Nucleic Acid Sandwich Hybridization Assay with Quantum Dot-Induced Fluorescence Resonance Energy Transfer for Pathogen Detection

2.1. Target Sequences and Oligonucleotide Probes

A label-free, 60-mer conserved H5 sequence (5′-AAG ATA GAC CAG CTA CCA TGA TTG CCA GTG CTA GGG AAC TTG CCA CTG TTG AAT AAA TTG-3′) was used to demonstrate the feasibility of the sandwich hybridization assay. Two oligonucleotides that specifically recognize two separate but neighbor regions of the target sequence were served as the capturing probe (5′-CAG TG G CGA G(T/C)T CCC T-3′) and the reporter probe (5′-ACT GGC AAT CAT GGT (A/G)GC-3′), respectively. The 3′ end of the capturing probe was modified with a thiol (-SH) linker for conjugation with Qdot 655 ITK™ amino (PEG) quantum dots, namely QD655 (Invitrogen, Carlsbad, CA, USA) (FRET donor). The 5′ end of the reporter probe was labeled with an Alexa Fluor 660 dye (FRET acceptor). A 60-mer hemagglutinin H1 sequence (5′-AGT ACC AAT GAA CTG GCG ACA GTT GAA TAG ATC GCC AAA ATC TGG TAA ATC CTT GTT GAT-3′) with 40.6% sequence similarity to the 60-mer H5 target was used as a non-target control in this study. All the sequences with and without terminal modifications were all synthesized and purified by Integrated DNA Technology Inc. (Coraville, IA, USA). The construction of a 630-nt H5 cDNA fragment is described as follows. First, the full-length H5 gene sequence from HK/156/97/H5N1 AIV (accession number AF028709) was synthesized by overlapping oligonucleotide assembly approach, in which 88 oligonucleotides (40-mer) encoding both strands of the H5 sequence were combined and assembled in a two-stage PCR protocol [28]. The H5 gene was then in vitro transcribed into its corresponding RNA, from which a 630-nt cDNA fragment, ranging from nt 1,112 to 1,741 in the H5 gene, was synthesized by reverse transcription. For spiking experiments, total nucleic acids from human bronchial epithelial cells (BEAS-2B cells) were extracted by High Pure Viral Nucleic Acid Isolation kit (Roche Diagnostics, Mannheim, Germany).

2.2. Conjugation of Capturing Probes with Quantum Dots

The protocol used for conjugation of capturing probes with Qdot 655 ITK™ amino (PEG) quantum dots (QD655) was modified from the steps reported in the literature [29] and is described briefly as follows. At first, 62.5 μL of QD655 (8 μM) was buffer exchanged to 250 μL of Buffer A (50 mM sodium phosphate and 150 mM sodium chloride, pH 7.2) using Amicon Ultra-4 (100 kDa cutoff) centrifugal filter units (4,000×g at 4 °C for 15 min). Then 250 μL of sulfo-SMCC crosslinker (2 mM) (Pierce, Rockford, IL, USA, in 1,000-fold excess, in Buffer A) was added into the QD suspension and allowed to react at room temperature for 1 h. The resulting sulfo-SMCC crosslinked QDs were filtered by a NAP-5 column (GE Healthcare, Buckinghamshire, UK) to remove excess cross-linker and eluted with 1 mL of Buffer B (50 mM sodium phosphate, 150 mM sodium chloride, 10 mM EDTA, pH 7.2). 5 μL of 1 mM capturing probes with 3′ thiol linkers were mixed with 245 μL of 0.1 M dithiothreitol at room temperature for 1 h. The mixture was then filtered by a NAP-5 column into 0.7 mL of Buffer A and reacted with the above-mentioned sulfo-SMCC crosslinked QDs at 4 °C overnight. In theory, the capturing probes were conjugated to QD655 in a molar ratio of 10:1 (probe:QD).

The conjugated products were washed twice with 4 mL of phosphate buffered saline (PBS), twice with a 4 mL of high salt buffer (1.0 M sodium chloride, 100 mM sodium citrate, pH 7.2), and then twice again with 4 mL of PBS using Amicon Ultra-4 (100 kDa cutoff) centrifugal filter units according to the vender’s recommendation. The success of QD655 conjugation was confirmed by 2% agarose gel electrophoresis (Figure 2). The QD conjugates were then blocked by 0.05% denatured bovine serum albumin at 60 °C for 1 h [30,31] to prevent false positive FRET signals due to possible non-specific adsorption of fluorophore-labeled oligonucleotides on the QDs. Finally, the blocked QD-conjugated capturing probes were purified using Amicon Ultra-4 100 kDa cutoff concentrator and resuspended in 50 μL of PBS.

2.3. Nucleic Acid Sandwich Hybridization

20 μL mixtures of the capturing probes (conjugated with QD665), the reporter probes (labeled with Alexa Fluor 660 dye) and the H5 target DNA at different molar ratios were prepared in a sample buffer (20 mM Tris-HCl, pH 8.4, 50 mM KCl, 7 mM MgCl₂) and transferred onto an ITO glass slide for 1.5 h of incubation at 37 °C before FRET detection. The concentrations of capturing and reporter probes used were quantified by the concentrations of the conjugated QD (denoted as [QD]) and the labeled Alexa Fluor 660 dye (denoted as [dye]), respectively (note: in the ideal case, the capturing probe-to-QD ratio and the reporter probe-to-dye ratio are 10:1 and 1:1, respectively). The concentration of H5 target is denoted as [H5]. The concentrations of QD, Alexa Fluor dye and target DNA were all quantified with a UV-visible spectrophotometer (WPA S2100, Biochrom Ltd., Cambridge, UK).

2.4. Detection of Sandwich Hybridization with QD-Induced FRET

The QD-induced FRET spectra of the capturing probe/H5 target/reporter probe complex were monitored by a homemade, low weight and cost-effective optical sensor system. The optical sensor system was constructed based on a single UV LED featuring a narrow emission band between 395 and 410 nm (EFEV-1AE1, flux(lm) = 160 mW, V_f = 3.0–4.0 V, Edison Opto, Taipei, Taiwan), an ITO glass slide having dimensions 7.62 × 2.54 × 0.1 cm³ with transmittance ≥85% in 400–700 nm and R_sh = 7 ohm/sq (Ritdisplay Corporation, Hsin-Chu, Taiwan), and a miniature 16-bit CCD-array spectrophotometer (USB 4000, Ocean Optics, Dunedin, FL, USA). An optical fiber with a diameter of 0.6 mm (QP600-2-UV-VIS multimode fiber, Ocean Optics) was striped and trimmed and then connected into a standard SMA 905 connector (Ocean Optics) to transmit the light. Moreover, additional optical filters (Newport Corp., Irvine, CA, USA) on excitation (500 nm shortpass) and emission (500 nm longpass) sites were used to reduce the background signal. Figure 1 shows the setup of the FRET detection device. The 20 μL, transparent microchamber on the ITO glass slide was constructed by applying a sticky 25 μL Microarray Gene Frame^® (AB-0576, area = 1 × 1 cm², thickness = 0.25 mm, ABgene, Epsom, UK) onto the ITO conductive surface along with a transparent, plastic cover slip. The assembly of Gene Frame^® and cover slip were thermally stable and remained well-sealed up to 97 °C. The hybridization mixture was transferred to the microchamber with a micropipette. Then the miniature CCD-array spectrophotometer, connected to a laptop computer, was used to collect the emission spectra detected by the optical sensor system.

3.1. Sandwich Hybridization Assay with a QD-Induced FRET Reporter System

In general, the application of FRET technique to quantification of a binding event can be carried out by either the emission-shift method or the donor-quenching method [32]. The emission-shift strategy can exhibit two-dimensional signals (decrease in QD emission and increase in dye emission) and was used in this work. Because the emission maximum of QD655 and the excitation maximum of Alexa Fluor 660 occur at 653 and 660 nm, respectively (Figure 3), the high overlap between the spectra of QD655 (donor) emission and Alexa Fluor 660 (acceptor) excitation could induce high-efficiency FRET, so the QD-dye FRET pair was used as the reporter system in the sandwich hybridization assay. Figure 4 illustrates the molecular design for the sandwich hybridization assay. As shown in Figure 4(a), capturing probes are conjugated to QD655, while each reporter probe carries an Alexa Fluor 660 dye label. If the sandwich hybridization occurs, the binding event between the two probes and the target will cause spatial approaching between QD and dye and then induce a FRET effect under UV excitation (Figure 4(b)). Thus, the FRET reporter system can turn target recognition into a fluorescence shift signal and the extent of the FRET emission shift should depend on the target concentration—An increase in target concentration leads to a decrease in QD655 emission but an increase in Alexa Fluor 660 emission. Here the FERT efficiency (E, %) and FRET value (R), as defined below, are calculated to quantify the percentage of energy transfer and the extent of the FRET emission shift, respectively. The optimal assay condition for H5 detection is determined accordingly:

3.2. Toward the Target Detection with the FRET Emission Shift

In theory, the aforementioned sandwich hybridization should result in a FRET emission shift from 653 nm to 690 nm, but we found that a reasonable high FRET value as well as a detectable emission shift occurs when the sandwich hybridization is done with a high dye-to-QD ratio. At the beginning of our study, we employed a high concentration of QD655 donor ([QD] = 1 μM) and Alexa Fluor 660 dye ([dye] = 1 μM) in the sandwich hybridization assay. As shown in Figure 5(a), the QD emission obviously decreases with the increase in target concentration, but the QD emission decrease does not turn into an observable dye emission increase (for [H5] = 1 μM, E = 34.8% and R = 0.05; the background FRET value is about 0.02). To investigate whether the QD emission decrease is due to the hybridization of QD-dye FRET pair with the H5 target, we carried out a negative control study in which 60-mer non-target H1 sequence ([H1] = 1 μM) was hybridized with QD-dye probe pair ([QD] = [dye] = 1 μM). The emission spectrum in the hybridization is not different from that for QD-dye pair ([QD] = [dye] = 1 μM) without the addition of H5 target ([H5] = 0 μM). This result indicates that the QD-dye FRET probe pair shows no significant response to the H1 sequence and the aforementioned decrease in QD emission should be due to the hybridization of the FRET pair to the H5 target.

Figure 5(b) displays the spectra with 10–30 μM dye and 1 μM QD for 1 μM target and shows that an obvious emission shift is detected when the dye concentration is 30 μM (In this case, E = 74.9% and R = 0.76). Thereafter, the FRET effect gradually approached to saturation since the further increase in dye concentration >30 μM did not significantly enhance the FRET emission shift signal (data not shown). The aforementioned result indicates that a high dye-to-QD ratio can enhance both the FRET efficiency and FRET value. Thus, if both the QD emission decrease at 653 nm and the dye emission increase at 690 nm are to be correlated to the target concentration in the sandwich hybridization assay, a high dye-to-QD ratio is required. Presumably, the corresponding reason is twofold. First, in theory, a high dye-to-QD ratio causes more acceptor fluorophores gathering around a QD donor particle during sandwich hybridization and enhances the FRET efficiency and thus the FRET value. Second, a reporter probe molecule carries only one Alexa Fluor dye. That is, the dye concentration is equal to the concentration of the reporter probe. On the other hand, the capturing probes were conjugated to QD655 in a molar ratio of 10:1, so the capturing probe concentration is considered 10 times of the QD concentration in the ideal case. The actual active sites on each QD surface that allow anchoring DNA probes might be even higher than 10 [33]. Hence, a high dye-to-QD ratio is considered favored, because it gives a comparable or excess quantity of reporter probes for the capturing probes on QDs. Thus, a dye-to-QD ratio of 30:1 was used in the subsequent sandwich hybridizations in order to obtain FRET emission shift spectra information.

3.3. Correlation of Target Concentration with FRET Emission Shift Data

Figure 6(a) shows a representative spectral dataset for a series of sandwich hybridization assays with 50 nM QD, 1.5 μM dye (i.e., a dye-to-QD ratio of 30:1) and different concentrations of 60-mer H5 target (0–50 nM). It can be seen that the target DNA concentration can be correlated with both signal decrease in QD emission at 653 nm and signal increase in dye emission at 690 nm. The average FRET values from triplicate experiments for the target concentrations at 0.5 nM, 5 nM and 50 nM are 0.21, 0.35, and 0.59, respectively and their corresponding relative standard deviations (RSD) are 11.6, 7.1, and 4.9%, respectively (inset of Figure 6(a)). The correlation coefficient between the target concentration and FRET value is 0.975. Since the background FRET value (when the target is absent) is 0.08 (RSD = 13.4%), the estimated limit of detection (LOD), a concentration corresponding to a FRET value three standard deviations above the background FRET value, is about 0.27 nM for 50 nM of QD donor used in the sandwich hybridization, which is comparable to or better than several QD-based FRET bioassays with LODs of 1–12 nM [19,20,31,34].

3.4. Demonstration of Clinical Feasibility for the QD-Induced FRET Reporter System

Accurate detection of target in complex mixture is highly required in clinical diagnosis. The nucleic acid-based influenza virus detection from throat or nasal swab specimens generally involves the extraction of influenza viral nucleic acids along with a huge background of human nucleic acids. Therefore, to investigate whether our sensor platform could be used to detect the target in clinical settings, a 630-nt H5 cDNA fragment was served as a real sample substituent and spiked into 5 μM human total nucleic acids extracted from human bronchial epithelial (BEAS-2B) cells in the sample buffer, in which the same sandwich hybridization experiments as indicated in Figure 6(a) were performed again and the result is shown in Figure 6(b). Apparently, the similar FRET responses to different target concentrations shown in Figure 6(a) were observed again. This indicates that the detection of H5 target in a non-target complex background is feasible. Notably, the average FRET values (inset of Figure 6(b)) for the target concentrations at 0 nM, 0.5 nM, 5 nM, and 50 nM were decreased to 0.06 (RSD = 12.0%), 0.12 (RSD = 11.3%), 0.19 (RSD = 5.6%), and 0.35 (RSD = 5.9%), respectively. The estimated LOD in this case is about 0.5 nM. The decrease in the LOD is likely due to the decreased hybridization efficiency of the target-probe hybrids in highly non-target rich mixture.