Vanessa Pilati*ab,
María Salvador
ac,
Leyre Bei Frailea,
José Luis Marqués-Fernándeza,
Franciscarlos Gomes da Silva
b,
Mona Fadel
a,
Ricardo López Antónd,
María del Puerto Morales
c,
José Carlos Martinez-Garcíaae and
Montserrat Rivas
ae
aDepartamento de Física, Campus de Viesques, Universidad de Oviedo, Gijón, 33204, Spain. E-mail: pilativanessa@uniovi.es
bComplex Fluids Group, Instituto de Física & Faculdade UnB – Planaltina, Universidade de Brasília, Brasília, 70910-900, Brazil
cDepartamento de Nanociencia y Nanotecnología, Instituto de Ciencia de Materiales de Madrid (ICMM), Madrid, 28049, Spain
dInstituto Regional de Investigación Científica Aplicada (IRICA) and Departamento de Física Aplicada, Universidad de Castilla-La Mancha, Ciudad Real, Spain
eInstituto Universitario de Tecnología Industrial de Asturias (IUTA), Gijón, 33203, Spain
First published on 8th July 2024
Lateral flow assays are low-cost point-of-care devices that are stable, easy to use, and provide quick results. They are mostly used as qualitative screening tests to detect biomarkers for several diseases. Quantification of the biomarkers is sometimes desirable but challenging to achieve. Magnetic nanoparticles can be used as tags, providing both visual and magnetic signals that can be detected and quantified by magnetic sensors. In the present work, we synthesized superparamagnetic MnFe2O4 nanoparticles using the hydrothermal coprecipitation route. MnFe2O4 presents low magnetic anisotropy and high saturation magnetization, resulting in larger initial magnetic susceptibility, which is crucial for optimizing the signal in inductive sensors. We functionalized the coprecipitated nanoparticles with citric acid to achieve colloidal stability in a neutral pH and to provide carboxyl groups to their surface to bioconjugate with biomolecules, such as proteins and antibodies. The nanomaterials were characterized by several techniques, and we correlated their magnetic properties with their sensitivity and resolution for magnetic detection in radiofrequency inductive sensors. We considered the NeutrAvidin/biotin model of biorecognition to explore their potential as magnetic labels in lateral flow assays. Our results show that MnFe2O4 nanoparticles are more sensitive to inductive detection than magnetite nanoparticles, the most used nanotags in magnetic lateral flow assays. These nanoparticles present high potential as magnetic tags for the development of sensitive lateral flow immunoassays for detecting and quantifying biomarkers.
Classical LFAs combine microfluidics and biorecognition strategies. In brief, a fluid to analyse, which can be a body fluid (blood, saliva, urine, serum, plasma), an environment sample such as wastewater, or a food sample, is deposited into a sample pad. The fluid flows along a nitrocellulose membrane strip by capillary action. The nitrocellulose has a test line containing specific biorecognition molecules that can capture the analyte, which is tagged by a nanolabel to be detected. If an analyte is in the fluid sample, it is trapped at the test line together with the nanolabels. Commercial LFAs are mainly based on gold nanoparticles that provide a reddish colour signal in the test line, which can be interpreted as a positive or negative result.
Quantifying the molecule of interest in the sample is often necessary for the early diagnosis of some illnesses and detecting contaminants and toxins. However, it remains a significant challenge in LFAs that use visible labels. Calibration difficulties arise from ambient lighting variations, humidity, and differences in paper colours. Importantly, in optical sensors, the signal arises mostly from nanolabels in the upper section of the test line, while a significant fraction of them within the paper remains undetected.4
Magnetic nanoparticles (NPs) may be used as labels to quantify the bio-analytes in the test line. Besides the visual signal, they can be detected by magnetic sensors, which are sensitive to all the NPs through the thickness of the membrane. In addition, manipulating magnetic NPs enables analyte separation in a complex sample and its pre-concentration to further increase the sensitivity.5–7 Magnetic tags have been used to quantify Bacillus anthracis spores,8Streptococcus pneumoniae biomarkers,7 cancer biomarkers,9 drug abuse,10 and histamine in wine,11 among others. Most magnetic LFAs are based on superparamagnetic iron oxide, usually magnetite (Fe3O4) nanoparticles and their assemblies.12
Several magnetic sensors with different physical principles have been explored to quantify the magnetic nanoparticles in the test line and, therefore, the target biomolecule. They can be mainly classified as magnetoresistive readers or inductive readers.13,14 The most common magnetoresistive sensors for reading LFAs are based on giant magnetoresistance (GMR)15–17 and tunnel magnetoresistance (TMR).18,19 Magneto-inductive sensors have also been used to achieve quantification in LFAs. A superconducting quantum interference device (SQUID) magnetometer is the most sensitive option.14 However, it demands expensive maintenance and fabrication, hindering its use as a point-of-care sensor. Atomic magnetometers have also been used to achieve high sensitivity in LFAs quantification.20 Other inductive strategies based on magnetic flux,21 magnetic permeability14,22–24 and non-linear magnetization25 have been developed to be low-cost and easily miniaturized while achieving high sensitivities for the detection of several biomarkers.14
In the case of inductive sensors based on magnetic permeability, a copper coil applies an alternating magnetic field to adjacent superparamagnetic particles. The particles increase the coil's self-inductance and resistance, analogous to the effect of a ferrite core on an inductor, and this change in impedance is measured with an impedance analyser.24,26 The magnetic signal can be optimized by increasing the magnetic permeability of the magnetic material μ = μ0(1 + χ) = B/H, being μ0 the vacuum magnetic permeability, χ the magnetic susceptibility, B the magnetic induction, and H the applied magnetic field. This can be achieved by increasing the size of the magnetic material while keeping its superparamagnetic character27 and also reducing its magnetic anisotropy.
For this work, we chose to investigate the radiofrequency inductive response of manganese-ferrite MnFe2O4 NPs to be used as magnetic nanolabels in magnetic lateral flow assays instead of the most common magnetite NPs. Bulk Mn-ferrites are soft magnetic materials with a very low magnetocrystalline anisotropy constant28 (K = 0.3 × 104 J m−3) compared to that of magnetite (1.1 × 104 J m−3). Bulk Mn-ferrite has a high specific saturation magnetization at ambient temperature (80 A m2 kg−1), close to that of magnetite (90 A m2 kg−1).28 Therefore, MnFe2O4 NPs are expected to present a high initial magnetic susceptibility, which is important for detection in inductive sensors. In addition, colloidal dispersions of Mn-ferrite NPs are dark brown/black, while those based on iron oxides like magnetite and maghemite NPs are usually brown/orange, and those based on gold NPs are red. Therefore, it is expected that Mn-ferrite can provide better optical contrast between the test line and the paper background compared to pure iron-oxide and gold NPs.
Several techniques can be used to synthesize Mn-ferrites nanoparticles, mostly based on “bottom-up” chemical approaches. The most common methods are coprecipitation,29,30 thermal decomposition,31,32 solvothermal,33 and hydrothermal methods.34 In this paper, we synthesized Mn-ferrite NPs using a hydrothermal coprecipitation route. It consists of the simultaneous precipitation of Fe3+ and Mn2+ in alkaline media under constant heating and stirring. Such a technique has a high yield at low cost and complexity. It does not require the use of highly toxic solvents or surfactants. Both aspects make hydrothermal coprecipitation a promising technique for industrial and commercial purposes. However, the control over size and shape is generally limited. Several parameters can potentially modify the final coprecipitated NPs. These include the nature and concentration of the metallic precursors35 and of the alkaline base,29,30 the reaction temperature,36 the stirring rate37 and the experimental setup. The control of these parameters is crucial for optimizing NP properties and ensuring reproducibility.
The coprecipitated NPs of the present work were functionalized with citric acid to achieve colloidal stability at neutral pH and provide carboxyl groups at the NP surface for their bioconjugation. We investigated the detection performance of our NPs using a radiofrequency inductive sensor and compared their magnetic signal with their magnetic properties. In addition, the NPs with higher sensitivity were bioconjugated with NeutrAvidin to test their performance as magnetic labels in lateral flow assays composed of a biotin-bovine-serum-albumin (biotin-BSA) test line in a well-known biochemistry NeutrAvidin-biotin model of biorecognition.
Lateral flow strips were prepared using a nitrocellulose membrane (UniSart CN95) from Sartorius Stedim Biotech (Spain), an absorbent pad (Whatman CF5) from GE Healthcare Life Sciences (UK), a backing card (HF000MC100) and a glass fiber sample pad (GFCP001000) from Millipore (USA). All components of the lateral flow strips were used without further modifications.
We checked the crystalline structure of the synthesized NPs by X-ray powder diffraction (XRPD) after air-dried evaporation of the liquid phase of the ferrofluid samples (80 °C). Measurements were conducted with a Seifert XRD 3000 T/T diffractometer (Rich. Seifert & Co., Ahrensburg, Germany) using a Mo emitter (Kα: λ1 = 0.0709316 nm and λ2 = 0.0713607 nm) and with 2θ ranging from 7° to 57°. The intrinsic width was measured using a LaB6 bulk sample (SRM 660 NIST). We conducted thermogravimetric analysis (TGA) to investigate the citric acid covering the NP surface. Measurements were taken in a Seiko (TG/DTA EXSTAR 6000 thermobalance). Samples were heated from 25 °C to 900 °C at 5 °C min−1 in airflow.
The size distribution of the NPs was investigated by transmission electron microscopy (TEM) measurements using a 200 keV JEOL-2000FXII microscope. A diluted and sonicated ferrofluid drop was deposited on a carbon-coated copper grid. We obtained size histograms by measuring approximately 1000 nanoparticles per sample using ImageJ software. The hydrodynamic diameter of the NPs was investigated by dynamic light scattering (DLS) measurements using a Zetasizer instrument (Malvern). Hydrodynamic volume and polydispersity index were estimated using a log-normal fitting of the DLS number distribution data.
ΔZ(f,χ′,χ′′,ψ) = fLψχ′′(f) + ifLψχ′(f), | (1) |
We investigated the inductive response of the synthesized NPs (before bioconjugation to NeutrAvidin) in the sensor by printing ferrofluids with different concentrations directly in the nitrocellulose (25 mm wide) with the backing card. Printing was made using an automatic micropipette coupled to a micropositioner, programmed to dispense the ferrofluids across the membrane at a 0.1 μL mm−1 rate. After drying, the nitrocellulose containing the deposited NPs was cut into 25 × 7 mm strips with a guillotine. We carried out five different depositions for each sample, with different ferrofluid concentrations, leading to different mass values of the deposited NPs. Our samples were measured using the same inductor and frequency (20 MHz) as in Salvador et al.,27 to compare their performance with magnetite NPs.
We define the relative magnetic signal S as the relative increase of the impedance due to the contribution of all the deposited NPs:
![]() | (2) |
![]() | ||
Fig. 1 (A) Scheme of NeutrAvidin bioconjugation with the NPs covered by free carboxyl groups. (B) Picture of a dipstick lateral flow strip printed with a biotin-BSA as a biorecognition element. The NPs@NeutrAvidin conjugate flows through the membrane and is retained in the biotin test line due to the strong NeutrAvidin/biotin affinity. The retained conjugates in the test line are seen thanks to the dark-brown color of the magnetic labels. (C) Scheme of the magnetic detection of the NPs in the nitrocellulose membrane by the radiofrequency inductive sensor (left). At right, typical curves from the test line of a magnetic LFA using the MnFe2O4 nanoparticles of this work (dark-brown line) and Fe3O4 nanoparticles covered by a double layer of oleic acid46 (orange test line). |
For the bioconjugation of the NPs with NeutrAvidin, first, 5 μL of ferrofluid (6.7 mg mL−1), previously sonicated for 5 minutes, was mixed with 10 μL of EDC (1000 μg mL−1) freshly prepared in a phosphate-buffered saline solution (PBS 1 mM, pH 7.4). The mixture was placed in a vortex stirrer for 1 min. We added 100 μL of a NeutrAvidin solution (1000 μg mL−1 in PBS 1 mM pH 7.4) to the mixture and put it in a refrigerated ultrasonic bath for different times: 1 min, 5 min, 10 min, 20 min, and 40 min. We repeated the same protocol without the EDC solution. We investigated different NeutrAvidin concentrations, ranging between 0–1000 μg mL−1.
For this, we chose the optimized reaction time, i.e., the bioconjugation protocol that led to the most intense test line.
For the LFAs, we mixed 20 μL of freshly prepared NPs@NeutrAvidin conjugate with 80 μL of a freshly prepared running buffer (RB) solution (10 mg mL−1 BSA, 0.5% Tween20 in PBS 10 mM, pH 7.4). The assays were carried out by vertically introducing the sample pad into an Eppendorf tube containing the RB and the NPs@NeutrAvidin conjugate (dipstick format), as illustrated in Fig. 1B. The magnetic nanolabels retained in the test line were detected and quantified by a radiofrequency inductive sensor, and the magnetic signal was analyzed to investigate the optimization of bioconjugation parameters. In addition, for the different NeutrAvidin concentrations, we measured the optical signal of the test line using a mobile telephone camera. We analyzed the gray value intensity profile using ImageJ software.7 We compared the optical and magnetic signals of our labels. All assays were performed in triplicates.
The TG/DTA measurements shown in Fig. 2A investigated the citric acid coating. From TG curves, a slight loss in mass occurs at low temperatures (until 100 °C), with an endothermic peak in DTA related to the desorption of absorbed water onto the NP surface. This is followed by a significant mass loss above 160 °C, with an exothermic peak at ∼230 °C. This is expected for the decomposition and removal of the citric acid molecules chemically attached to the NP surface.50 We obtained a higher mass loss for the S2 sample in this temperature range, likely due to a larger amount of citric acid. In addition, a secondary weight loss with an exothermic peak occurs at ∼300 °C, mainly for S2 sample, which would be related to a double layer of citric acid, one stronger bonded to NPs than the second one, as reported for citric acid coated magnetite NPs.51 Such results confirm the success of citric acid coating and the availability of carboxyl –COOH reactive groups for bioconjugation. Near pH 7, –COOH undergoes deprotonation, leading to carboxylate –COO− anions at the NP surface.51 This provides additional electrostatic repulsion among the NPs, essential for colloidal stability. Following the more significant mass decrease in both samples until ∼300 °C, two minor bumps with mass growth appear in TG curves. Given that the experiments were conducted in an airflow atmosphere, these measurements are probably associated with the oxidation process of Mn2+ to Mn3+ and Mn3+ to Mn4+ at higher temperatures. This seems more pronounced for the S1 sample, which has a smaller citric acid content (7.5%) than S2 (11.6%) and a higher surface area due to its smaller mean diameter, as it will be shown later by TEM measurements (8 nm vs. 10 nm).
![]() | ||
Fig. 2 (A) TG/DTA measurements of the two synthesized samples. (B) XRPD diffractogram of the two samples. |
Sample | Stirring rate (rpm) | D XRPD (nm) | <a> (nm) | D 0 TEM (nm) | σ TEM | D 0 DLS (nm) | σ DLS |
---|---|---|---|---|---|---|---|
S1 | 360 | 7.7 | 0.849 | 8 | 0.34 | 27 | 0.26 |
S2 | 170 | 9.4 | 0.849 | 10 | 0.25 | 29 | 0.20 |
We have determined the lattice parameter a from Bragg's lawnλ = 2dhklsinθ, with dhkl being the interplanar distances of the {hkl} family planes (h,
k and l being the Miller indices) and n the order of diffraction. We also considered the cubic symmetry of the spinel structure, where dhkl = a/(h2 + k2 + l2)1/2. We determined a for all peaks and calculated the average of them. Results are also shown in Table 1; for the two samples, the values are very close to those of Mn-bulk ferrite28 (0.850 nm). We estimated the density of the NPs considering the volume occupied by the cubic cell (a3) and the atomic mass inside it.48 Hence, we obtained 5008 kg m−3 for the two samples, fairly close to that of bulk ferrite28 (4990 kg m−3).
Fig. 3A and B are TEM micrographs of the two samples. The NPs are quasi-spherical in both samples. The size-distribution histogram (Fig. 3E) fits well to a log-normal probability function – the best fitting yields the values in Table 1 for the characteristic size and polydispersity. The sample synthesized at a low stirring rate (S2) has the largest particles and a narrower size distribution.
We estimated the two samples' hydrodynamic diameter and polydispersity index by DLS measurements (Fig. 3F and Table 1). The resulting values are 27 and 29 nm with PDIs around 25%, which are slightly larger than the TEM sizes, indicating some NP clustering (∼2–5 NPs per cluster). Indeed, DLS results are compatible with the small NP assemblies in the TEM micrographs of significantly diluted and sonicated samples (Fig. 3C and D). Aggregation is undesirable for many NP bio-applications because it can compromise the performance of magnetic nanoparticles (for magnetic imaging or drug delivery, for example) and even lead to significant challenges in in vivo applications (vascular risks associated with clot formation). However, for in vitro magnetic detection in LFA, it can be beneficial. Some agglomeration (as long as the agglomerates can flow through the membrane pores) can enhance the detection sensitivity for quantification because it increases the number of NPs and magnetic signal per biomolecule.7,46
Sample | M (300 K)S (A m2 kg−1) | M (5 K)S (A m2 kg−1) | H (5 K)C (kA m−1) | <TB> (K) | K ZFC/FCeff (104 J m−3) | K LAS (300 K)eff (104 J m−3) | χ′ | χ′′ | |χ| |
---|---|---|---|---|---|---|---|---|---|
S1 | 76.3 | 115.2 | 11.6 | 29 | 3.8 | 1.4 | 12.8 | 4.1 | 13.4 |
S2 | 78.2 | 115.3 | 9.8 | 30 | 1.8 | 1.5 | 13.8 | 3.8 | 14.3 |
In nanoscale, a surface of disordered spins induced by the reduced coordination at the nanoparticle surface increases the low-temperature saturation magnetization,57 therefore reducing the expected squareness MR/MS at low temperatures. However, this surface effect is more pronounced for ultrasmall NPs (∼3 nm).57 The MS values for both our samples are very close to the bulk Mn-ferrite, suggesting a very small surface effect on low-temperature magnetization. Indeed, the NP magnetic cores estimated by fitting the Langevin–Chantrell approximation58 to the room temperature data are 7 nm and 9 nm for S1 and S2 NPs, respectively, very close to TEM values. Since measurements were conducted in powder, we associate the decrease in the reduced remanent magnetization mainly to magnetic dipolar interactions, which can produce a demagnetizing effect on the system's remanence.59,60 Further investigations on the nature and intensity of the magnetic interactions would provide valuable insights into clarifying this issue.
We evaluated the magnetic anisotropy of our samples through the blocking temperature. The blocking temperature characterizes the transition from a (i) blocked regime, when the magnetic moments are frozen in the easy magnetization axis (pointing to one of the possible directions), to a (ii) superparamagnetic (SPM) regime, when the magnetic moment has enough thermal energy to flip from one direction to the other during the measuring time.
The transition between the two regimes depends on the NP volume, magnetic anisotropy, measuring time, and magnetic field strength. We measured the ZFC-FC curves to investigate the blocking temperatures, as illustrated in Fig. 4C. In monodisperse systems, the blocking temperature corresponds simultaneously to the peak of the ZFC curves and the point where the ZFC and FC curves intersect. However, for polydisperse NPs, these temperatures are notably distinct. When there is a volume distribution, there will be a distribution of blocking temperatures. Therefore, the broader distribution, the wider transition range between blocked to SPM regimes.
The distribution of blocking temperatures was investigated by calculating the derivative of the difference between ZFC and FC curves.61,62 The mean blocking temperatures TB are indicated in Fig. 4C and their values reported in Table 2. A secondary peak is evident in the derivative curve at T = 12 K for both samples, more pronounced for S2. Since NPs have a well-fitted log-normal size distribution, the presence of a secondary peak in the derivative curve could be related to a superspin glass state at low temperature63,64 due to strong dipolar interactions.65 We estimated the effective anisotropy constant Keff values for both samples by considering the values in Table 2 and using the following equation:66
![]() | (3) |
To go further in understanding the very small coercivity observed at room temperature, we estimated the critical diameter for superparamagnetic behaviour at room temperature by assuming T = 300 K in the expression DC (T) = (150kBT/πKeff)1/3. We found that for both S1 and S2 samples, NPs larger than 24 nm are blocked at room temperature. Such sizes correspond to less than 0.1% of S2 NPs and 1.2% of S1 NPs. These estimations suggest that most NPs of the size distribution are in a superparamagnetic regime at room temperature.
![]() | ||
Fig. 5 (A) Typical measurement of a line of ferrofluid sample deposited onto the nitrocellulose membrane, measured by the inductive magnetic sensor. On the inset: nitrocellulose and backing card measurement, without deposited NPs (left) and “coffee-ring” pattern (right). (B) Magnetic sensor signal as a function of the deposited NP mass. The inset shows the sensor sensitivity using our MnFe2O4 NPs and previous results with 89 nm highly dense Fe3O4 nanoclusters (gray striped bar)7 and 10 nm coprecipitated Fe3O4 NPs (gray bar).46 (C) AC initial magnetic susceptibility measured at high frequencies. Both real and imaginary components are shown for the two samples. |
Measurement is carried out after drying for 24 h. The term “magnetic signal” refers to the relative increase of the impedance due to the contribution of all the deposited NPs (see the highlighted area in Fig. 5A). It is calculated using eqn (2) and considering the modulus of the complex impedance |Z|. We deposited the ferrofluid samples (before bioconjugation) with different concentrations onto the nitrocellulose membrane and measured their sensor signal as a function of NP mass. Results are presented in Fig. 5B. We obtained a linear trend for this range of NP mass, which agrees with previous studies with the same sensor.42 The slope of the best-fitted straight line corresponds to the sensor sensitivity, which is the relative increase of the impedance per unit mass of NPs as follows:
![]() | (4) |
The obtained values are shown on the inset of Fig. 5B: 226% per mg of S1 NPs and 243% per mg of S2 NPs. The slightly superior performance of S2 NPs compared to the S1 NPs can be attributed to the higher NP mean diameter, which can produce a higher magnetic initial susceptibility. Both values are higher than the obtained ones for magnetite measured with the same sensor: 23% per mg of 10 nm coprecipitated Fe3O4 NPs covered by a double layer of oleic acid46 and 121%/mg of magnetite nanoclusters coated with polyacrylic acid (PAA).7 We also determined the resolution, which is the minimum mass that can be detectable by the sensor, given by
R = 3σNoiseN/Σ, | (5) |
We measured the frequency dependence of the complex initial magnetic susceptibility χ = χ′ − iχ′′ in high frequency. The real part is proportional to the induced in-phase magnetization, while the imaginary part is proportional to the out-of-phase magnetization (magnetic losses). Results are shown in Fig. 5C.
The dashed line in Fig. 5C highlights the frequency of 20 MHz, which was the same used for the measurements in the magnetic sensor, providing a good signal-to-noise ratio. The real part of the susceptibility χ′ is slightly higher for S2 NPs. Regarding the imaginary part (χ′′), there are similarities between the two samples, including a presumed resonance frequency of 630 MHz. The higher susceptibility for S2 samples is likely related to the larger mean NP diameter. Nonetheless, the values for the two components of initial magnetic susceptibility (see Table 2) are an estimate because the measurements were made on powders, where the magnetostatic interactions among the particles are strong and would reduce the intrinsic initial magnetic susceptibility.71 The resonance peaks in the imaginary part are expected to occur when 2πfτ = 1, τ being only the Néel relaxation τ = τ0exp(KeffV/kBT), since particles cannot move, and τ0 the spin relaxation time. We estimated the effective anisotropy constant value by assuming the resonance frequency f = 630 MHz and τ0 = 10−10 s. The results for S1 and S2 samples are 1.4 × 104 J m−3 and 0.7 × 104 J m−3, respectively, both are very close to those presented in Table 2, which were obtained by LAS fitting to the 300 K data.
The biological fluids were tested with the LFA; in all of them, the proteins labelled with magnetic NPs were retained at the biotin line, forming a visible test line (see the upper part of Fig. 6C). The lines were quantified using a radiofrequency inductive sensor and a mobile telephone camera.
We started with a high concentration of NeutrAvidin (1000 μg mL−1) to test the bioconjugation protocol immediately after different reaction times in a refrigerated ultrasonic bath. Results are shown in Fig. 6A. The magnetic signal is proportional to the number of nanolabels retained in the test line, which depends on the amount of NeutrAvidin tagged with NPs. We obtained a fast bond between the citrate-coated S2 NPs and the NeutrAvidin protein. For the EDC-mediated reaction (gray bars in Fig. 6A), we observed a high signal after 1 min, which decreased with sonication time. Without using the EDC crosslinker (orange bars in Fig. 6A), we achieved an optimized signal after 5 and 10 min of sonication. We kept these two optimized conjugates resting for 150 min without sonication and we repeated the assays, obtaining a reproducible magnetic signal at test line, as shown in the inset of Fig. 6A. This suggests that NPs@NeutrAvidin bonding occurs fast and is stable at rest using the present protocol. However, a reduction in test line signal occurs when increasing sonication time, suggesting a partial rupture of the S2-NeutrAvidin conjugate, likely related to the NeutrAvidin desorption from the NP surface.
Fig. 6B depicts the magnetic and optical signal as a function of NeutrAvidin concentration (8–1000 μg mL−1). For such an investigation, we chose the optimized sonication time of 5 min without EDC chemistry. We also checked the specificity of the NeutrAvidin/biotin biorecognition in our LFA by using the same protocol, however, without NeutrAvidin (blank control sample). The test lines of our LFAs are shown in the upper part of Fig. 6B. No signal was obtained for the blank sample, confirming the specificity of the assays. The magnetic signal increases with increasing NeutrAvidin concentration until a saturated value of ∼1700 μΩ mm above 250 μg mL−1 of NeutrAvidin, corresponding to ∼949 ng of S2 NPs retained in the test line. The trend is linear for the range of NeutrAvidin concentrations until ∼31 μg mL−1, as seen on the inset of Fig. 6B, together with the fitting parameters, corresponding to 474 ng of S2 NPs in the test line. For the smaller NeutrAvidin concentration we tested (8 μg mL−1), we found 133 ng of NPs in the test line.
The optical signal obtained with a mobile telephone camera had very similar behavior. We got a better coefficient of determination R2 and lower standard deviation with the magnetic measurements than with the optical data. This is mainly related to variations in optical signal because of different ambient light.
The hydrodynamic diameters obtained by DLS (see Fig. 6C) increased from 29 nm (before bioconjugation) to 38 nm (after 1 min of NeutrAvidin bioconjugation). This difference corroborates the bioconjugation with NeutrAvidin, which has a diameter of 4–5 nm (60 kDa),72,73 and the absence of a significant post-bioconjugation clustering.
The magnetic signal obtained for our 10 nm MnFe2O4 NPs (S2) bioconjugated with NeutrAvidin (1000 μg mL−1) outperformed previously published results with different Fe3O4 NPs with the same NeutrAvidin concentration and measured with the same sensor and frequency: 573 μΩ mm for 10 nm NPs covered with oleic acid, 630 μΩ mm for 7.9 nm lauric acid-coated NPs and 1250 μΩ mm for 9.5 nm NPs covered with myristic acid.46 Also, we achieved similar values to the reported one for 89 nm Fe3O4 nanoclusters covered with polyacrylic acid.7 Since S2 NPs are smaller, this may be attributed to the heightened sensitivity of MnFe2O4 NPs for detection in inductive sensors.
The fast binding of NeutrAvidin to the NPs in the bioconjugation protocols we tested probably occurred due to an electrostatic union between negative ions on the NP surface and positive charges of the NeutrAvidin in PBS (1 mM, pH 7.4), expected because of its isoelectric point74 of 6.3. Indeed, chemical bonding between proteins and the nanoparticle surface may occur in several ways, including electrostatic union between negative charges on the NP surface and positive charges in the protein and covalent binding between carboxyl on the NP surface and amines in the protein, among others.75 The latter is generally intermediated by the EDC/NHS chemistry76 and occurs slowly, but is generally preferred in immunoassays due to its greater stability. Therefore, for future development of real immunoassays, the bioconjugation protocol must be optimized considering the biomarkers of interest for a specific immunoassay architecture.
We further investigated the use of 10 nm S2 NPs as magnetic tags in LFAs using the NeutrAvidin/Biotin model of biorecognition. In our protocol, we achieved an intense signal on the test line just after 5 min of bioconjugation without employing EDC chemistry, suggesting an electrostatic bonding between the NP surface and the NeutrAvidin. The magnetic signal obtained for the 1 mg mL−1 of NeutrAvidin was compared to that obtained for magnetite nanoclusters on our inductive sensor. Our findings suggest that MnFe2O4 nanoparticles and their assemblies have strong potential for further development magnetic lateral flow immunoassays for biomarkers detection and sensitive magnetic quantification.
Footnote |
† Electronic supplementary information (ESI) available: Table ESI 1 – Resulting parameters from the law of approach to saturation fitting for 300 K and 5 K magnetization measurements. See DOI: https://doi.org/10.1039/d4na00445k |
This journal is © The Royal Society of Chemistry 2024 |