Enzyme-free ultrasensitive fluorescence detection of epithelial cell adhesion molecules based on a toehold-aided DNA recycling amplification strategy

Epithelial cell adhesion molecules (EpCAMs) play a significant role in tumorigenesis and tumor development. EpCAMs are considered to be tumor signaling molecules for cancer diagnosis, prognosis and therapy. Herein, an enzyme-free and highly sensitive fluorescent biosensor, with a combined aptamer-based EpCAM recognition and toehold-aided DNA recycling amplification strategy, was developed for sensitive and specific fluorescence detection of EpCAMs. Due to highly specific binding between EpCAMs and corresponding aptamers, strand a, which is released from the complex of aptamer/strand a in the presence of EpCAMs which is bound to the corresponding aptamer, triggered the toehold-mediated strand displacement process. An amplified fluorescent signal was achieved by recycling strand a for ultrasensitive EpCAM detection with a detection limit as low as 0.1 ng mL−1, which was comparable or superior to that of reported immunoassays and biosensor strategies. In addition, high selectivity towards EpCAMs was exhibited when other proteins were selected as control proteins. Finally, this strategy was successfully used for the ultrasensitive fluorescence detection of EpCAMs in human serum samples with satisfactory results. Importantly, the present strategy may be also expanded for the detection of other targets using the corresponding aptamers.


Introduction
Nowadays, cancer has become a worldwide problem which threatens public health and is the leading cause of death in China and the second leading cause of death in the United States. 1,2 Efforts in the ght against cancer will need sustained clinical and basic research to improve the effectiveness of diagnostic techniques and screening programs, which is critical for reversing the cancer epidemic worldwide. 2 Thus, the development of effective methods for cancer therapy has attracted increasing worldwide attention in the medical eld. Among them, the early identifying and quantifying of carcinoma biomarkers could provide an easier and more effective way to monitor the progression of carcinomas, which is of great importance for early accurate diagnosis and effective therapy for cancer. 3,4 EpCAM, a glycosylated transmembrane protein which is normally expressed in many epithelial tissues throughout the body, mediates epithelial-specic intercellular cell adhesion and is involved in cell signal transduction, proliferation, migration, differentiation and invasion. [5][6][7] However, in subsequent studies, overexpression of EpCAM was also found in human colon carcinomas, 8 breast cancer, 9,10 pancreatic cancer, 11 gallbladder cancer, 12 gastric cancer 13 and so on, but low levels or no expression in normal healthy tissues. 14,15 Thus, the overexpression of EpCAM has been regarded as a prognostic tumor biomarker associated with a poorer prognosis in a wide variety of different carcinomas and reects the existence and growth of tumors in the human body. 7 Due to this differential expression of EpCAM between human cancers and normal healthy cells, EpCAM plays a signicant role in tumorigenesis and tumor development and it is considered to be one of the prognostic tumor signaling molecules for cancer diagnosis, prognosis and therapy. 16,17 Therefore, the early sensitive and reliable detection of EpCAM is of great signicance for the early clinical diagnosis of tumors. Considering the signicant role of EpCAM in the early diagnosis of tumors, more attention has been given to developing quantitative methods for the detection of EpCAM in the past few decades. Many diagnostic strategies relying on anti-EpCAM antibodies have been developed. 18,19 Among these, enzyme-linked immunosorbent assays (ELISA) represent the major approach for the sensitive detection of EpCAM. 7,20,21 However, this antibody-based method is usually labor intensive, complicated, expensive, time-consuming and even requires highly skilled personnel. To date, only a few novel strategies, including electrochemical biosensors 20 and uorescence biosensors, 17,22 have been developed for the sensitive determination of EpCAM. Among them, uorescence biosensors are particularly attractive due to their high sensitivity, easy readout, simplicity and the feasibility of quantication. 23,24 Aptamers are single-stranded functional DNA or RNA structures that are obtained in vitro from large random-sequence nucleic acid libraries by the systematic evolution of ligands by exponential enrichment (SELEX) technology. [25][26][27] They are capable of easily recognizing and binding specic targets including metal ions, small molecules, proteins and even whole viruses or cells. 28,29 In comparison with antibodies, aptamers possess numerous unique advantages including design exibility, ease of modication, easy and controllable labeling, high specicity, high purity, long-term stability and so on. 4,24 Based on these merits, they have been attracting increasing research efforts as alternative bio-recognition elements to antibodies for biosensor design. Meanwhile, the ratio of aptamer to target is 1 : 1 in almost all technologies, resulting in low sensitivity and a high error rate. The sensitivity of these reported aptamerbased uorescence detection systems is compromised. To overcome these limitations, signal amplication strategies, including enzyme-aided signal amplication (nicking endonucleases, exonucleases, DNAzymes, etc.), 23,30-33 catalyzed hairpin assembly (CHA), 34 molecular machines, 35,36 the hybridization chain reaction (HCR), 37 rolling circle amplication, 38 nanoparticle-assisted amplication 39 and toehold-aided DNA recycling amplication, 40,41 have recently been developed to achieve the sensitive detection of biomolecules in the eld of bio-analytical sciences. Among these signal amplication strategies, toehold-aided DNA recycling amplication has the advantages of being enzyme-free, easy to use and inexpensive, having continuous signal turnover capability and inherent modularity and being easy to scale up, 32,33 and is especially intriguing for signal amplication. Toehold-aided DNA recycling amplication has overcome the disadvantages of the specic reaction conditions and reaction time dependent enzyme activity of enzyme-aided signal amplication, 34 and reversed the low specicity caused by great background signals due to nonspecic CHA products in the absence of a target. 42 Toehold-aided DNA recycling amplication is a controllable independent process based on a toehold-mediated strand displacement process, without the participation of various enzymes or nanomaterials, which does not have the disadvantages of expensive price, poor stability, complicated operation, specic reaction conditions or reaction time dependent enzyme activity. 34 It was rstly pioneered for the construction of a tweezer-like dynamic molecular machine by Yurke et al. 36,40 A toehold, a short single-strand overhanging domain of doublestranded complex to which the target sequence attaches and then compels one DNA strand in a double-stranded complex to migrate away, triggers the strand displacement process. 33,43 Toehold-mediated strand displacement exhibited high sequence-dependence and was successfully applied to the construction of nanomachines, 36,44 molecular self-assembly, 45 logic gates, 46,47 signal amplication, 40 neural networks 48 and so on. To meet the demands of the specicity, sensitivity and feasibility of EpCAM detection, the development of an enzymefree signal amplication strategy is extremely urgent.
Considering the specicity of aptamer-based biosensors and the intriguing characteristics of toehold-aided DNA recycling amplication, the combination of the two strategies is promising for the specic and sensitive detection of EpCAM. Herein, we report an aptamer-based enzyme-free approach for the ultrasensitive uorescence detection of EpCAM using a toeholdaided DNA recycling amplication strategy. The toehold-aided DNA recycling amplication strategy, with cyclic reuse of the initiation strand (strand a) for the direct uorescence detection of EpCAM, was developed successfully for the ultrasensitive detection of EpCAM at levels as low as 1.0 ng mL À1 , with a linear range from 2 ng mL À1 to 150 ng mL À1 . Moreover, the suitability of this approach for the sensitive determination of EpCAM in real serum was also investigated, with recovery in the range of 109.2-114.8%. Therefore, the developed strategy will become a promising and reliable method for the ultrasensitive detection of EpCAM in the early clinical diagnosis of cancers and medical research.

Reagents and materials
EpCAM, bovine serum, CD86 and CD63 were purchased from Cusabio Biotech Co. Ltd. The aptamers and synthetic DNA sequences (Table 1) were all purchased from Sangon Biotechnology Co. Ltd. (Shanghai, China, www.sangon.com) and puri-ed using HPLC. The other reagents employed were of analytical grade and used without further purication. All reagents were diluted to the required concentration with working buffer (20 mM Tris-HCl, 100 mM NaCl, 10 mM KCl and 10 mM MgCl 2 ; pH 7.5) before usage. Healthy human serum was obtained from the Dongfeng General Hospital. Ultrapure water prepared with a Millipore water purication system (18.2 MU cm resistivity, Milli-Q Direct 8) was used in all runs.

EpCAM sensing procedure
Prior to the experiments, the mixtures of aptamer/strand a and strand b/strand c/strand d hybridized strands were heated at 90 C for 5 min, and then slowly cooled down to room temperature. Equimolar concentrations of strand b, strand c and strand d were mixed. Next, 100 mL EpCAM at different concentrations was incubated with 100 mL aptamer/strand a for 30 min at 37 C. This was followed by the addition of the 100 mL mixture of strand b/strand c/strand d. Subsequently, 100 mL strand e was introduced and the solution was incubated at 37 C. Finally, the solution was diluted to 1 mL and the uorescence intensity of the solutions was measured.

Fluorescence measurements
The uorescence detection of the mixture was carried out using a Hitachi F-4600 spectrophotometer (Hitachi Co. Ltd., Japan, www.hitachi.co.jp) equipped with a xenon lamp excitation source at room temperature. The excitation was set at 495 nm and the emission spectra were collected from 510 nm to 600 nm. The uorescence intensity at 518 nm was used to investigate the optimal experimental conditions and evaluate the performance of the proposed sensing system. In the control experiments, the measurement process was the same as above, except for the addition of EpCAM. Unless otherwise noted, each uorescence measurement was repeated three times, and the standard deviation was plotted as an error bar. The quantitative assay of EpCAM was realized using the uorescence intensity. F 1 and F 0 are the uorescence intensities at 518 nm in the presence and absence of EpCAM, respectively.

Design principles
In the present study, a enzyme-free uorescence amplication strategy for sensitive EpCAM detection on the basis of a combined aptamer-based EpCAM recognition and toeholdaided DNA recycling amplication strategy was developed. As illustrated in Scheme 1, strand a, which triggers the toeholdmediated strand displacement reaction, is rstly hybridized with the EpCAM aptamer sequence to form an aptamer/strand a duplex, preventing the occurrence of a strand displacement reaction in the absence of EpCAM. Strand b, strand c and strand d are hybridized to form a b-c-d duplex with weak uorescent emission, due to the quencher-contained strand b quenching the uorescence of the uorescence reporter of strand d. In the presence of EpCAM, strand a can be dissociated from the aptamer/strand a duplex due to the highly specic affinity between EpCAM and the corresponding aptamer. The liberated strand a further hybridizes with the toehold domain of the b-c-d duplex and triggers the toehold-aided strand displacement reaction, leading to the release of strand b to form an a-c-d duplex, which increases the uorescent signal. Upon addition of strand e, it hybridizes with the toehold domain of the above formed a-c-d duplex and displaces strand a and strand c. The liberated strand a hybridizes again with the toehold domain of the b-c-d duplex and triggers a toeholdaided DNA recycling amplication, leading to signicantly amplied uorescence emission for the ultrasensitive detection of EpCAM.

Feasibility analysis of the developed method for EpCAM detection
To further verify the feasibility of the toehold-aided DNA recycling uorescent signal amplication strategy, uorescence measurements were performed to record the uorescence emission spectra of different mixtures. As shown in Fig. 1, when compared with the highly uorescent signal of DNA strand d (curve a), an extremely weak uorescent signal for the b-c-d duplex was obtained, which was attributed to the quencher-contained strand b quenching the uorescence of the uorescence reporter of strand d (curve b vs. curve a). On the addition of strand e, a very slightly increased uorescent signal was observed (curve c vs. curve b), indicating the partial dissociation of strand b from the b-c-d duplex in the presence of strand e. The reason for this may be that, although the binding capacity of strand e to strand d is greater than that of strand b and strand c to strand d, there were no unpaired bases to hybridize with strand e when strand d was rst hybridized with strand b and strand c, resulting in the slow reaction of strand e displacing strand b and strand c over a short period of time. Similarly, the incubation of aptamer/ strand a, b, c, d and e showed negligible uorescence intensity changes compared with that of strand b, c, d and e (curve d vs. curve c). On the addition of strand e, a signicant enhancement in uorescence intensity was further exhibited in the presence of EpCAM (curve f vs. curve e). This apparent signal enhancement further indicated the successful release of strand a from the a-c-d duplex and subsequent toehold-aided DNA Table 1 The aptamer and synthetic DNA strand sequences used in this work recycling amplication in the presence of strand e. A signicant enhancement in uorescence intensity was also observed in curve g, indicating the occurrence of the toehold-mediated strand displacement reaction in the presence of strand a. The gel electrophoresis results also conrmed the above results (Fig. S1 †). As we all know, Gibbs free energy can also reect the stability of a DNA duplex. A smaller Gibbs free energy value indicates better stability for the hybridization of two complementary strands. The occurrence of toehold-aided strand displacement was rst based on the strands with smaller Gibbs free energy displacing the strands with larger Gibbs free energy in the presence of unpaired bases in complementary strands. 43 Therefore, the Gibbs free energy of the formation of different duplexes was also analyzed using online soware (http://www.nupack.org/). Fig. 2 shows the secondary structures and the free energy of the formation of the corresponding duplexes: strand a/aptamer, strand a + strand d, strand d + strand b, strand d + strand e and strand d + strand c, which further veries the aforementioned results.

Optimization of reaction conditions
In order to achieve optimal sensing performance using the proposed toehold-aided DNA recycling amplication strategy for EpCAM detection, several reaction conditions such as the concentration of aptamer/strand a, the concentration of strand d, the concentration of strand e and the reaction time were optimized. The uorescence intensity and the value of F 1 /F 0 were selected to evaluate the effects of the aforementioned reaction conditions on the sensing performance of the developed method, where F 1 and F 0 were the uorescence intensities of the solutions in the presence and absence of EpCAM, respectively. As depicted in Fig. 3(a) 2 The secondary structures and the free energy of the formation of the corresponding duplexes at 37 C (the incubation temperature) were analyzed using online software (http://www.nupack.org/); strand a + aptamer, strand a + strand d, strand d + strand b, strand d + strand e and strand d + strand c. efficiency, because a great deal of the initiation strand (strand a) or product sequences released in the reaction need to hybridize with adequate quantities of strand e or the b-c-d duplex in order to trigger toehold-aided DNA recycling amplication. 23 Secondly, high concentrations of the b-c-d duplex and strand e may give rise to an increase in background signal because of nonspecic amplication. Therefore, the concentrations of the b-c-d duplex and strand e were also investigated. As shown in Fig. 3(b) and (c), uorescence intensity increased gradually along with increasing concentration of the b-c-d duplex in the range from 60 nM to 140 nM and increasing concentration of strand e in the range from 80 nM to 120 nM in the presence or absence of EpCAM (50 ng mL À1 ). The F 1 /F 0 value reached a maximum when the concentrations of the b-c-d duplex and strand e are both at 100 nM. Therefore, a concentration of 100 nM was selected as the optimized concentration of both the b-c-d duplex and strand e.
The reaction time is another important reaction condition affecting uorescence intensity. The plots depicted in Fig. 3(d) represent changes in uorescence intensity and F 1 /F 0 values along with reaction time varying from 5 min to 25 min at time intervals of 5 min. The F 1 /F 0 value reached a maximum when the reaction time was 15 min and then decreased gradually because of an accelerated increase in background uorescent signal. Thus, 15 min was conrmed as the optimized reaction time.

Sensitivity for EpCAM detection
Under optimized reaction conditions, the sensitivity of the proposed toehold-aided DNA recycling amplication strategy for EpCAM detection was evaluated at different concentrations. As shown in Fig. 4(a), uorescence intensity gradually increased along with the concentration of EpCAM, from 0 to 300 ng mL À1 . By plotting the curve of uorescence emission intensity vs. concentration of EpCAM at an emission wavelength of 518 nm (shown in Fig. 4(b)), a good linear relationship between the uorescence intensity and the concentration of EpCAM was obtained, in the range from 2 ng mL À1 to 150 ng mL À1 , with a regression coefficient (r 2 ) of 0.9804 and a detection limit of 0.1 ng mL À1 (obtained according to the 3s rule). This low detection limit which provided excellent sensitivity was comparable to the reported immunoassays and electrochemical microuidic immunosensor, 20 and higher than that of the reported uorescence biosensor for EpCAM detection. 17 The excellent sensitivity and broad linear range indicated that this toehold-aided DNA recycling amplication strategy was satisfactory for the ultrasensitive uorescence detection of EpCAM.

Specicity for EpCAM detection
Due to high recognition and specic affinity between the aptamers and the targets, the aptamer-based biosensors exhibit signicant specicity. In the present work, three different relevant proteins, including BSA, CD86 and CD63, with a concentration of 100 ng mL À1 at 2-fold higher than that of EpCAM, were spiked respectively. The measurements were performed under the same conditions to validate the specicity of the proposed method for EpCAM detection. As shown in Fig. 5, in the presence of other control proteins (100 ng mL À1 ), slight uorescence changes were observed in the absence of EpCAM, while a signicant enhancement of uorescence emerged in the presence of EpCAM (50 ng mL À1 ), which indicated the excellent specicity of the proposed strategy for the detection of EpCAM.

Determination of EpCAM in real samples
To further verify the potential applicability of the present strategy, the detection of EpCAM in biological samples by spiking human serum and 50% serum (diluted with buffer) (human serum obtained from Dongfeng General Hospital) with various concentrations of EpCAM was performed according to the EpCAM sensing procedure. As shown in Fig. 6, a signicant increase in uorescence in the presence of 50 ng mL À1 EpCAM in undiluted serum was observed when compared with the blank test (in the absence of EpCAM), while a negligible change in uorescence intensity in serum was observed when compared with 50% serum or buffer (50 ng mL À1 EpCAM), which indicated that the detection of EpCAM in serum is free of The fluorescence intensity vs. the concentration of EpCAM, from 0 to 300 ng mL À1 , at an emission wavelength of 518 nm. Inset: the linear relationship between fluorescence intensity and EpCAM concentration in the range from 2 ng mL À1 to 150 ng mL À1 . Error bars: SD, n ¼ 3. Fig. 5 Selectivity investigation of the proposed method for the detection of EpCAM (50 ng mL À1 ), BSA (100 ng mL À1 ), CD86 (100 ng mL À1 ), CD63 (100 ng mL À1 ). Error bars: SD, n ¼ 3.

Conclusions
In summary, a uorescence biosensor for ultrasensitive EpCAM detection was rstly constructed by combining toehold-aided DNA recycling amplication with aptamer-based target recognition, without the participation of enzymes. Due to a signicant uorescent amplication signal in response to EpCAM and aptamer-based EpCAM recognition, the sensitive and specic detection of EpCAM was achieved, with a detection limit as low as 0.1 ng mL À1 . In addition, this approach has been successfully applied in the specic and sensitive detection of EpCAM in real samples, indicating that it will become a reliable method for EpCAM detection in the early clinical diagnosis of cancers. Moreover, the present enzyme-free ultrasensitive uorescence biosensing strategy may be also a promising strategy for the direct detection of other biomarkers by selecting the corresponding aptamers in early clinical diagnosis.

Conflicts of interest
There are no conicts to declare.