GoldMag nanoparticles (5 mg/ml) with a nanoflower structure (21) and lateral flow strips were provided by Xi'an GoldMag Nanobiotech Co., Ltd. (Xi'an, China). Buffers were prepared according to standard laboratory procedures. Water (18.2 MΩ cm) purified by the Barnstead Nanopure Water System (Thermo Fisher Scientific, Inc., Waltham, MA, USA) was used for all sample preparation. All chemicals listed below were of analytical grade. Cetyltrimethylammonium bromide (CTAB; 99% purity), dimethyl sulfoxide (DMSO) and PSS sodium salt (molecular weight, 70 kDa) were obtained from Sigma-Aldrich (Merck Millipore, Darmstadt, Germany). A mouse anti-digoxin antibody (catalog no. MAD53-310C) was purchased from Meridian Life Science, Inc. (Cincinnati, OH, USA). An anti-fluorescein isothiocyanate (FITC) antibody (catalog no. bs-0366R) and a goat anti-mouse IgG antibody (catalog no. bs-0296 G) were obtained from Beijing Biosynthesis Biotechnology Co., Ltd. (Beijing, China). Streptavidin was obtained from Promega Corporation (Madison, WI, USA). All labeled oligonucleotides were synthesized by Invitrogen (Thermo Fisher Scientific, Inc.). The sequences of each oligonucleotide are listed in Table I. HotMaster Taq DNA Polymerase kit was purchased from Tiangen Biotech Co., Ltd. (Beijing, China), and dNTP and uracil-DNA glycosylase (UDG) polymerase were from ShineGene Bio-Technologies, Inc. (Shanghai, China). The Lowry protein assay kit, bovine serum albumin (BSA) and calf serum were from Sigma-Aldrich (Merck Millipore).

The Fourier transform-infrared (FT-IR) spectrum of particles was recorded using a Nicolet 5700 FT-IR spectrometer (Thermo Fisher Scientific, Inc.), followed by drying. A Hitachi H-600 transmission electron microscope (TEM; Hitachi, Ltd., Tokyo, Japan) was used to acquire images of particles, whereas particles size and zeta potential were characterized by dynamic light scattering using Zetasizer Nano ZS (Malvern Instruments Ltd., Malvern, UK). A 2550 UV-visible spectrophotometer (Shimadzu Corporation, Tokyo, Japan) was used to determine the surface plasmon resonance (SPR).

A pure core/shell of Fe3O4/Au nanoparticle was obtained by dispersing GoldMag nanoflower (Fe3O4/Au/Fe3O4) with cationic surfactant, CTAB (22). After ultrasonic treatment at 45 Hz for 20 min, 10 mg nanoflower particles were gently mixed with 6 ml CTAB (5 mmol/l) and sonicated for 40 min. Subsequently, the CTAB-GMNs (GoldMag nanoparticles) were separated magnetically through the application of an external permanent magnet, and the supernatant (solution containing uncapped CTAB) was removed. A PSS solution (6 ml, 0.1 mg/ml) was added to the CTAB-capped GMNs particles under sonication for 30 min and then left to stand for 2 h. PSS-GMNs were magnetically separated and the supernatant was discarded; this step was repeated twice. Then, the PSS-GMNs were suspended in 7 ml of deionized water.

PSS-GMNs (1 mg) were equilibrated in the 600 µl of phosphate buffer (1X PB, pH 7.2), containing 40 µg of the targeted moiety (anti-digoxin or anti-FITC antibody). This mixture was shaken at 180 rpm for 1 h at 22°C. After 1 h of incubation, the unconjugated antibodies were removed by washing in a magnetic field. A blocking buffer (1X PB buffer, pH 7.2, containing 3% BSA and 5% calf serum) was added to the conjugates, and the mixture was incubated for 1 h. After incubation for 2 h, anti-digoxin antibody or anti-FITC antibody functionalized PSS-GMNs conjugates, were magnetically separated and then suspended in buffer (1X PB, pH 7.2, containing 1% BSA) at 2–8°C prior to use. The conjugation efficiency was calculated by determining the concentration of anti-digoxin or anti-FITC antibody in the solution prior to and following coupling using the Lowry protein concentration assay, with BSA as a protein standard (23).

Streptavidin and goat anti-mouse IgG were printed on a porous nitrocellulose membrane to form the test line (T line) and control line (C line), respectively, using a HM3010 BioJet dispenser (BioDot, Inc., Irvine, CA, USA). To detect the genotype of each SNP, two complementary strips were run using half the volume of the same PCR product separately. One strip detects wild-type alleles (WT channel) and the other detects mutant alleles (M channel). The probe solution containing PGMNs with an anti-digoxin antibody or anti-FITC antibody was dispensed on the conjugate pad of WT channel and M channel of the LFA, respectively. These strips were placed in a card box and stored in a sealed aluminum foil bag with desiccant silica gel at room temperature. The strips remain stable for 12 months.

According to the principle of ARMS-PCR (24), the reverse primers were designed as biotin-labeled common primers, and the forward primers are allele specific primers with the nucleotide at their 3′ terminus corresponding to the SNP site. To detect the genotype of each sample, two complementary reactions (ApoE 158 tube and ApoE 112 tube) were run separately. The primers were combined in two reaction mixtures to yield predicted amplification products of 451 bp (Reaction ApoE 158: ApoE 158C, ApoE 158T) and 588 bp (Reaction ApoE 112: ApoE 112T, ApoE 112C). Each PCR reaction was performed in a total volume of 50 µl containing 10X PCR buffer (10 mM Tris HCl, 1.5 mM MgCl2), 2.5 µl DMSO, 0.2 mM dNTP mixture (dATP, dCTP, dGTP and dUTP), 0.5 U HotMaster Taq DNA Polymerase, 0.5 U of UDG polymerase, 0.1 µM common reverse primer and 4 µl DNA template (4 µl TE buffer without DNA served as a control in all experiments). The ApoE 158 reaction mixture also contained 0.05 µM ApoE 158C and ApoE 158T forward primers. Similarly, the ApoE 112 reaction mixture contained 0.05 µM ApoE 112T and ApoE 112C forward primers. An Applied Biosystems 2720 PCR Thermal Cycler (Thermo Fisher Scientific, Inc.) was used for the amplification. The amplification was commenced with a UDG incubation step (50°C for 2 min), initial denaturation/UDG inactivation step (95°C for 5 min), followed by 30 cycles of 95°C for 30 sec and 65°C for 1 min, and a final extension at 65°C for 5 min. Subsequent to PCR amplification, the whole reaction volume (50 µl) was loaded on the sample pad of the WT channel and M channel of LFA strips and the results were analyzed.

The reference DNA samples with different ApoE genotypes, confirmed by direct sequencing (Sangon Biotech Co., Ltd., Shanghai, China), were used to validate the method. The 3′ penultimate or antepenultimate nucleotide (underlined in Table I) was mismatched to enhance specificity of the assay (25). The sensitivity was evaluated by varying the concentrations of DNA samples.

Human whole blood samples (n=305) were obtained from the Shaanxi Provincial People's Hospital (Xi'an, China). All subjects were self-reported to be from the Chinese Han populations and unrelated to each other. Informed written consent was obtained for participation in the present study, which was approved by the Human Subjects Ethical Committee of Northwest University. Genomic DNA was extracted from 200 µl whole blood sample by Whole Blood Genomic DNA Isolation Kit from Xi'an GoldMag Nanobiotech Co., Ltd. (Xi'an, China) according to manufacturer's instructions. All DNA samples were tested to determine the genotypes of ApoE in a double-blind trial. Each sample was analyzed using PGMNs-based LFA strips and DNA sequencing (Sangon Biotech Co., Ltd.) as a comparison.

Statistical analyses were performed using Microsoft Excel version 14.0 (Microsoft Corporation, Redmond, WA, USA) and SPSS software version 16.0 (SPSS, Inc., Chicago, IL, USA). Statistical differences were determined using a Chi-squared test. P<0.05 was considered to indicate a statistically significant difference.

GoldMag nanoparticles were synthesized and characterized as described previously (21). The general scheme to functionalize nanoflower GMNs with CTAB and PSS coating, is represented in Fig. 1A. The addition of cationic surfactant CTAB to the Fe3O4/Au/Fe3O4 particles inhibited the absorption of additional Fe3O4 petals to the Fe3O4/Au core shell structure. CTAB stabilized the core/shell structure by neutralizing layers of surface positive charge to prevent aggregation. However, CTAB-coated particle dispersions are frequently destabilized in salt and buffer solution, resulting in partial aggregation and low recovery yields (26). In the present study, CTAB-capped coated Fe3O4/Au particles were stabilized by wrapping the CTAB layer with PSS (Fig. 1A).

The successful attachment of PSS was demonstrated in FT-IR spectra (Fig. 1B). The PSS spectrum has typical absorption features of neat polystyrene and sulfonate group (SO₃⁻) (27,28), including a primary absorption of the stretching vibrations that correspond to asymmetric stretching of -CH₂- at 2,916 cm⁻¹, C-C stretching of sp² hybridized carbon atoms at 1,597 cm⁻¹, para-disubstituted benzenes (C=C) at 1,438 cm⁻¹, symmetric vibrations of -SO₃ groups in PSS chains at 1,158 cm⁻¹ and stretching of hydroxyl groups of -SO₂-OH over the region 3,700–3,000 cm⁻¹. The spectra for CTAB-coated GMNs exhibited the distinctive absorption peaks centered at 2,920, 2,850, 1,401, and 960 cm⁻¹, which represented symmetric (2,920 and 2,850 cm⁻¹) and asymmetric (1,401 cm⁻¹) stretching of the C-H bond in CTAB and quaternary amine (960 cm⁻¹) stretching of CTAB (19), respectively. For PGMNs, the characteristic absorption bands of PSS were observed, whereas the distinctive peaks of CTAB were not detected. This result demonstrated the presence of the polymer on the particle, indicating that the PSS functionalized GMNs had been successfully prepared.

The stepwise conjugation of functional groups on the GMNs was then monitored by measuring the surface charges at different stages of synthesis. As presented in Fig. 1C, a positive zeta potential (+27.5 mV) in neutral pH (water) of CTAB-capped GMNs, is attributed to the positive charge of the trimethyl ammonium group [-N (CH₃)₃⁺] of CTAB. Following the PSS coating of the nanoparticle surfaces, the zeta potential shifted to negative values (−37.6 mV). This reversal of zeta potential upon the PSS coating indicates the presence of PSS on the nanoparticle surfaces, the PSS exhibits a constant negative charge above pH 2 (18,29). Negative zeta potential formulations help repel each particle in the suspension, ensuring long-term stability and avoiding particle aggregation (30). When PGMNs were added into sodium chloride solution with an ionic strength of 30 mmol/l or 1X PB buffer at pH 7.2, a slight decrease in zeta potential of PGMNs was observed due to lowering of ion screen effects of PSS in the electrolyte solution.

For assay development, PGMNs were conjugated to anti-digoxin antibodies or anti-FITC antibodies to construct GoldMag probes. Bioconjugation of particles with targeted antibodies was confirmed by dynamic light scattering measurement, SPR band analysis and the protein concentration assay. As demonstrated in Fig. 1D, the mean hydrodynamic diameter of particles increased from 68 to 95 nm (PGMNs-anti-digoxin) or 107 nm (PGMNs-anti-FITC) following conjugation. The increased hydrodynamic size following the conjugation suggested that antibodies were effectively coupled to the particles (31). UV-vis data (Fig. 1E) indicated that antibody-labeled PGMNs exhibited a marked red shift compared with the unlabeled ones, from 538 to 547 nm (PGMNs-anti-digoxin) or 552 nm (PGMNs-anti-FITC) where the plasmon resonance band appears (32). This shift is likely to be caused by the surface chemistry change of the nanoparticles from PSS to antibody (33). The optimization of PGMNs-antibody conjugates was defined by the protein concentration assay as described in a previous report (23), and the appropriate amount of anti-digoxin or anti-FITC antibodies immobilized on the PGMNs was about 60 and 48 µg/mg, respectively.

The principle of visual detection of ApoE genotypes using PCR-GoldMag LFA system is illustrated in Fig. 2. The method consists of the following two steps (Fig. 2A): i) Amplification of one sample in two reactions, ApoE 158 and ApoE 112 tubes; and ii) PCR products of each tube are loaded onto the sample pads of two complementary LFA strips following amplification. One strip detects wild-type alleles (termed ‘WT channel’) and the other detects mutant alleles (termed ‘M channel’). The LFA strip is composed of five parts (Fig. 2B): A sample pad, a conjugate pad, a strip of nitrocellulose membrane, an absorbent pad and a plastic backing. Detection of each PCR product is performed using two complementary strips (WT and M channel) with conjugate pads pre-dispensed with PGMNs-anti-digoxin antibody conjugates and PGMNs-anti-FITC antibody conjugates, respectively.

To reduce the labor and cost, the multi-ARMS-PCR analysis was used to amplify wilt-type and mutant target SNP alleles in a single reaction tube, by simultaneously using allele-specific forward primers and a common, biotin-labeled reverse primer (Fig. 2C). Wild-type primers were labeled with digoxin and mutant primers were labeled with FITC. Genotyping of ApoE 158 and ApoE 112 was performed in two separate PCR tubes, with ApoE 158-specific primers to produce 158-specific amplicons (451 bp) and the other with ApoE 112-specific primers to produce 112-specific amplicons (588 bp). The PCR amplicons were synthesized with the 3′-end of the primer complementary to the template. Thus, amplicons from the wild-type DNA were biotin and digoxin labeled, whereas the mutant amplicons were biotin and FITC labeled (Fig. 2C).

The visual detection of the amplicons was performed on the LFA strips. No purification of the PCR products was required prior to detection by LFA. In the test procedure, the PCR products migrate through the membrane strip. PCR target fragments, if labeled with digoxin, form a complex with the pre-fabricated PGMNs-anti-digoxin antibody conjugates on the adjacent conjugate pad of the WT channel strip. Similarly, PCR target fragments labeled with FITC form a complex with the pre-fabricated PGMNs-anti-FITC antibody conjugates of the M channel strip. The subsequent DNA-PGMNs-antibody conjugates migrate across the membrane strip until captured by pre-immobilized streptavidin on the T line forming the positive result of a red band. The red band at the T line of the WT channel strip indicates the presence of wild-type fragments, and the red band at T line of the M channel strip indicates presence of mutant fragments. The excess of PGMNs-antibody conjugates is captured at the C line by goat anti-mouse IgG that suggests the successful performance of the LFA system.

The final genotyping result of a sample for one locus requires visual inspection of the color development on the T lines of both the WT and M channel strips (Fig. 2C). For wild-type sample, a distinct red band is only visible on the T line of the WT channel, whereas the T line of the M channel does not have a color band. By contrast, for homozygous mutant samples, the red band appears exclusively on the M channel strip and not on the WT channel strip. However, when red bands with similar intensities are present on the T lines of WT and M strips, it indicates a heterozygous mutant sample. The detection of different ApoE phenotypes and negative control (no DNA) is illustrated in Fig. 2D.

To obtain the optimal performance, potential variables, including ARMS, LFA preparation and detection conditions, were considered. All primers were designed to avoid cross-hybridization. In order to ensure the specificity of the primers, an additional mismatch at the penultimate or antepenultimate nucleotide of the 3′ terminus of allele specific forward primers was introduced based on principle of ARMS-PCR (24). To increase the specificity and to reduce the PCR running-time, a combination of primer annealing and extension steps in the PCR was performed. By using a two-step PCR, the running-time was greatly reduced (30 cycles within 60 min).

The specificity of ApoE genotyping was an important consideration for the present study. For specificity analysis, six ApoE genotypes (E2/E2, E3/E3, E4/E4, E2/E3, E2/E4, and E3/E4) were analyzed using the LFA system (Fig. 3A). In addition, ApoE genotyping results detected by DNA sequencing were performed as a comparison (Fig. 3B). The GoldMag-based LFA can clearly differentiate one base mutations between wild-type target DNA and mutant target DNA, allowing successfully distinction between the six ApoE genotypes.

Under the optimum conditions, the sensitivity of LFA was evaluated by using various quantities of DNA samples with known ApoE genotypes. Different quantities of target DNA were used for sensitivity analysis (5, 10, 20, 50, 100, 200, 400, 800 and 1,000 ng, and 0 ng as a negative control). Fig. 4 presents the representative images of LFAs using different concentrations of ApoE 158 (Fig. 4A) and ApoE 112 (Fig. 4B) reference DNA templates. No red band was observed on the T line of the LFA in the absence of DNA (control). The signal intensity correspondingly increased as the target DNA concentration increased. The red bands on the T line were observed with ≥10 ng of target DNA, which was considered to be the limit of detection of this assay.

To test the robustness of the proposed method for the ApoE genotypes, 305 clinical samples from a Han Chinese cohort were analyzed using the PCR-GoldMag LFA. In addition to the LFA, direct DNA sequencing was performed. All analyzed samples produced an unambiguous ApoE genotype. Of the 305 samples analyzed in the current study, the allele frequency of E2, E3 and E4 was 6.98, 80.48 and 12.54%, respectively. The frequency of each of the ApoE genotypes is presented in Table II. All genotyping results (100%) were in accordance with the results of direct DNA sequencing. In the subjects (n=305), the overall allele distribution was similar to a previous study (34), with no significant difference in allele distribution (Chi-squared test, P=0.598). These results indicate the reliability of using the PCR-GoldMag LFA method for ApoE genotyping.