Optical modulation biosensing system for rapid detection of biological targets at low concentrations.

In many sensitive assays, target molecules are tagged using fluorescently labeled probes and captured using magnetic beads. Here, we introduce an optical modulation biosensing (OMB) system, which aggregates the beads into a small detection area and separates the signal from the background noise by manipulating the laser beam in and out of the cluster of beads. Using the OMB system to detect human interleukin-8, we demonstrated a limit of detection of 0.02 ng/L and a 4-log dynamic range. Using Zika-positive and healthy individuals' serum samples, we show that the OMB-based Zika IgG serological assay has 96% sensitivity and 100% specificity.

. Schematic representation of a typical two-site sandwich immunoassay. A target molecule, such as a human interleukin-8 (IL-8) protein, is first captured by an antibody that is conjugated to a magnetic bead, and then detected by a fluorescently labeled antibody. The number of target molecules in the samples is proportional to the number of fluorescent molecules that are attached to the magnetic beads. detection area significantly increases the signal. Hence, MMB-assisted assays are characterized by high sensitivity and shorter and simpler testing protocols [14,15].
In MMB, to efficiently separate the signal from the background noise, the beads have to be manipulated from side to side relative to the fixed laser beam for at least 12 cycles. Thus, the total acquisition time is extended to 12 seconds, resulting in prolonged exposure of the beads to a high intensity laser beam. Such exposure could lead to significant photobleaching of the sample [18] that may decrease the signal over time and increase the variability of the results. Moreover, the long data acquisition time hinders the use of the system for high throughput applications.
Here, we suggest a new approach, termed optical modulation biosensing (OMB), that produces the relative movement between the magnetic beads and the laser beam by fixing the magnetic beads and modulating the laser beam. Modulating the laser beam is much faster than modulating the aggregated beads, and therefore the data acquisition time is significantly shorter (∼250 ms). Using the OMB system to detect human Interleukin-8 (IL-8), we demonstrate a limit of detection of 0.02 ng/L and a 4-log dynamic range. This level of performance is comparable with that of the most sensitive devices but is achieved without their bulk and cost [19]. We further demonstrated the clinical sensitivity and specificity of the OMB system by detecting anti-Zika virus (ZIKV) Immunoglobulin G (IgG) antibodies in serum samples from quantitative reversetranscriptase polymerase chain reaction (qRT-PCR) or neutralization Zika-positive patients and healthy individuals. The results were compared to the state-of-the-art EUROIMMUN ELISA test.

Optical modulation biosensing (OMB) system setup
The OMB system (Fig. 2) uses a 532 nm collimated laser diode module (CPS532, Thorlabs Inc, Newton, NJ, USA) working at 0.25 mW. The laser beam diameter (3.5 mm) is reduced by two plano-convex lenses (not shown) and projected onto a scanning galvo mirror (GVS211, Thorlabs Inc, Newton, New Jersey, USA) at a 45°angle. The beam is reflected by a dichroic mirror (Di02-R532, Semrock Inc, Rochester, New York, USA) into a microscope objective lens (10x, 0.25 NA, 16.5 mm focal length, Newport Corporation, Irvine, California, USA). The objective lens focuses the beam to a 150 µm diameter spot on a rectangular borosilicate sample cell, which contains the magnetic beads and has inner dimensions of 8 mm x 0.4 mm x 70 mm (W2540, VitroCom, Mountain Lakes, New Jersy, USA). The emitted fluorescence is collected using the same objective lens, then passed back through the dichroic mirror and two emission filters (Semrock, FF01-575/25 for phycoerythrine, and FF01-560/25 for ATTO 532). The fluorescence is spatially detected by a camera (GS3-U3-23S6M-C, Flir Integrated Imaging Solutions, Richmond, BC, Canada), with a frame rate of 50 frames per second (fps) for 2 s (i.e., a total of 100 frames).

Fig. 2.
Schematic representation of an optical modulation biosensing system. The laser beam from a collimated 532 nm laser diode module is projected on a scanning galvo mirror at a 45°angle and reflected by a dichroic mirror into a microscope objective lens. The objective lens focuses the beam on a rectangular borosilicate sample cell, which contains the magnetic beads. The emitted fluorescence is collected using the same objective lens, passes back through the dichroic mirror and two emission filters, and is detected by a camera. Magnetic beads are aggregated by an alternating magnetic field gradient, applied to the sample by two electromagnets.
The magnetic beads are aggregated by an alternating magnetic field gradient applied to the sample by electromagnets A and B (Fig. 2), located on each side of the sample cell. After 120 seconds, the aggregated beads are fixed to side A by turning off the current to electromagnet B and keeping the current to electromagnet A constant. Subsequently, the laser beam is set in a periodic lateral motion on the sample cell and the camera is turned on.
The lateral movement of the laser beam is generated by connecting the scanning galvo mirror's motor to a power source (GPS011, Thorlabs Inc, Newton, NJ, USA) and a function generator (AFG3022B, Tektronix, Beaverton, OR, USA). The mirror is rotated by a square waveform at a frequency of 2 Hz with an amplitude of 225 mV, which moves the laser beam back and forth over a distance of 250 µm on the sample cell (Fig. 3). To prevent overheating, the motor is attached to a heat sink (GHS003, Thorlabs Inc, Newton, New Jersy, USA).

Image analysis
The laser beam is modulated at 2 Hz and the camera acquisition frame rate is 50 fps. Therefore, within two seconds, there are four modulation cycles and 25 data points per cycle. During the first half of the modulation cycle (250 ms) the beam is located at side B (illuminating the background solution), whereas during the second half of the modulation cycle, the beam is located at side A (illuminating the beads). For each measurement, 100 frames were acquired over a period of two seconds. Each frame was divided into two areas, the side with the fixed beads (A) and the side without the beads (B). To identify the position of the beam in each area, the frames were cropped and thresholded. A binary mask of the beam's position in each area was applied to all the frames. The mean grey value (MGV) in each frame was calculated and graphed as a function of time.
In general, when the laser beam is located at side B (i.e., the background solution, Fig. 3(a)), the signal intensity at side B, I B , equals the fluorescence intensity of the background, I background , and is given by where I R is background noise originating from Raman scattering of solvent molecules, I UF is residual fluorescence from unbound fluorescent molecules or other impurities, and I dark current is signal originating from the dark current of the camera. When the laser beam is located at side A (i.e., on the beads, Fig. 3(b)), the signal intensity at side A, I A , equals the fluorescence intensity of the beads, I beads , and is given by where I signal is the net fluorescence intensity of the fluorophores that are attached to the beads and I AF is the autofluorescence of the magnetic beads. The signal that originates from the autofluorescence of the magnetic beads, I AF can be significantly minimized, down to 1% of its initial value, by photobleaching the beads [18]. Thus, the net fluorescence intensity of the fluorophores can be estimated by subtracting Eq. (1) from Eq. (2): When the laser beam is located at side A or side B, the corresponding signal intensity at the opposite side (i.e., I B or I A , respectively) is approximately I dark current .
To obtain the required number of ATTO 532 molecules/bead, we diluted the stock solution of the biotinylated fluorescent dye as follows. According to manufacturer's data, the stock solution of 1 mg/mL of biotinylated ATTO 532 contains 4.42 · 10 14 molecules/µL. To reach a concentration of ∼1 * 10 14 molecules/µL, we added 3 µL of the stock solution to 10 µL of PBST buffer and vortexed for 15 s. Then, 10 µL of this initial dilution (i.e., a total of 1 · 10 15 molecules) were further diluted in 90 µL of PBST buffer, resulting in a concentration of 1 · 10 13 molecules/µL. A series of nine additional dilutions (10 µl of the previous dilution and 90 µL of the PBST buffer) were prepared, down to a concentration of 1 · 10 4 molecules/µL. Each sample contained 2.5 µL of the appropriate dilution, supplemented with 97.5 µL of the assay buffer containing ∼25,000 M280 streptavidin superparamagnetic beads. The samples were incubated and constantly mixed for one hour at room temperature. As the blank measurement (i.e., beads without fluorescent molecules), we used 2.5 µL of PBST buffer without the fluorescent molecules.
Following a single buffer replacement, the beads were analyzed using the OMB system. The measured fluorescent signal was calculated according to Eq. (3), and was directly related to the number of dye molecules on each bead. Before their use in the assay, the magnetic beads were photobleached for 18 h in the buffer solution [18].

Dose response of human interleukin 8 (IL-8) assay
The OMB system was further evaluated utilizing a two-site sandwich immunoassay, using human interleukin 8 (IL-8) as the target biomarker ( Fig. 1). The assay was carried out in assay buffer, using a commercial IL-8 assay kit (BioRad, Hercules, CA, USA, CXCL 171BK31MR2). The kit consisted of biotinylated detection antibodies and magnetic beads that are coupled to capture antibodies. A Bio-Plex Pro Reagent Kit III (Bio-Rad, #171304090M), consisting of streptavidincoupled phycoerythrin fluorescent dye (SA-PE) and reaction buffers, was used with the assay kit in the course of the experiment. To obtain a dose-response measurement in buffer, 50 µL of phosphate buffered saline supplemented with 0.1% (v/v) of Tween 20 containing magnetic beads (∼6000 beads/reaction) was mixed with 50 µL of assay buffer containing varying concentrations of IL-8 (Biolegend, San Diego, CA, Cat.# 574202), resulting in final concentrations of 0.05, 0.1, 1, 100, 1 × 10 3 , and 1 × 10 4 ng/L. The reaction mixtures were incubated for one hour at room temperature. The initial incubation was followed by a 30-minute incubation with 50 µL of 1x biotinylated detection antibody solution and then a 20 min incubation with 80 µL of 1x SA-PE solution. During the incubation with SA-PE, the samples were protected from light. All incubation steps were performed in a 96 well plate on a rotary shaker at room temperature. To remove the unbound SA-PE from the solution, a single buffer replacement was performed at the end of the protocol, using a MagJET magnetic separation rack (ThermoFisher Sci, Waltham, MA, USA, Cat.# MR03). Following the single buffer replacement, the beads were analyzed using the OMB system. The measured fluorescent signal was calculated according to Eq. (3), and was directly related to the number of dye molecules on each bead.

OMB-based anti-Zika IgG clinical assay
A total of 25 serum samples from qRT-PCR or neutralization Zika-positive patients and 25 serum samples from healthy individuals were obtained from the National Center for Zoonotic Viruses at the Central Virology Laboratory of the Ministry of Health at Sheba Medical Center, Israel. The Zika-positive samples, acquired from day 7 onward post symptoms onset, were collected from Israeli travelers presenting at the Institute of Tropical Medicine at Sheba Medical Center after returning from Zika endemic areas. Zika virus qRT-PCR and neutralization test procedures were described previously [20]. Patient information, including age, gender, and the day the sample was taken after the onset of symptoms, was obtained from the electronic medical record. All the samples (50 µL each) were stored at −20°C, delivered to Bar-Ilan University on dry ice, and then thawed once for the OMB assay. The study was approved by IRB of Sheba Medical Center.
Each OMB-based assay consisted of ∼25,000 tosylactivated magnetic beads (M-280, Thermo Fisher Scientific, USA), which were pre-conjugated to the Zika NS1-suriname strain (The Native Antigen Company, UK) using Thermo Fishers' standard coupling procedures. The beads were incubated for one hour, at room temperature with 2 µL of serum sample diluted in buffer. After a single washing step, a fluorescently labeled detection antibody (donkey F(ab')2 Anti-Human IgG -H&L (PE), Abcam plc., UK) was added to the beads, followed by another washing step. Duplicates from the final solution were measured using the OMB system. The samples were also tested by the Euroimmun ZIKV ELISA (Euroimmun, Germany) according to the manufacturer's recommendations [20].

Evaluating the effect of modulation frequency on signal photobleaching
To evaluate the effect of the laser modulation frequency on the signal photobleaching, we used a sample containing magnetic beads covered with 10 5 ATTO 532 fluorescent molecules per bead. The sample was split to three and each was tested in the OMB system using a modulation frequency of 2, 5, or 10 Hz. The signal was recorded and analyzed according to the previously described procedure.

Results
A typical graph of the data acquired from a sample with a low concentration of target molecules is shown in Fig. 4. Moving the laser beam from one side of the cuvette to another produces a clear distinction between the fluorescent signal of the beads and the background noise of the sample matrix. The signal intensity of the magnetic beads (I A ), at the right side of the sample holder (marked in a solid red line, Fig. 4), is relatively high when the laser beam hits the beads ( Fig. 3(b)), and it is low when the laser beam hits the other side ( Fig. 3(a)). The signal intensity of the background solution (I B ), at the left side of the sample holder (marked in a solid blue line, Fig. 4), is low when the laser beam hits the background solution ( Fig. 3(a)), and it is lower still when the laser beam hits the other side ( Fig. 3(b)). When the laser beam illuminates the beads (Fig. 4, solid red line), the fluorescent molecules become photobleached and at the end of the first illumination cycle, the signal drops by 15% compared to the initial value. A slight recovery of the fluorescent signal can be seen in the following cycles (Fig. 4, red line), but at the beginning of the fourth cycle, the fluorescent intensity of the beads is approximately ∼9% less than the intensity at the beginning of the first cycle. To avoid the influence of the photobleaching effect, the fluorescent intensity of the beads, I beads , was extracted from the first data point of the second half of the first modulation cycle. Hence, the total acquisition time was ∼250 milliseconds. When the laser beam illuminates the background solution (Fig. 4, solid blue line), the fluorescence intensity of the background, I background , is nearly constant with no evidence of photobleaching.
The results of the ATTO 532 dose response experiment are presented in Fig. 5. The calculated limit of detection (LoD) for the ATTO 532 fluorescent dye is 95 fM, and the coefficient of variance is less than 16% for all the samples.

Fig. 5. Dose response of biotinylated ATTO 532 with streptavidin-coupled superparamagnetic beads.
Measurements were obtained using the OMB system. Error bars represent the standard deviation of three measurements (n = 3). Error bars of the blank measurement represent the standard deviation of six measurements (n = 6). The limit of detection (LoD) was calculated as three standard deviations over the average signal of the blank measurement. Figure 6 depicts the results of the IL-8 dose response experiment. Here, the calculated LoD is 0.02 ng/L, and the coefficient of variance is less than 22% for all tested samples. This LoD is slightly better than the LoD (0.04 ng/L) achieved previously by the MMB system for the same type of assay [21]. The results of the OMB anti-Zika IgG serological assay are presented in Fig. 7. The information for each sample, including the results of the EUROIMMUN ELISA and the OMB assays, is listed in Table 1. The receiver operating characteristic (ROC) cut-off (10.85) was determined based on previous results obtained using the MMB system taking into account minimal cross reactivity with other flaviviruses, such as dengue and West-Nile viruses [20]. The OMB-based IgG assay was able to detect 24 out of 25 Zika-positive samples (96% sensitivity), whereas the EUROIMMUN ELISA was able to detect 19 out of 25 (76% sensitivity). Both the OMB-based assay and the EUROIMMUN ELISA were able to detect all the negative samples as negative (100% specificity). It should be noted that the sample that was misidentified by the OMB-based assay as negative was acquired on day 8 post symptoms onset (i.e., possibly before the seroconversion) and was also misidentified as negative by the EUROIMMUN ELISA (Table 1).  The fluorescence signal (MGV) of the OMB system for different laser modulation frequencies is presented in Fig. 8. When the modulation frequency increases, the illumination cycles become shorter, effectively reducing the photobleaching effect. Thus, the signal drop between the first data points of the first and fourth modulation cycles decreases. For example, when the laser beam is modulated at 2 Hz, the signal between the first data points of the first and fourth cycles drops by 5%. When the laser beam is modulated at 5 and 10 Hz, the signal drops by 3% and 1.8%, respectively.

Discussion
Since its introduction in 2008, the MMB technology has been successfully applied in academic and clinical research to detect a variety of biomarkers [20][21][22][23]. High sensitivity, a short processing time, and relative simplicity make the MMB platform extremely appealing for point of care applications. However, to separate the signal from the background noise, the MMB system requires accumulation of sufficient modulation cycles that takes approximately 12 seconds. In OMB, the relative movement between the magnetic beads and the laser beam is achieved by immobilizing the magnetic beads at one spot and moving the laser beam back and forth, in and out of that spot. Hence, the acquisition time could be reduced to a fraction of a second, and the washing and separation steps were eliminated.
While the data acquisition time of the OMB system is much faster than the MMB system, their analytical and clinical performances are similar. For example, the minimum number of fluorophores per bead that can be detected by the OMB system (172 molecules per bead) and the LoD for IL-8 (0.02 ng/L) are similar to those achieved by the MMB system (65 molecules per bead and 0.04 ng/L) [21]. In addition, using qRT-PCR or neutralization ZIKV-positive samples, we show that the clinical sensitivity of the OMB-based assay is similar to the MMB system (96% vs. 97%) [20], much higher than the EUROIMMUN ELISA (76%).
The advantages of laser beam modulation, which is the core principle of the OMB system, over magnetic beads modulation, which is the core principle of the MMB system, are not limited to a much faster data acquisition time. Previously, we reduced the size and power consumption of the MMB system by replacing the electromagnet with a small permanent magnet that aggregates the magnetic beads from the entire sample into a small detection volume, and thereby significantly increases the signal [24]. However, a single permanent magnet cannot move the beads relative to the laser beam, and therefore to eliminate background noise from unbound fluorescent molecules, the assay incorporated five washing and separation steps [24]. In the OMB system demonstrated here, we use two electromagnets to aggregate the magnetic beads and fix them to one side of the sample holder using one of the electromagnets. However, aggregating and fixing the beads can potentially be done using a single permanent magnet, which will eliminate the need for the two electromagnets [24] and significantly reduce the bulk and power consumption of the OMB system. Ongoing research focuses on combining aggregation of the magnetic beads using a small permanent magnet and modulating the signal using the OMB technology.
The background noise in fluorescence-based assays originates from unbound fluorescent molecules and Raman scattering of the solvent molecules [25]. In heterogeneous assays, to minimize the background noise originated from unbound fluorescent molecules, the protocol includes many washing and separation steps [16]. However, red-shifted photons from the Raman scattering of the solvent molecules remain a challenge. Here, by alternating the laser beam back and forth from the beads to the background solution, we remove background noise of both unbound fluorescent molecules and Raman scattering. The first measurement provides information about the total fluorescence, which is primarily due to the specific fluorescence of the magnetic beads with captured fluorophores. The second measurement provides information about the non-specific background fluorescence, which originates from the fluorescence of the unbound fluorescent molecules, the Raman scattering of the solvent molecules, and the dark current of the camera. By subtracting the average background noise from the average total fluorescence, we improve the SNR and the sensitivity of the assay. Moreover, the modulation eliminates the need for multiple washing and separation steps, and thereby significantly simplifies and shortens the protocol. Simply turning the laser beam on and off will not remove background photons that originate from the Raman scattering of the water molecules. In addition, it will not eliminate the need for washing and separation steps that remove the unbound fluorescent molecules.
The number of beads in the OMB assay was determined empirically based on the trade-off between the aggregation time and sensitivity. The aggregation time depends on the magnetic forces acting on the beads and the sensitivity depends on the number of fluorescent molecules per bead. The magnetic force acting on a magnetic bead is proportional to the bead's magnetic moment and the gradient of the magnetic field. Once two or more beads aggregate, their combined magnetic moment increases proportionally to the number of aggregated beads, and subsequently so does the magnetic force acting on them [26]. In contrast, the drag force, which is proportional to the radius of the clump of beads, increases much less. Thus, increasing the number of beads in the assay will shorten the aggregation time, but will reduce the number of fluorescent molecules per bead.
Signal averaging increases the strength of a signal relative to noise that is obscuring it. By averaging a set of replicate measurements, the signal-to-noise ratio (SNR) will be increased [27]. In both MMB and OMB, the signal from the fluorescent molecules that are attached to the magnetic beads is averaged by exciting at least 2800 beads simultaneously. The simultaneous excitation of multiple beads is achieved by aggregating the beads (∼2.8 µm in diameter) inside a much larger excitation laser beam (150 µm in diameter).
In MMB, due to the magnetic manipulation, the beads are mixed in the sample, and whenever they pass through the laser beam, different regions of the beads' surface are exposed to the light. In comparison, in the OMB system, the beads are fixed to one position, and therefore whenever the laser beam is on the beads, it excites only one region of the beads, which may generate side effects. For example, due to continuous exposure of one region of the beads to the laser beam, the fluorescent molecules that are attached to the beads at that location can become photobleached (Fig. 4). In addition, at low concentrations, the fluorescent molecules are not homogenously spread on the magnetic beads' surfaces (i.e., the fluorescent molecules can be concentrated on regions of the beads that are not exposed to the light). Hence, images of one specific region of the beads surfaces may not correctly represent the number of fluorescent molecules on the beads. Photobleaching can be avoided by increasing the laser beam manipulation frequency, thereby shortening the time the laser beam illuminates the magnetic beads. Here, by increasing the modulation frequency from 2 Hz to 10 Hz, the signal drop between the first and fourth cycles was significantly reduced. It should be noted that the maximum laser manipulation frequency is limited by the frame rate of the camera to several hundred hertz. Here, for all modulation frequencies, we used a frame rate of 50 fps. Increasing the frame rate would have resulted in additional data points in each illumination cycle. Moreover, by increasing the modulation frequency the data acquisition from several modulation cycles can be completed within less than 10 milliseconds.
In both OMB and MMB, an accurate estimation of I signal , the net fluorescence intensity of the fluorophores that are attached to the beads, assumes that the average noise generated when the laser beam illuminates the beads equals the average noise generated when the laser beam illuminates the background solution. However, when the laser beam illuminates the beads, a fraction of the laser intensity is backscattered or absorbed and does not reach the background solution. Hence, it is possible that the Raman scattering of solvent molecules (I R ) and the residual fluorescence from unbound fluorescent molecules or other impurities (I UF ) are not the same for the case in which the laser beam illuminates the background solution (Eq. (1)) and the case in which the laser beam illuminates the beads (Eq. (2)). These differences can be minimized by decreasing the depth of the detection area. A shallower liquid layer is characterized by lower levels of Raman scattering (I R ) and fewer unbound fluorescent molecules (I UF ). Consequently, their relative contribution to the overall fluorescence of the sample is reduced.
Several optical and microfluidic devices have been developed for detection of biological analytes. For example, the Luminex xMap technology also utilizes a magnetic bead-based assay that can be used with either a flow-cytometry device (e.g., Luminex 200, BioRad Bio-Plex, FLEXMAP 3D) or a magnetic capture-based device (e.g., MAGPIX). The xMap detection technology enables detection of up to 500 analytes within the same well. However, to simultaneously detect multiple analytes, the magnetic beads have to be pulled into the device and examined individually. Thus, these devices include a complicated fluidic handling system, are expensive, bulky, and require frequent and costly maintenance and calibration procedures.
In OMB, the magnetic beads are aggregated from the entire sample volume into the detection area. Thus, the fluorescent signal increases by orders of magnitude. On the one hand, aggregating the beads and examining them as a clump, eliminates the need for a complicated fluidic system, and thereby significantly reduces the cost and complexity of the system. On the other hand, it limits the number of analytes that can be detected in one well.
Compared with the highly multiplex systems, the primary use of the OMB system is to detect an analyte of interest at very low concentrations using a compact and cost effective design. Here, the LoD of the OMB system is 172 fluorescent molecules (Atto 532) per bead. In comparison, the declared LoDs of the MAGPIX and the Luminex 200 are 700 and 1000 fluorescent molecules (phycoerythrine) per bead. It should be noted that compared to phycoerythrine, Atto 532 has much lower extinction coefficient (∼115,000 M −1 cm −1 vs. ∼1,960,000 M −1 cm −1 ) and similar quantum yield (0.9 vs. 0.82). Thus, the LoD of the OMB system is at least 4-6 times better than the MAGPIX and the Luminex 200.

Conclusions
We present a novel optical modulation biosensing system that can rapidly detect and quantify low concentrations of biomarkers. Using Il-8 as our target molecule, we demonstrated a 0.02 ng/L LoD and 4-log dynamic range. In addition, we detected anti-Zika IgG antibodies in clinical samples with much higher clinical sensitivity (96%) than the state-of-the-art EUROIMMUN ELISA (76%). This level of performance is comparable to that of the most sensitive devices but is achieved without their bulk and cost. Moreover, the data acquisition time of the OMB system is ∼250 milliseconds, much shorter than the ∼12 seconds of the MMB system, and therefore the OMB technology is less susceptible to the photobleaching effect and is much more suitable for high throughput applications. Future research will focus on combining the optical beam modulation with a permanent magnet for rapid detection of target molecules in a conventional 96-well plate.