Separation of solution flow using a low-cost adhesive tape
The key strategy of using an adhesive tape in the fabrication of the LFIA device was to separate two solutions independently flowing from two different inlet pads. Solution movements on devices without and with the adhesive tape were investigated by applying milli-Q water and green food dye-colored water on the first and second pad, respectively. By using the device without the adhesive tape (Fig. 1a), a controlled solution movement from each inlet towards the NCM alone could not be achieved after applying the water (purple arrow) and the green liquid (orange arrow) on the first (ii) and second (iii) inlet pads, respectively. Uncontrolled flow behavior including counter-flow occurred at the confluence point of the two liquids. After the absorption of all liquids by the absorbent pad (iv–v), the flow of water and green liquid on the NCM appeared non-uniform. On devices incorporating the adhesive tape (Fig. 1b), the water (purple arrow) smoothly flowed from the first inlet pad toward the NCM, followed by the green liquid (orange arrow) from the second inlet pad also moving toward the NCM without interfering with the flow of the water (ii–v).
Device working principle
The principle of the sensitivity improvement of AuNP-based LFIAs via the double-sample inlet device is shown Fig. 1c. Briefly, after loading the sample on the sample pad located under the adhesive tape (i), ferritin molecules present in the sample liquid flow to the NCM (ii) and are captured by the immobilized antibody on the T line (iii). After the sample is also loaded on the conjugate pad (iv), ferritin molecules bind to some of the labeled antibodies located at the end of the conjugate pad, forming the target analyte-labeled antibody complex, while some free labeled antibody flows to the NCM (v). Finally, the free labeled antibody is captured by the ferritin bound immobilized antibody at the T line together with the target analyte-labeled antibody complex, while remaining free labeled antibodies are bound at the C line, leading to the appearance of red color bands on T and C lines (vi). Based on the above procedure, there are two ways for the occurrence of binding interaction between the target analyte and the two antibodies, which might improve the sensitivity of the assay.
Furthermore, the principle of the sensitivity improvement of AuNP-based LFIAs using the gold enhancer (KAuCl4 and HA) is illustrated in Fig. 1d. Before gold enhancement, the target analyte, the labeled and the immobilized capture antibodies can bind into a sandwich form, resulting in creation of red color signals. After the flow of the gold enhancer solution 8, AuNPs are enlarged by the reduction of Au3+ ions to bulk metal in the presence of HA as a reducing agent. In the reduction process, the Au3+ is rapidly deposited on the surface of the AuNP-labeled antibody, to form larger size AuNPs, resulting in the appearance of purple color signals.
Parameter optimization
To achieve the best device efficiency with the use of suitable amounts of reagent and sample fluids as well as device materials for reducing operation costs, various parameters were systematically optimized: the concentration of anti–h ferritin 8803 for conjugation to AuNPs, the volume of the labeled antibody solution used for deposition onto conjugate pads, blocking conditions of the conjugate pad and the NCM, concentrations of anti–h ferritin 8806 and GAM antibody solutions applied on the NCM and finally, the concentrations of gold enhancer solutions 28. The appropriate concentration of anti–h ferritin 8803 for conjugation to AuNPs during preparation of the labeled antibody was selected at 100 µg/mL, evaluated from the lowest concentration providing an unchanged solution color after adding 10% (w/v) NaCl and the highest red intensity (see Supplementary Fig. S1 online).
In the case of colorimetric immunoassays, the amount of the two antibodies binding to the target analyte significantly affects the analytical efficiency 1. There must be sufficient antibody molecules for target binding, in order to achieve a high signal intensity in colorimetric detection that can be clearly and easily evaluated via the naked eye. Various volumes of the above-mentioned optimized labeled antibody solution (4–16 µL) were evaluated by assaying 50 ng/mL of ferritin, a sufficient target concentration level for the simple and clear evaluation of color intensities at both T and C lines, using the double-sample inlet device. Results (Fig. 2a) indicated that 8 µL of the labeled antibody displayed a high visual color intensity equal to the 12 and 16 µL cases. Hence, to keep the device costs minimal, 8 µL were selected as the optimal volume for the labeled antibody applied on the device. In addition, for the improvement of labeled antibody release from the conjugate pad and movement on the NCM, the effects of the blocking solution on the conjugate pad and the NCM were investigated for three different conditions: (1) unblocked, (2) blocking with 3% (w/v) bovine serum albumin (BSA) and (3) blocking with 1x blocking buffer. Applying 50 ng/mL of ferritin samples on the double-sample inlet devices (see Supplementary Fig. S2 online) with BSA-blocked conjugate pad (see Supplementary Fig. S2a online) and unblocked NCM (see Supplementary Fig. S2b online) enabled clear naked eye observation of both T and C lines and were selected as optimal blocking conditions.
Furthermore, concentrations of the immobilized capture antibody (0.5–1.5 mg/mL) and the GAM (0.25–1.25 mg/mL) applied to fabricate the T and C lines were studied to select the smallest concentrations exhibiting good analytical performance. These investigations were performed by applying 20 ng/mL of ferritin samples (an average cut-off level of ferritin in humans) on the double-sample inlet device. The results indicated that 1.0 mg/mL of the immobilized capture antibody (Fig. 2b) and 0.5 mg/mL of GAM (Fig. 2c) are the minimal concentrations showing appropriate color intensities and were therefore chosen as optimal values for the preparation of the T and C lines on the NCM.
To fabricate the gold enhancement device, the concentrations of KAuCl4 (0–70 mM) and HA (0–400 mM) were investigated in assays performed with 20 ng/mL of ferritin samples. Accordingly, 50 mM KAuCl4 (Fig. 2d) and 200 mM HA (Fig. 2e), which were the minimal concentrations exhibiting appropriate color intensity enhancement with low background color interference, were selected as optimal values for gold enhancement. Generally, KAuCl4 and HA had to be separately applied on the device at different positions, because their mixed solutions immediately turned to purple color. For this reason, the HA/PVA/PEO blend film was utilized for applying HA in this study in contrast to directly depositing and drying of HA on the gold pad (see Supplementary Fig. S2c online). Results showed that the use of the HA/PVA/PEO blend film provided a color signal improvement, whereas, by directly depositing HA, KAuCl4 and HA interacted and adsorbed on the gold pad, resulting in purple color development on the gold pad instead of the enhancement of the color intensity at T and C lines on the NCM. For further investigation, the HA/PVA/PEO blend films were soaked in water, and the pH of the solution was measured every 30 s as an indicator of HA release from the blend films compared to the pH of the 200 mM HA solution (pH = 3.01). As the results in Fig. 2f demonstrate, the solution pH decreased with increasing film soaking times from 0–180 s and with increasing HA concentrations used for film preparation, confirming the gradual release of HA from the HA/PVA/PEO blend films. Moreover, the pH of the solution was stable after 180 s with a pH value of 4.34 measured for films prepared from the optimal concentration of 200 mM HA. From these results, it could be confirmed that the use of the HA/PVA/PEO blend film enables a delayed release (180 s) and flow of HA on the device. Therefore, after application of water on gold pads modified with polymer blend films, KAuCl4 reaches the T and C lines to react with AuNPs before the release of HA over 180 s, which could significantly improve the assay sensitivity. All optimal parameters influencing the overall device performance are summarized in Supplementary Table S1 online.
Characterization of HA/PVA/PEO blend film and gold enlargement
To confirm the successful preparation of the polymer blend film, the surface morphology of the HA/PVA/PEO blend film was characterized by SEM and compared to that of a PVA/PEO blend film. The SEM image of the PVA/PEO blend film displayed a smooth surface (Fig. 3a). After the addition of HA (Fig. 3b), a 3D fractal structure was found within the PVA/PEO blend film, confirming the successful immobilization of HA 29. Furthermore, FTIR of the HA/PVA/PEO blend film was also investigated and compared to that of the PVA/PEO blend film (Fig. 3c). There is a strong broad band in the 3200–3400 cm−1 region attributed to O–H and N–H vibrations of the COOH and HNOH groups of the HA/PVA/PEO blend film, respectively. The band at approximately 2885 cm−1 can be assigned to C–H bonds in the PVA/PEO polymer structure. The small band at around 1241 cm–1 is assigned to CH2 symmetric twisting. Strong bands observed at 1086 cm−1 were recognized as crystallization sensitive bands, referring to the C–O stretching vibration. A sharp band at 840 cm–1 is attributed to C–O stretching 30-32. The characteristic peak for amino groups of HA is at approximately 1636 cm−1, confirming the achievement of the immobilization of HA within the PVA/PEO blend film 33.
To characterize the gold enlargement complex by the reduction of Au3+ ions to bulk metal in the presence of HA on the device, the surface morphologies of T lines on the NCM before (the initial AuNPs) and after gold enlargement (gold enhanced AuNPs) were investigated using SEM. The SEM images indicated that the size of AuNPs (Fig. 3e) significantly increased after gold enlargement compared to the initial AuNPs before the enlargement (Fig. 3d). The gold enlargement complex with the average size of 3.09 ± 0.43 µm (n = 20) exhibited a narrow size distribution and uniform particles size 34,35, assuming the high reproducibility of the reaction.
Selectivity and storage stability
The selectivity of the proposed devices was investigated by comparing the results obtained with different potential interferents existing in human serum and blood 36, including 10% (w/v) BSA, 1 µg/mL of alpha–fetoprotein (AFP), c–reactive protein (CRP), creatinine (CR), homocysteine (HCY), myoglobin (MB), human serum albumin (HSA) and human immunoglobulin G (HIgG), with those of 20 ng/mL of ferritin. Results in Fig. 4a indicated that there were no measurable color intensities on the T line after applying solutions of the interferents, verifying a high selectivity toward ferritin due to the highly specific binding interaction between two monoclonal antibodies and the antigen.
In addition, the storage stability of the gold enhancement and double-sample inlet devices was also investigated after keeping these devices at RT in a desiccator for 0–4 weeks and testing with 20 ng/mL of ferritin. As illustrated in Fig. 4b, the observed changes in the Dred intensity value for 20 ng/mL of ferritin remained above 93% of their initial intensities, verifying the stability of the proposed devices.
Analytical performance
The proposed devices were utilized for the measurement of various concentrations (0–500 ng/mL) of ferritin, and the overall assays were completed within 10 minutes without any requirement of external equipment or instruments. The increase of the color intensity that appeared on the T line correlated with the increase of the ferritin concentration, and the visual limit of detection (LOD) was determined as the minimal concentration of ferritin resulting in naked eye observable color intensity on the T line. From the photographs in Fig. 5a, visual LODs achieved with the conventional (see experimental procedure in Supplementary S1 and Fig. S3 online), gold enhancement and double-sample inlet devices were respectively found to be 10, 0.5 and 2 ng/mL, demonstrating that the gold enhancement and double-sample inlet devices provided 20-fold and 5-fold greater sensitivities than the conventional device, respectively. In terms of quantitative analysis (Fig. 5b), ferritin concentration-dependent Dred intensities were examined in triplicate, indicating that the Dred intensities obtained with the gold enhancement and double-sample inlet devices were evidently higher than those of the conventional device. Quantitative calibration curves of the Dred intensity as a function of the logarithm of the ferritin concentration shown in Fig. 5c displayed linearities with good correlations for the gold enhancement device (0.5–500 ng/mL, y = 50.495x+22.847 R2 = 0.992) and the double-sample inlet device (2–500 ng/mL, y = 32.942x–8.984, R2 = 0.996), while the small error bars representing the standard deviations (n = 3) indicated high reproducibility. After reaction completion, the results were founded that there was some solution residue in the conjugate pad area. However, in the calibration curves, small error bars were observed with an acceptable relative standard deviation (RSD) below 3.9 %. Therefore, this solution residue did not affect the accuracy of the detection. The calculated LODs of the gold enhancement and double-sample inlet devices obtained from the calibration curves were respectively found to be 0.03 and 0.05 ng/mL (LOD = 3SD/slope) 8. Compared to other previous reports of ferritin LFIAs as summarized in Supplementary Table S2 online, the proposed device provided improved detection limits and a satisfactory concentration response range.
Furthermore, the proposed device was also compared several previous reports on signal amplification of AuNP-based LFIAs using a gold enhancer as shown in Table 1. The signal amplification achieved in these previous reports was 5–100 folds after gold enhancement, and the overall reaction was complete within 15–20 min, showing that our proposed device gave a shorter analysis time with a high signal amplification value. To achieve 8–100 folds signal amplification 9,34,37,38, the procedure applied in these previous reports required multiple steps of sample loading until the reaction was complete, gold enhancer solution application and washing. In addition, the enhancement process required the use of freshly prepared gold enhancer solutions, leading to more demanding assays and short-term stability. Panraksa et. al. 8 also reported a one-step gold-enhanced AuNP-based LFIA by fabricating a wax-printed patterned LFIA platform that required an expensive wax printer and provided a minimal signal amplification value (5-fold). Furthermore, no device storage study was conducted because of presumably short-term stability. Therefore, the proposed gold enhancement device integrating the adhesive tape and the HA/PVA/ PEO blend film exhibited numerous benefits, including rapid analysis (10 min), long-term stability (at least one month) and simple fabrication and use, which could be suitable and acceptable for on-site measurements.
Table 1
Comparison of the signal amplification of AuNP-based LFIAs using a gold enhancer
Analyte
|
Reactant
|
Enhancement System
|
LOD
|
Linear range
|
Signal amplification (folds)
|
Analysis time (min)
|
Ref.
|
Avian influenza and Newcastle disease virus
|
1% (w/v) HAuCl4/ 10 mM HA
|
Washing and soaking in freshly prepared enhancer
|
2−12
|
21–2−9
|
100
|
15
|
37
|
Escherichia coli O157:H7
|
1% (w/v) HAuCl4/ 10 mM HA
|
Washing and adding freshly prepared enhancer
|
5 × 103 CFU/mL
|
–
|
8
|
20
|
38
|
Salmonella Enteritidis
|
1% (w/v) HAuCl4/ 10 mM HA
|
Adding freshly prepared enhancer to sample pad
|
104 CFU/mL
|
103–108 CFU/mL
|
100
|
20
|
9
|
Ralstonia solanacearum
|
1% (w/v) HAuCl4/ 2 mM HA
|
Applying freshly prepared enhancer on NCM
|
3 × 104 cells/mL
|
–
|
33
|
15
|
34
|
C–reactive protein
|
200 mM KAuCl4/ 750 mM HA
|
Single step operation by wax-printed sequential flow device
|
0.001 µg/mL
|
0.1–5 µg/mL
|
5
|
15
|
8
|
Ferritin
|
50 mM KAuCl4/ 200 mM HA
|
Double inlet device based on adhesive tape and polymer blend film
|
0.03 ng/mL
|
0.5–500 ng/mL
|
20
|
10
|
This work
|
Real sample application
The practical applicability was tested by directly applying human serum samples on the proposed devices without any pretreatment or dilution. Results in Table 2 reveal that the concentration levels of ferritin in the three sera measured with the proposed gold enhancement device are not significantly different from those obtained by the enzyme-linked immunosorbent assay (ELISA) standard method (see experimental procedure in Supplementary S2 online). Test values from three-independent measurements were found to be 98–101% accurate with relative standard deviation (RSD) below 2.5%. These results verified that the proposed device provided high accuracy and could be utilized for early screening and measuring the amount of ferritin in biological samples.
Table 2
Results for using the proposed device for colorimetric immunoassay of ferritin in human serum samples (n = 3)
Sample
|
Ferritin level (ng/mL)
|
Accuracy
(%)
|
RSD
(%)
|
Proposed device
|
ELISA
|
Serum 1
|
16.14 ± 0.2
|
16.53
|
98
|
1.3
|
Serum 2
|
62.59 ± 0.9
|
63.87
|
98
|
1.5
|
Serum 3
|
23.37 ± 0.6
|
23.13
|
101
|
2.5
|