Graphene oxide was a kind gift from Dr. Yang’s lab (The Hong Kong Polytechnic University, HK, China). Carboxyl-modified 525nm QDs were purchased from Invitrogen, Ltd. (USA). Phosphate buffered saline (PBS) pH 7.4 and bovine serum albumin (BSA) were prepared accordingly. N-hydroxysuccinimide (NHS) and 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Microcon molecular weight cut-off (MWCO) spin filters were obtained from Millipore Corporation (Bedford, MA, USA). Luria Bertani (LB) broth was purchased from Qingdao Hope Bio-Technology Co., Ltd. (China) and prepared by following supplier’s instruction. DNA extraction kits were purchased from Qiagen (Germany). A 30-mer single strand invA oligo (5′-CTTTCGTCTGGCATTATCGATCAGTACCAG-3′) and a 26-mer single strand control oligo (5′-GTGAAATTATCGCCACGTTCGGGCAA-3′) were extracted from the highly conserved Salmonella typhi murium invA gene (GenBank: M90846.1). Single-base mismatched oligo (M1: 5′-CTTTCGTGTGGCATTATCGATCAGTACCAG-3′), double-base mismatched oligo (M2: 5′-CTTTCGTGTGGCATTATCCATCAGTACCAG-3′) and the control oligo were synthesized for the specificity test of the developed system. Amine-modified capture A (5′-ATGCCAGACGAAAG/Aminolinker C7/-3′) and capture B (5′-/Aminolinker C6/CTGGTACTGATCGA-3′) that were complementary to the invA oligo were designed and synthesized. The primers that can specifically amplify the desired part of the invA gene were synthesized as well. The sequence of the forward primer is F-invA: GCCTACAAGCATGAAATGGCAGAAC and the reverse primer is R-invA: TCATCGCACCGTCAAAGGAACC. The length of the amplified product is about 649 bp. All the oligonucleotides listed in Table 1 were synthesized by Beijing Genomics Institute (Shenzhen, Guangdong, China) and prepared according to the supplier’s instruction.

Salmonella Typhimurium (S. Typhimurium) was inoculated in LB broth and maintained at 37°C overnight. 1.5 mL of the overnight culture was used to extract genomic DNA and suspended into 50 μL distilled water with a concentration of 0.6 μg/μL. 0.5 μL of the extracted Salmonella genomic DNA (∼0.3 μg) was used as template for PCR reaction. 10 μL of the purified PCR products (100 ng/μL) were used for the following detection analysis.

The process of conjugation between GO or QDs with capture A or B was realized by EDC/NHS assisted covalent bonding where EDC [1-Ethyl-3-(3-dimethylaminopropyl)-carbodiimide] is a zero-length cross-linking agent used to couple carboxyl or phosphate groups to primary amines and NHS (hydroxysuccinimide) was used as a stabilizer. The conjugation between GO and capture A was performed as described in a previous study where capture A was derived from the positive strand of and complementary to Salmonella invA gene (Shi et al., 2015). Briefly, freshly prepared NHS (5 mM) and EDC (1 mM) were added into the GO solution (5 mg/mL), vortex for 2 min and sonication for 15 min. Then the treated GO was mixed with 30 μM capture A, sonicated at room temperature (RT) for 1 h. The generated GO-capture A conjugate was further purified and washed by DI-H₂O by centrifugation at RT at 10,000 rpm for several times. In order to prevent unspecific binding of QDs with GO-capture A, the GO-capture A conjugate was further treated with 0.5 mg/mL BSA at RT for 30 min and then rinsed with DI-H₂O.

For the conjugation of QDs with capture B, which was derived from positive strand of and complementary to Salmonella invA gene, most of the procedures were carried out following the manufactures’ instructions, but with modifications. 50 μl QDs stock solution (8 μM) was diluted in 1xPBS (pH7.4), mixed with 20 μM capture B, followed by immediate addition of EDC (1 mg/mL). The mixture was then mixed by rotating at RT at dark environment for 2 h. The unbound capture B was removed by using a centrifugal filter and then washed by 1xPBS (pH7.4) for several times. 0.1% N₃Na was added in the final products and stored at 4°C in dark environment for future use.

Fourier transform infrared spectrum (FT-IR) spectra of GO, GO-capture A, QDs and QD-capture B were measured with a PerkinElmer Spectrum 100 FT-IR spectrometer (PerkinElmer Inc., USA). Zeta potentials of GO, GO-capture A, QDs and QD-capture B were characterized by a ZetaPlus Zeta Potential Analyzer (Brookhaven Instruments Corp., USA).

In the system, the invA oligo worked as a bridge to bring the GO-capture A and QD-capture B conjugates close enough by being complementary to both capture A and B, the energy emitted by QDs would be quenched by GO in the form of decreased fluorescence intensity. 0.5 mg/mL BSA passivated GO-capture A (60 μg/mL) was first mixed with invA oligo and incubated at 55°C for 10 min, then a desired concentration of QD-capture B (150 nM) was added, the mixture was incubated at 55°C for another 10 min. The total reaction volume was 50 μl. The fluorescence intensity was measured by using a Cary Eclipse Fluorescence Spectrophotometer (Agilent Technologies, USA) with the excitation wavelength set as 320 nm and emission range as 480–580 nm. For the specificity test of the system, the invA oligo was replaced either by M1 or M2 or control oligo in the reaction mixture, and for the application assay of the developed system, the relative PCR product was added in the place of invA oligo. All the assays were repeated for at least five times.

Standard Error of the Mean (SEM) was used to express the error in replicates. For group comparisons, the statistical test used was unpaired two-tailed t-test (P < 0.05). The statistical analyses were performed with GraphPad Prism 5 (GraphPad Software, Inc., USA).

The invA gene of Salmonella is highly conserved and has been used as a target for the detection of Salmonella previously (Zahraei Salehi et al., 2005). The principle of the developed Salmonella invA gene detection system is illustrated in Figure 1. In the system, carboxyl QDs (donor) and GO (quencher) were first conjugated with the capture B and A, respectively, with the aid of EDC/NHS. Upon the addition of the complementary invA oligo of Salmonella, the QDs and GO conjugates could be brought into close proximity to make the FRET pair work, the energy emitted from excited QDs would be quenched by GO. Based on this mechanism, the invA gene of Salmonella can be detected via the measurement of the fluorescence intensity change in the developed system.

After conjugation of GO and QDs with the corresponding oligonucleotides, the conjugates of GO-capture A and QD-capture B were analyzed by zeta potentials. When compared with that of GO, the zeta potential value of GO-capture A was about -25 mV, much closer to that of capture A alone, and both zeta potential peaks overlapped (Supplementary Figure S1A); so did the QD-capture B zeta potential value, which was about -12.5 mV, closer and overlapped with that of capture B alone as well (Supplementary Figure S1B), indicating that the surface of GO and QDs were covered by the corresponding captures after conjugation. In addition, the successful conjugation between GO and capture A or QDs and capture B were confirmed by FTIR spectra analysis as shown in Supplementary Figure S2. The characteristic amide vibration absorption peak could be obviously detected at around 1655 cm^-1 both in GO-capture A (Supplementary Figure S2A) and QD-capture B (Supplementary Figure S2B), suggesting that an amide bond has formed between the carboxyl group on GO or QDs and the amine group of oligonucleotides.

Moreover, in order to check whether the capture B oligo could alter the emission pattern of QDs or not under the excitation wavelength of 320 nm, the emission spectra of QDs and QD-capture B were compared, and both of them were found to display very similar emission patterns with the emission peak at about 520 nm (Figure 2).

Graphene oxide has relatively large surface area for binding but does not exhibit binding specificity. In order to reduce or avoid undesired interaction between GO and QD conjugates, the GO conjugates were first passivated by BSA. After passivation, the non-specific interaction between GO and QD conjugates could be decreased but could not be completely avoided (Figure 3A). To investigate the quenching efficiency of developed biosensor in the detection of invA gene, experimental and control assays were carried out. In the experimental assays, various concentration of synthesized invA oligo was studied with the passivated GO-capture A concentration fixed at 60 μg/mL, in the control assays, the only difference was that the passivated GO-capture A was replaced by the same concentration of passivated GO (without capture A). Theoretically, before saturation, the higher concentration of the invA oligo included in the reaction mixture, the higher quenching efficiency would be observed. The formula of quenching efficiency is Q = (F0 - Fq)/F0^∗100%, in which the F₀ represents the fluorescence intensity of QD-capture B before quenching and the F_q means that after quenching. About 50% quenching efficiency could be detected with as low as 10 nM invA oligo, but the quenching efficiency was not increased significantly with the higher concentration of invA oligo. For comparison, in the control assays, almost no quenching efficiency could be detected (Figure 3B). The limit of detection (LOD) of the present system in detecting invA gene of Salmonella was further determined which was ∼4 nM (Figure 4).

The biosensor specificity in detecting invA gene of Salmonella was analyzed by using mismatched oligonucleotides and control oligo. The quenching efficiency of 400 nM invA oligo was ∼52%, and that of 400 nM M1 and M2 were ∼45 and ∼43%, respectively, while that of control oligo was smaller than 10%, a significant difference from invA oligo (P < 0.0001, two-tailed t-test; Figure 5), suggesting that the system is very specific for detecting invA gene, while its discrimination power is not very high when the oligo is within couple of nucleotide difference from invA gene. Consistently, a higher selectivity was reported by applying similar detection approaches (Yang et al., 2008; Shi et al., 2015). In addition, J.S. Kim and colleagues witnessed very high Salmonella spp. detection specificity by using invA gene as target, with all of 10 Salmonella spp. could be specifically detected, but not the 40 non-Salmonella strains (Kim et al., 2011).

The invA gene has previously been used as target to specifically detect Salmonella spp. in food samples with high sensitivity (Kim et al., 2011). In order to check the possibility of the present biosensor in detecting the invA gene from Salmonella of environmental samples, specific primers were used to amplify the invA gene which cover the complementary fragment of probes that are conjugated onto GO and QDs. Compared with the same concentration of the invA oligo, 400 nM PCR product of invA gene could cause almost same degree of decrease in fluorescent intensity, and the fluorescent intensity varied accordingly with the change of the concentrations of PCR product of invA gene. The fluorescence intensity pattern from using 25 nM PCR product was almost the same as that of control assay, and for the assay with 300 nM PCR product, the intensity peak (around 520 nm) was about half of that from the control assay (Figure 6), indicating that the present biosensor has the potential to be applied in detecting the invA gene of Salmonella in food or environmental samples.