Enhanced detection of quantum dots labeled protein by simultaneous bismuth electrodeposition into microfluidic channel

In this study, we propose an electrochemical immunoassay into a disposable microfluidic platform, using quantum dots (QDs) as labels and their enhanced detection using bismuth as an alternative to mercury electrodes. CdSe@ZnS QDs were used to tag human IgG as a model protein and detected through highly sensitive stripping voltammetry of the dissolved metallic component (cadmium in our case). The modification of the screen printed carbon electrodes (SPCEs) was done by a simple electrodeposition of bismuth that was previously mixed with the sample containing QDs. A magneto‐immunosandwich assay was performed using a micromixer. A magnet placed at its outlet in order to capture the magnetic beads used as solid support for the immunoassay. SPCEs were integrated at the end of the channel as detector. Different parameters such as bismuth concentration, flow rate, and incubation times, were optimized. The LOD for HIgG in presence of bismuth was 3.5 ng/mL with a RSD of 13.2%. This LOD was about 3.3‐fold lower than the one obtained without bismuth. Furthermore, the sensitivity of the system was increased 100‐fold respect to experiments carried out with classical screen‐printed electrodes, both in presence of bismuth.

the sensitivity of the electrochemical immunoassays. Among these nanoparticles, are the quantum dots (QDs), also known as semiconductor crystals, they have been mainly used as fluorescence labels; however they can also be detected by dissolving them liberating the metal ions present in their core using the same strategies applied for heavy metals detection [3].
Different materials have been used for metal detection, such as mercury [4], bismuth (Bi) [5], graphene [6], etc. The high toxicity of mercury film electrodes, which have been applied for the determination of metals by stripping voltammetry [7,8], makes necessary the search for more environmental-friendly materials with similar performance. One interesting and cheap alternative is the bismuth, which was used for first time by Wang et al. [9], as bismuth film electrode (BiFE) on glassy-carbon or carbon-fiber substrate for the detection of microgram per litter levels of cadmium, lead, thallium, and zinc by stripping voltammetry. Apart the low toxicity of BiFEs compared to mercury, they offer other advantages such as ability to form alloys with different metals, wide potential ranges, and low sensitivity to dissolved oxygen. They have already been applied for the detection of several metals. Calvo-Quintana et al. coupled Bi-modified electrodes to a small portable instrument for the detection of lead at legal limits (20 g/kg) in a complex matrix such as milk [10]. Caldeira et al. [11] applied Bi films for the determination of traces of Tl(I) separately and together with Zn(II) and Pb(II), with a LOD of 2 nmol/L.
Regarding the biosensing applications, the detection of Zinc (Zn) in patients with problems in the functional immune system has been evaluated by Papautsky's group. Those Zn concentrations decrease in critically ill patients and could potentially benefit from Zn supplementation as a therapeutic strategy. The authors reported the Zn faster detection by anodic stripping voltammetry on bismuth microfabricated electrodes in serum as well as in bovine serum extract [12,13].
To the best of our knowledge, bismuth has not been considered as an electrode modifier into microfluidic platforms, in order to increase the sensitivity of the QDs-based electrochemical immunosensors. However there are some related examples made in batch like the work made by Chen's group, which consists in the development of an electrochemical biosensor for the neutravidin recognition by using QDs as labels, which were detected by using an electrodeposited bismuth film, achieving a LOD of 5 nM [14]; and a biosensors based on bismuth modification for protein and DNA detection, which was proposed by Kakabakos's group. In this work, the authors used stripping voltammetry technique for the in-batch CdSe@ZnS QDs detection, which was used as labels for the prostate-specific antigen (biomarker) and DNA quantification [15].
Other approach by using a bismuth film was used as a sensing platform for immunoreaction assay that includes the interactions between IgE and anti-IgE molecules. IgE protein was deposited on a carbon paste electrode via bismuth (III) cations without using any membrane or functional reagent. The immobilized reagents and interaction of anti-IgE on this formation were monitored by electrochemical impedance spectroscopy [16].
Bi can be deposited onto screen-printed electrodes (SPEs) following different procedures, in different conditions and on different materials depending on the target analyte or the aim of analysis. The fabrication can be done using "ex situ," "in situ," and "bulk" approaches. In the "ex situ" method, the electrodes are immersed in a solution containing Bi 3+ and a potential is applied to reduce it to Bi for its deposition [17]. In the in situ method, both Bi and the target analyte are electrochemically deposited on the electrodes, forming alloys followed by anodic stripping of metals from the electrode [18]. The third option is based on the incorporation of bismuth oxide to the ink for the fabrication of the SPE, which is then reduced to metallic bismuth for the formation of a Bi layer [19]. In our group, a phenol detection system by immobilization of a mushroom tissue onto SPE and multiwall carbon nanotube (MWCNT) modified SPE (SPE/MWCNT) via Bi deposition was performed. This sensor achieved a LOD for phenol of 0.48 mM and 1.17 mM for SPE/Bi/Tissue and SPE/MWCNT/Bi/Tissue, respectively [20].
In the presented work, a versatile platform able to operate as a sensitive immunoassay for HIgG detection within a microfluidic system is developed (Supporting Information Fig. S1). The modification of SPE by in-flow simultaneous electrodeposition of a thin Bi layer together with the QDs labelled protein has been performed and evaluated so as to improve the QDs electrochemical signal. The developed system is useful not only for QDs detection with interest for immunosensing but for other applications with interest for environment control such as heavy metal detection.
Reagents: Streptavidin−CdSe@ZnS quantum dots (QD655), tosylactivated magnetic beads MyOne (1 m of diameter), Human IgG, anti-Human IgG produced in goat (␣-HIgG), and biotin anti-Human IgG (biotin ␣-HIgG) produced in goat were obtained from Invitrogen (Spain). PBS, Tween 20, casein, BSA, and bismuth (III) nitrate pentahydrate (reagent grade 98%) were acquired from Sigma Aldrich (Spain). PBS supplemented with 5% w/v of casein and 0.005% v/v of Tween 20 was prepared as blocking buffer for the channel wall treatment. PBS supplemented with 5% w/v of BSA and 0.005% v/v of Tween 20 was prepared as blocking buffer to avoid unspecific absorption onto the magnetic beads. PBS supplemented with Tween 20 at 0.05% v/v was used as washing buffer (PBST). PBS was employed as immunobuffer, except for QDs conjugation. In this case, commercial buffer Borate buffer, pH: 8.3, 0.05% sodium azide) was used. Rare earth (neodymium iron boron, 1.72 T) permanent magnets were purchased in Magnetladen, Germany.
Apparatus: Electrochemical experiments were performed using an electro-chemical analyzer Autolab 20 (Eco-Chemie, The Netherlands) that was connected to a personal computer using a software package GPS 4.9 (General Purpose Electrochemical System). SEM analyses were performed by using an EVO (Carl Zeiss NTS GmbH, Germany). Bare and bismuth modified screen-printed carbon electrode (SPCE) were analyzed both in drop and flow configurations.
Enhanced quantum dots electrochemical detection by using bismuth: Chips were produced following the same protocol previously reported by our group [21]. Both measurements, in batch and in-chip were tested in order to compare the sensitivity of the systems. CdSe@ZnS QDs (10 L at 5 nM of concentration) were first partially dissolved by using HCl 1 M for 2 min (adding 50 L). After dissolving them, 200 L of acetate buffer (pH 4.5) was added to the solution. For the comparison between samples with and without bismuth, 40 L of bismuth (at different concentrations) and PBS was added, respectively. The first experiment consisted in the optimization of the bismuth concentration for both platforms. Since the cadmium reduction potential (-1.1 V) is also suitable for the bismuth electrodeposition the same potential value is used. The sample containing bismuth was introduced in the same solution, ensuring the simultaneous reduction of the cadmium (II) and bismuth (III) (within 120 s) (Supporting Information Fig. S2).
Immunoassay performance: Briefly, 40 g of tosylactivated magnetic particles (15 L from stock solution) were washed three times with borate buffer and suspended  (2) immobilization of magnetic beads already modified with the capture antibody and blocked, (3) conjugation step of the model protein and detection antibody already modified with QDs. Between each step, a washing step is performed by using PBST. After the last washing step, in (4) HCl 1 M is introduced from the outlet to dissolve the external layer of the QDs and liberate the Cd ions from the QDs core to be detected in step (5) after the introduction of acetate buffer.
in 135 L of the same buffer. These magnetic beads (MBs) were functionalized with capture antibody (anti HIgG 10 g/ml) by electrostatics by using a borate buffer pH 8.5 overnight under agitation (700 rpm, 37ºC). The conjugated MBs were washed with PBST three times in order to remove the antibody excess, followed by a blocking step using PBS supplemented with Tween 20 at 0.05 % v/v and BSA at 5% for 2 h under agitation conditions (700 rpm, 25ºC). The blocked MBs were washed again with PBST three times by using the magnetic separator. Then, the microchannel was blocked by using PBS supplemented with Tween 20 at 0.05% v/v and casein at 5%. This solution was introduced in the microchannel for 10 min at 5 L/min. After the channel surface blocking, PBST was introduced for 10 min more at the same flow rate in order to clean the excess of the blocking solution.
The working electrode was activated by cyclic voltammetry (ten cycles at a potential range from -0.8 to 0.8 V and a scan rate of 100 mV/s in PBS buffer). Magnetic beads already modified with antibodies were immobilized into the mixer by placing a permanent magnet behind the mixer outlet, with the optimized volume and flow-rate (2 min at 5 L/min).
A 7.5 L of QDs (2 M from stock solution) were suspended in 492.5 L of QD incubation buffer (50 mM borate pH 8.3, 0.05% sodium azide), as well as 45 L of Anti-Human IgG (gamma-chain specific)-biotin. They were mixed and incubated for 30 min (750 rpm 25ºC) in batch. Consequently the excess of antibody was removed by filtering using eppendorf filters (Eppendorf R Safe-Lock microcentrifuge tubes, with pore size of 100 KDa from Millipore) and centrifuged at 15 000 rpm for 5 min [22].
An optimized micromixer (500 m diameter and 50 m thickness) with two inlets was fabricated and used for the immunoassay performance (more details in Results section). In one of the inlets, a syringe containing the analyte (Human IgG) is located while the antibody-QDs were located in the second inlet. Both of them were simultaneously introduced at a flow-rate of 2.5 L/min during 15 min. After the mixing and incubation, the channel was washed with PBST at a flow rate of 2.5 L/min for 15 min. Then, a dissolving solution (HCl 1 M) was introduced from the outlet in order to dissolve the QDs (labels in the immunoassay). This procedure was performed at flow-rate of 5 L/min for 5 min (the time needed to fill the entire channel at this flow-rate). The dissolving solution was remained in static mode for 3 min. Finally, the mixer was connected to the measurement chip (a simple channel with integrated screen printed electrode), then, acetate buffer (pH. 4.5) was introduced from the two inlets (micromixer) with a flow rate of 5 L/min and the electrochemical measurement was performed without applying a conditional potential as a new chip was used for each measurement, being not necessary to reoxidize the cadmium deposited onto the electrode after the measurement. The deposition potential was defined at -1.1 V, and the deposition time was set at 400 s to allow all the solution to reach the electrode surface. The stripping measurement was done from -1.1 V to -0.15 V with  A graphical representation of the steps involved in the immunoassay performance and detection is shown in Fig. 1.
In Fig. 2A, the QDs signal for different bismuth concentrations in drop format is shown. The continuous lines show how the peak is increased when bismuth is added, while the dotted lines represent the system saturation after 50 g/ml bismuth concentration. In Fig. 2B, the same results are shown for in-chip measurements. In the microchannel system, the saturation is achieved with smaller bismuth concentration (25 g/ml) compared to the drop format.
In batch measurements, the use of bismuth causes an increase of the electrochemical signal of around 1.2folds, with a sensitivity of 7 × 10 −6 A/nM and R 2 of 0.997 (Fig. 3A), while in chip, the increasing is 5.5-fold with a sensitivity of 5 × 10 −6 A/nM and R 2 of 0.915 (Fig. 3B). In order to compare the two systems, current density was calculated. By comparing the chip with the batch measurements, the achieved sensitivity is 100-fold better in flow (sensitivity of 20 × 10 −6 A/nM) than in batch mode (sensitivity of 0.2 × 10 −6 A/nM) (Fig. 3C). This means that the chip format helped to improve the sensitivity due to the fact that the sample is permanently flowing onto the working electrode, and the analyte is close to the electrode surface due to the channel dimensions.
This increasing of sensitivity can be also attributing to the more uniform film of bismuth obtained by in-flow electrodeposition. Indeed, the modification of the SPCEs after the deposition of bismuth in flow and static modes has been evaluated by scanning electron microscopy. Homogeneous distribution of the Bi on the electrodes surface can be observed when the deposition is in flow mode due to the continuous renovation of the Bi solution onto the electrode and the diffusion phenomena that occurs faster than in static mode (Fig. 3A, B, C).
In order to demonstrate the use of bismuth in microfluidics as strategy to improve the detection of QDs as labels in a magneto-immunoassay, a micromixer was integrated with a detection channel (a simple channel with integrated screen printed electrode). In order to check the mixing performance, trypan blue dispersion in water (obtained from Sigma Aldrich) were used in different mixers geometry in order to mimic the experimental conditions ( Figure S3A). In Supporting Information Fig. S3B a higher dispersion rate of trypan blue is observed after the mixing process (at the end of the channel) in the chosen mixer (Supporting Information Fig. S3B-I) in comparison with the others. In the same Figure, it is also possible to observe the diffusion interphase between trypan blue suspended in water compared with water only. This diffusion interphase is less defined when the flow rate increases from 2.5 to 50 L/min, even that, the complete mixing is achieved at the end of the channel for the tested range of flow-rates (Supporting Information Fig. S3B-ii,iii. The conjugation parameters in flow were also optimized, in order to choose the shortest incubation time without losing efficiency (results do not shown). All incubation steps were performed by changing the flow rate from 1 to 5 L/min. The most efficient conjugation was obtained using flow rates up to 5 L/min, otherwise, the higher flow could reduce the efficiency of binding event inside the micromixer. Different incubation time intervals such as 5, 10, and 15 min were also evaluated, times higher than 15 min did not show any significant improvement. A proper electrochemical signal was observed at about 15 min. Indeed the first 5 min were necessary to fill the entire micromixer at this flow rate, being necessary 10 min more to ensure the binding between the immobilized magnetic beads and the detection antibody labeled with QDs. The effect of washing time (set at 5 L/min) was also studied finding that 10 min was a sufficient time to clean the channel from antibodies excess.
For the detection, acetate buffer (pH. 4.5) was introduced from the two inlets with a flow rate of 5 L/min and the electrochemical measurement was performed as it was explained in the experimental part (Immunoassay performance) (without conditional potential because for each measurement, a new chip was used, so it was no necessary to reoxidize the cadmium deposited onto the electrode after the measurement). The deposition potential was defined at -1.1 V as already reported by our group [23], and the deposition time was set at 400 s to allow all the solution containing the QDs, reach the electrode surface. More than 400 s in this application does not improve the electrochemical signal because the whole QD-conjugated has been flowed through the detection chip within this time. The stripping measurement was done from -1.1 V to -0.15 V with a step potential of 5 mV/s and amplitude of 0.3 V. This measurement was performed in the presence or not of bismuth in acetate buffer.
In order to confirm the advantages of the use of bismuth in the immunoassay performance, a magneto-immunoassay with three different concentrations of analyte (0, 50, and 100 ng/mL) was measured. After the cleaning and dissolving steps, the measurements were done in acetate buffer in presence or absence of bismuth. It can be observed that the sensitivity in the case of immunocomplex (labeled with QDs) is lower than in the case of QDs alone due to the fact that the biological sample inhibits the electrochemical signal and affects the electrode performance along the time; however the increasing of the signal for the measurement with the presence of bismuth was about 3.3 times better with a RSD of 13.2% in relation to the initial LOD of 10 nM with a RSD of 5.7%.
The increase of the sensitivity for an immunoassay based on QDs as electrochemical labels was demonstrated as a proof of concept with the performance of a complete magneto-immunoassay for human IgG determination (optimized protocol in the previous sections) into a microfluidic system, by addition of bismuth in the acetate buffer at the last step. It is an easy and promising strategy to improve the LOD with a simple addition in the measurement buffer without the addition of more steps in a magneto-immunoassay (Fig. 4).
Different parameters (e.g. bismuth deposition, flow rate, incubation time, mixer geometry, etc.), are optimized to achieve sensitive and reproducible measurements. The LOD for human Immunoglobulin G (HIgG) in presence of bismuth is very low (3.5 ng/mL) with a RSD of 13.2 %. This LOD is 3.3 lower than the one obtained without bismuth (10 ng/mL). Moreover, the sensitivity of the system is increased 100-folds in comparison with the batch experiments carried out with classical SPCEs (with the sample containing Bi).
The device takes advantage of the speed and low cost of the conventional immunoassay technologies. Under optimal conditions, the device was capable of detecting three times less concentration than the conventional system, in 6.6 min by using the sample containing the bismuth solution at the optimized concentration.
The exposed devices coupled with a portable electrochemical analyzer are promising for in-field quantitative testing for disease-related protein biomarkers. The developed device thus provides an efficient tool for rapid and quantitative detection of protein that can also be extended to DNA analysis.