bare AuNPs diameter was 40 nm ± 2 nm, but dispersion of PNA-AuNPs was significantly better than the pure AuNPs, which is because strong intermolecular repulsions among the negatively charged phosphodiester backbones of PNA probe increase the stability of PNA-AuNPs 34. These results demonstrated a successful PNA probe modification on the AuNPs surface.

Figure 1

(A) UV-absorption spectrum of AuNPs (black curve) and PNA-AuNPs (red curve); TEM images of AuNPs (inset a) and PNA-AuNPs (inset b). (B) TEM image of lead phosphate-apoferritin nanoparticles.

As shown in Fig. 1B, the individual particles are discrete, spherical and uniform with a diameter of about 12 nm, and the dense lead phosphate cores appear black and are surrounded by an amorphous protein shell. It seems to confirm that lead ions were encapsulated into the core of the apoferritin. We calculated the loading amount of metal ions per apoferritin by comparing the concentrations of apoferritin and lead ions in the nanoparticle solution. The concentrations of apoferritin and metal ions were determined by the BCA method and ICP, respectively. The numbers of lead ions are 1300 per apoferritin with a standard deviation of ±90 (n=6).

Evaluation of immunoaffinities

Nonspecific adsorption is one of the most important factors in development of nanoparticle-based DNA biosensor. Electrochemical quartz crystal microbalance (EQCM) and electrochemical methods were used to monitor the interaction among PNA, ctDNA and anti-5-mC antibodies. As shown in Fig. 2A, the frequency changes (Δf) included the changing trends and well reproduced EQCM response. Before adding samples, the EQCM was initially equilibrated with Tris-HCl for 30 min, until stable baseline for Δf was achieved. After attaching ctDNA, as expected, it resulted great decrease in Δf response, indicating successful capture of ctDNA onto the PNA-AuNPs modified SPE surface (curve a). Further decrease of Δf resulted from the introduction of LPA-anti-5-mC which was recognized by ctDNA based on immnointeraction. We also observed that two recognitions were completed both within 30 min. Besides, we used ncDNA as control for comparison (curve b). It showed that the resulting response of Δf is very weak, which further proved strong interaction between PNA-AuNPs and ctDNA.

Figure 2

(A) EQCM measurements for recognizing (a) ctDNA and (b) ncDNA, respectively. (B) SWV responses of the DNA biosensor for (a) PBS, (b) 1% BSA, (c) 1000 fM ncDNA, (d) 10 fM ctDNA and (e) 1000 fM ctDNA.

Furthermore, Square-wave voltammetry (SWV) measurements were used to confirm the above immunoaffinities. Here, 20 fM ctDNA, 1000 fM ncDNA, PBS and 1% BSA were served as controls to be simultaneously examined as the challenging proteins. As shown in Fig. 2B, there are no detectable signals in PBS (curve a) and very small currents in BSA (curve b), 1000 fM ncDNA(curve c) and 10 fM ctDNA (curve d). However, a well-defined voltammetric peak from 1000 fM ctDNA loading on SPE is distinguished (curve e), indicating the strong affinity among PNA, ctDNA and anti-5-mC antibodies to form sandwich-like PNA-AuNPs/ctDNA/LPA-anti-5- mC complex. These observations suggest that the effects of non-specific adsorption could be neglected.

Optimization of detection conditions for ctDNA

Since the PNA-AuNPs conjugates were used here as capturing platform, the amount of PNA on the surface of AuNPs influenced the amount of captured ctDNA, and this accounted for the amount of LPA-anti-5-mC. One can see from Fig. 3A that the voltammetric currents increased upon raising the concentrations of the PNA on the surface of AuNPs and achieved the highest signal at 6 mM PNA. Thus, 6 mM PNA was used in the experiments.

Figure 3

Effects of parameters on immunosensing responses. (A) Concentration of PNA, (B) incubation time including PNA capture (black) and LPA-anti-5-mC recognition (red).

The current responses also depended on the incubation time of both PNA capturing and LPA-anti-5-mC recognition of ctDNA. As shown in Fig. 3B that both the responses increased with increasing incubation time and tended to steady values after 30 min (black and red curve), which was used for the construction of DNA sensor. These results were consistent with the EQCM results above.

Electrochemical measurements on the DNA biosensor

For the detection of ctDNA, high sensitivity and selectivity play important roles in improving the treatment. In view of this, the proposed DNA biosensor was challenged with different concentrations of ctDNA, as shown in Fig. 4. The SWV currents (from lead ions) increased with the increasing concentration of ctDNA from 20 fM to 10000 fM. The resulting calibration plot of current versus known concentration of ctDNA (Fig. 4 inset) is linear over 50 fM to 10000 fM range, with the detection limit of 10 fM, which is comparable to the previous reported value 22, 23. This method shows more advantages in its simplicity, rapidity and low cost. Since the average concentration of ctDNA in cancer patient plasma is more than 200 fM 3, the proposed device is sensitive enough to detect low-dose ctDNA. A series of six measurements of 1000 fM ctDNA yielded reproducible electrochemical responses with an RSD of 5.3% (data not shown), indicating acceptable reproducibility. In addition, the DNA biosensor could be stored at 4 ◦C. In this way, over 80% of the initial response remained after one month, indicating acceptable stability.

Figure 4

Typical electrochemical response of the DNA biosensor with the increasing concentration of ctDNA (20, 50, 100, 200, 500, 1000, 2000, 5000, 10000 fM from a to i). The inset shows the resulting calibration plot of ctDNA.

To explore the use of the biosensor for monitoring biological matrices, human plasma samples were spiked with ctDNA to final concentrations of 50, 200, 1000, 5000 and 10000 fM. Analysis of 100 μL of sample was performed using the DNA biosensor. The results were summarized in Table 1, which showed the recoveries are calculated to be in the range of 99.0-104.2%. Furthermore, the proposed DNA biosensor was used to measure ctDNA level in serum collected from four cancer patients before and after therapy. Figure 5 shows the ctDNA levels of four volunteers as measured by our device. For the calculation of the analyte concentrations, the current normalized responses of the DNA biosensor were correlated with the calibration curves depicted in Figure 4. As shown in Figure 5 and Table 2, when the treatment was beneficial, as determined by decrease in tumor size or reduction of pain, a decrease concentration of ctDNA also was observed. Conversely, when the treatment was unsuccessful, ctDNA levels tended to remain unchanged or even increased. Successfully detecting the ctDNA from cancer patient serum displays promise of our DNA biosensor for various biomonitoring applications in the near future.

Table 1

Comparison of DNA biosensor values with known concentrations of ctDNA spiked in human plasma

Sample No.	1	2	3	4	5
Known (fM)	50	200	1000	5000	10000
DNA biosensor (fM)	52.11±1.09	203.69±3.41	991±10.81	5087.72±99.78	9899.09±129.06
Recovery (%)	104.2%	101.8%	99.1%	101.8%	99.0%

Figure 5

Detection of ctDNA in the serum samples collected from cancer patients. Red color: before chemotherapy; Green color: after chemotherapy. The concentration of ctDNA was calculated from the calibration curve.

Table 2

The patient informations and ctDNA levels (fM) before and after treatment

Patient No.	1	2	3	4
Cancer type	Breast	Lung	Ovary	Rectum
Age	49	55	51	36
Gender	Female	Male	Female	Female
Treatment approach	Radiotherapy	Chemotherapy	Chemotherapy	Chemotherapy
Before therapy	3890.09±172.32	839.9±47.72	939.01±49.99	4502.02±214.38
After therapy	1060.11±43.33	1680.68±52.91	690.67±20.09	1979.55±96.68
Effect of therapy	Decreased	Increased	Unchanged	Decreased

Conclusions

In summary, we have successfully developed a dual biomarker and double amplification strategy-based detection platform and have demonstrated its use in an ultrasensitive and selective DNA biosensor for the analysis of ctDNA of PIK3CA gene. In this detection system, PNA probe and anti-5-mC were used to recognize the tumor-specific mutation and methylation of ctDNA, respectively, in order to improve the selectivity of the sensor. AuNPs played the role of nanocarrier to co-immobilize PNA probes at a high ratio, which could enrich more ctDNA, enhancing detection signal greatly. Additionally, apoferritin with the hollow structure was used to form metallic phosphates inside the cavity as voltammetrically distinguishable signal reporters to further improving detection sensitivity. Among the various technologies developed to analysis ctDNA to date, this is the first strategy to combine dual biomarker and an electrochemical biosensor for quantitative detection of ctDNA. More significantly, we have demonstrated the use of sensor to detect ctDNA both in complex biological matrices and cancer patient serum with satisfactory results. Therefore, we anticipate that this platform can be extended for the quantification of other nucleic acids or proteins, and we believe that it offers promising potential in the field of cancer diagnosis and personalized therapy.