NT-proBNP detection with a one-step magnetic lateral flow channel assay

Fabrication of lateral flow channel immunoassay and drying of reactive reagents

The LFA consists of a three-layer built-up: (1) the support foil (Melinex® 329 foil, 175 µm (Dupont Teijin Films, www.dupontteijinfilms.com)), (2) the spacer foil (Melinex®329 foil, 250 µm, with double-sided adhesive tape (Henkel Adhesives, www.henkel-adhesives.com) on both sides), and (3) the cover foil (Hostaphan RN 100, 125 µm (Mitsubishi Polyester Film, www.m-petfilm.de)) (see Supplementary Information (ESM) Fig. S1). The capillary (dimension, 90 mm × 1.5 mm × 0.28 mm) was cut in the spacer foil, and the inlet and outlet (r = 0.5 mm) were cut in the cover foil with the laser MicroLine 6000 P (www.lpkf.com) at a frequency of 30 kHz and 3W.

To prepare the ready-to-go, one-step immunoassay, the reactive reagent capture Ab and detection Ab-fluorescence NP were mixed with the dispensing buffer to give a 2.5 µg∙mL⁻¹ and 2% (w/v) solution, respectively, and a sucrose concentration of 6% (w/v). For the MNPs, the dispensing buffer was adjusted through replacing sucrose by additives such as trehalose, dextran, Tween 20, Tween 80, and polyethylene glycol. The drying and the additives’ content were optimized with respect to prevention of aggregation and high assay signal. The MNPs were mixed with the adjusted dispensing buffer to give a 2% (w/v) MNP solution. Capture Abs, Ab-fluorescent NPs, and MNPs were applied separately (2 µL each) onto the support of the immunoassay and dried in a drying cabinet at 50 °C for 10 min (FED 400 E2, www.binder-world.com). The support, spacer, and cover foil were then bond together as depicted in the ESM (Fig. S1).

Analysis of the capturing efficiency of the applied magnetic field

To investigate the capture efficiency of the small and large MNPs by the magnet, 200 µl of MNPs 2% (w/v) in HEPES buffer was transported in different flow rates to the magnet in the detection zone of the lateral flow channel. The magnet is placed at a distance of 175 µm from the bottom wall of the channel and at a distance of 6.5 cm from the application zone to mitigate the impact of initial forces that may arise upon applying the sample to the sensor. After capturing the MNPs, the downstream sample was collected, and its transmittance was analyzed in the photometer.

Analysis of the biotin-binding capacity of the MNPs

A biotin-5-fluorescein working solution (1 µmol∙L⁻¹) was prepared in HEPES buffer. Five hundred microliter of the biotin-5-fluorescein working solution was mixed with 50 µl of MNPs (0.3 µm, 2% (w/v)) and 50 µl of MNPs (3 µm, 2% (w/v)) and incubated in the dark at room temperature for 15 min. Afterwards, the MNPs were pelleted by centrifugation (5000 × g for 10 min) and the fluorescence of the supernatant was measured (λ_ex = 490 nm, λ_em = 524 nm). The biotin-binding capacity (pmol biotin/mg MNP) was then calculated from the difference of the fluorescence intensity in the presence and absence of the MNPs.

Sample preparation

Human blood samples were drawn by venipuncture from volunteers and provided by Roche Diagnostics GmbH in compliance with safety and ethical regulations. The hematocrit was obtained by micro-hematocrit capillary tubes. The blood samples were collected in EDTA-K3 vacutainer tubes and were used within a day.

NT-proBNP assay

A sandwich immunoassay with spiked samples was conducted to investigate the assay performance of the bioassay. For analysis, NT-proBNP biomarker samples were prepared in HBS buffer, human plasma, and human whole blood through dilution of a stock solution to produce concentrations of 7.5, 15, 30, 60, 125, 250, 500, 1000, 2000, 4000, 6000, and 9000 pg∙mL⁻¹. A volume of 15 µL of the spiked sample was applied on the sample application area. The sample was transported with a flow rate of 2 µL/min by the external vent control. The immunoassay was performed at room temperature, and fluorescence signals could be obtained after 10 min. For the calculation of the limit of detection (LOD), the logistic fit parameter for the lower curve asymptote A and the standard deviation of the blank SD (blank) was used:

The concentration of NT-proBNP in the sample is determined by correlating the fluorescence intensity to a calibration curve constructed using known concentration standards.

The functionality of the dried reagents was assessed by replicating the bioassay over a 8-week period. To conduct these measurements, the sucrose-based dispensing buffer was utilized to dry the capture Ab and Ab-fluorescent NPs, and the dextran-based dispensing buffer was utilized to dry the MNPs for 10 min at 50 °C. The one-step LFA was stored at 4 °C in an airtight capsule.

Biotin-binding capacity

We examined the impact of different sizes and surface areas (sa) of MNPs on their ability to improve diffusion-limited interactions in polymer channel lateral flow assays. First, we compared the biotin-binding capacity for biotin-5-fluorescein between small MNPs (d = 0.3 µm, sa = 4.57 × 10⁻⁶ m²∙mg⁻¹) and large MNPs (d = 3 µm, sa = 3.81 × 10⁻⁸ m²∙mg⁻¹) (ESM, Fig. S2, dark blue). Interestingly, both MNPs provided similar biotin-binding capacity of 2600 µmol∙g MNP⁻¹ and 2400 µmol∙g MNP⁻¹, respectively. Despite smaller MNPs having a larger surface area, they may exhibit a lower streptavidin coverage density compared to their larger counterparts, suggesting that, in this case, the size of the MNPs does not affect the binding capacity. In addition, to account for the factor of steric hindrance of antibodies, the fluorescence signal of the immunoassay in the LFA was investigated (ESM, Fig. S2, light blue). Similar fluorescence signals were obtained for both MNPs for a given analyte concentration, using the same MNP concentration. These findings suggest that interestingly, the size of the particles and surface area do not significantly affect factors such as biotin-binding capacity and steric hindrance during the immunoreaction.

Capture efficiency by external magnet

The primary goals of the magnetophoretic capture process are to decrease the count of untrapped particles and to ensure that the flow rate is adequate for achieving optimal capture efficiency and collection quality during magnetic separation. The capture efficiency of the MNPs was determined by measuring the transmittance of the supernatant after the capture process in the lateral flow channel, taking into account the particle size and flow rate (Fig. 2a). A 100% transmittance of the supernatant is defined as the state where all MNPs have been captured by the magnet, resulting in a maximum capture efficiency and a MNP-free supernatant. Conversely, if MNPs are not adequately captured by the magnet, the supernatant will contain MNPs, leading to a decrease in transmittance. As expected, for both MNPs, the capture efficiency decreased with increasing flow rate. Moreover, at the lowest flow rate, both MNP samples demonstrated full capturing. Interestingly, at higher flow rates, differences between nanoparticles and microparticles were observed. The transmittance of the small MNPs strongly decreased with increasing flow rate, while the transmittance of the large MNPs decreased only slightly. The optimal flow rate for the 0.3 µm particles (1 µl∙ml⁻¹) was generally lower than that for the 3 µm magnetic particles (3 µl∙ml⁻¹). The larger particles with greater susceptibility experienced a greater magnetic attraction and therefore endured higher flow velocities. This correlates to earlier findings indicating a dependency of particle size or magnetic core size of coated particles on flow behavior under flow conditions [35, 37]. As a result, the selection of larger MNPs for subsequent experiments was driven by the potential to apply a wider range of flow rates in the assay. Furthermore, we enhanced the slightly lower specific surface area of these larger MNPs and their increased steric hindrance by augmenting the particle count.

Open in a separate window

Fig. 2

Optimization of the capture efficiency by the external magnetic field in the lateral flow channel in dependency of the particle size and flow rate. Plot of the transmittance of the sample against the flow rate and particle size (a). Two hundred microliter of MNP (2% (w/v)) in HEPES buffer was transported with different flow rates through the channel, and the transmittance of the sample was then analyzed in the photometer. Plot of the assay signal against the rehydration time of the dried reagents (b). Control c represents the assay signal with Ab-fluorescent NPs, capture Ab, and MNPs in solution. Standard deviations were calculated based on three parallel measurements on three different lateral flow channels (n = 3)

Development of a dry-reagent lateral flow channel immunoassay

The biotinylated polyclonal capture antibodies and monoclonal probe antibodies employed in this study were identical to the antibodies of Roche’s fifth-generation hs-BNP assay in the Elecsys®/cobas e™ platform. Hence, the antibodies’ selectivity and specificity were presumed to be adequate without conducting additional tests. The initial experiments aimed to assess the viability of drying probe antibodies and probe antibodies labeled fluorescent NPs and MNPs. Since Lutz et al. demonstrated that the functionality of the antibodies could be effectively preserved after the drying and rehydration process [38], here only the effect of drying on MNPs was investigated. Nanoparticles are known to easily adsorb or aggregate irreversibly due to the reduced spacing and the compromised electrostatic repulsion between the nanoparticles during the drying process. This impairs the biological functionality of the particles [39–41]. Thus, to prevent aggregation and maintain the long-term functionality, various additives such as polyethylene glycol (PEG), surfactants such as Tween 20 and Tween 80, and various sugars were included in the dispensing matrix (ESM, Fig. S3). After drying 2 µL of MNPs (2% (w/v)) in the dispensing buffer on the PET support and rehydration with 15 µL of NT-proBNP-spiked HEPES (c_[analyte] = 1 ng∙mL⁻¹), we observed the formation of aggregates and significant irreversible adsorption of MNPs onto the PET support with all additives except of dextran. Drying the MNPs in only HEPES buffer (blank) diminished the signal to only 5%. In all other drying buffers, the fluorescence signal dropped to only 55% of the original signal. The drying of MNPs in the dextran-based matrix resulted in the most favorable fluorescence signals, indicated by a signal drop to only 92% in comparison to the control. This observation suggests that dextran successfully prevented the irreversible adsorption of MNPs on the support and minimized aggregation, indicating its efficacy in maintaining the integrity of the MNPs during the drying process. Subsequently, the rehydration time of the dried reagents was assessed by exposing them to different durations of rehydration in the channel. Here, we determined the optimal condition with respect to the reagents effective functionality and fluorescence signal in the immunoassay (Fig. 2b). The most favorable signal was obtained by rehydrating the dried reagents with 15 µL NT-proBNP-spiked HEPES buffer (c_[analyte] = 1 ng∙mL⁻¹) for 3 min. In fact, the fluorescence signal was minimally affected by drying when compared to the control, which represents Ab-fluorescent NPs, capture Abs, and MNPs in solution. This suggests that the rehydrated probe antibodies retain their biological function and that the Ab-fluorescent NPs do not form aggregates that would hinder their fluorescence quantification.

Optimization of analyte–capture antibody interactions

It is known that heterogeneous immunoassays can have lengthy detection times due to the duration required for analytes to reach surface-bound receptors [42–44]. Harnessing the advantage of MNPs, these were spotted in the beginning of the channel, hence allowing analyte–antibody interactions throughout the flow through the channel. This was compared to MPNs spotted in the detection zone to demonstrate the advantage of prolonged incubation and diffusional access between capture antibody and analyte (ESM, Fig. S4). The concentration of capture Abs, Ab-fluorescent NPs, and MNPs strongly influenced the sensitivity of the immunoassay. For a 15 µl NT-proBNP-spiked HEPES buffer (1 ng∙ml⁻¹), we found optimal concentrations of 2.5 µg∙ml⁻¹, 2% (w/v) and 3 µg, respectively (Fig. 3a–c). It is worth noting that as the concentration of MNPs increases, the fluorescence intensity also increases. This is due to the increased density of streptavidin recognition sites for the specific binding of the biotinylated sandwich complex, which leads to an improvement in the detection sensitivity. To further optimize the diffusional transport to the reactive sites, the duration for which the MNPs are in contact with the sample was modified by changing the flow rate as this influences capture efficiency, analyte–antibody interaction, and shear forces [45]. Thus, the fluorescence signal after the immunoreaction for a given analyte concentration at different flow rates was investigated (Fig. 3d). First, 15 µl of NT-proBNP-spiked HEPES buffer (1 ng∙ml⁻¹) was applied on the ready-to-go lateral flow channel, which subsequently initiated the rehydration of the dried reagents. The sample was transported with flow rates of 4, 3, 2, and 1 µL∙min⁻¹, and the fluorescence intensity was measured at the detection zone. It was found that lower flow rates resulted in overall more favorable conditions combining superior capture efficiency and increased interaction time. Furthermore, the rehydration ability of the dried reagents such as capture Abs, Ab-fluorescent NPs, and MNPs was visually observed to be better at lower flow rates due to the longer rehydration times. In general, the highest fluorescence signal was obtained with the optimal flow rate of 2 µl∙min⁻¹.

Open in a separate window

Fig. 3

Optimization of the immunoassay by varying the concentration of the dried reagents and the flow rate. Plot of the assay signal against the capture antibody (cAb) concentration (a), plot of the assay signal against the Ab-fluorescent NP (dAb) concentration (b), plot of the assay signal against the MNP concentration (c), and plot of the assay signal against the flow rate (d). Error bars represent mean values ± 1 σ (n = 5)

Performance of the lateral flow channel assay

Finally, the previously optimized conditions were transferred to generate dose–response curves for NT-proBNP in buffer and undiluted human blood. The spiked buffer measurements yielded a LOD of 26.7 pg∙mL⁻¹ with an error of only 11% (n = 4) (data not shown). In spiked human blood, a LOD of 43.1 pg∙mL⁻¹ and a mean standard deviation of 18% (n = 4) were obtained (Fig. 4). As anticipated, the sensitivity and standard deviation were slightly affected due to the partial blocking of MNPs by adsorbed serum proteins and an increased background signal originating from these proteins. Notably, the mean error of the proposed assay is expected to benefit from enhanced fabrication processes when specifically tailored for large-scale production. In contrast to the established POCT assays presently employed in clinical environments, like the commercially accessible NT-proBNP tests developed by Roche Diagnostics [46], one of the key benefits of our assay is its ability to deliver similar results (detection range, 60–9000 pg∙mL⁻¹) using a minimal sample volume of just 15 µL. This is a remarkable improvement compared to the Roche Cobas h 232 that requires up to 10 times more sample volume (15 µL vs. 150 µL). While experiments revealed that the limit of detection decreases with higher volume input, the primary objective was to showcase the viability of using small volumes. Specifically, our sensor can detect 0.076 fmol of NT-proBNP, whereas the Roche Cobas h 232 detects 1.06 fmol. To facilitate accurate addition of 15 µL by the patient, the implementation of capillary tubes, a conventional method for precise measurement of sample volumes, is also envisioned for the diagnostic test. This advancement not only preserves high sample volumes but also simplifies the sampling process, making it more convenient for both medical assistants and patients. Additionally, the utilization of a bead-based immunoassay in this lateral flow channel overcomes the limitations associated with diffusion in traditional microfluidic sensors. By leveraging the unique properties of magnetic nanoparticles, the assay enables efficient and targeted analyte capture and enhancing sensitivity. Recently, a newly commercially available MNP-based NT-proBNP assay has been demonstrated that is technologically the closest to our lateral flow channel assay [47]. This microfluidic chip enables NT-proBNP detection from 20 µL blood within 12 min with a range of 50–9000 pg∙mL⁻¹. The assay demonstrated an average error of 4.1% within the NT-proBNP range of 59–4559 pg∙mL⁻¹, indicating its superior analytical precision. However, the testing protocol relies on additional washing steps, which in turn raises the complexity and expenses associated with the assay. Another similar MNP-based system is the well-established Cobas Elecsys® for the ultra-sensitive detection of NT-proBNP [48]. The platform provides a superior detection range of 5–35,000 pg∙mL⁻¹ with only 15 µL sample and an assay time of 9–18 min. Nevertheless, this system operates within a laboratory setting, being costly, necessitating sample preparation, and unsuitable for point-of-care implementation. The materials utilized in LFAs, such as fleece fiber and membrane materials, often exhibit variations and inconsistencies in fabrication process. These variations can impact factors such as sample flow rate and the release of dried reagents, ultimately affecting sensor performance. Thus, the incorporation of a compact pump system for controlled and automated fluid transport in this assay provides a crucial advantage in terms of precise sample transportation. Despite the need for an additional handheld device, this controlled flow system ensures consistent and uniform sample movement, reducing the impact of sample variability and improving the reliability of results. This feature is particularly important in applications where accurate and reproducible measurements are paramount.

Open in a separate window

Fig. 4

Dose–response curve of NT-proBNP concentration in undiluted whole blood with logistic fit (blue line), 95% confidence interval (shaded blue curve) (a), and fluorescence images representing NT-proBNP concentration (b). Standard deviations were calculated based on three parallel measurements using four flow channels, while outliers were removed after Q-test (confidence interval 95%). Error bars represent mean values ± 1 σ (n = 3)

We thank Berkin Yanarsoenmez (Roche Diagnostics, Mannheim) for providing the fluid control system.