2.1.

Antibody and Virus Reagents

M13K07 phage and anti-M13 monoclonal antibody were used as a model system for this study. M13K07 is well characterized and can be handled with minimal safety concerns. M13K07 virus (M13) was obtained from the Vanderbilt Molecular Recognition and Screening facility (Nashville, Tennessee) in a stock concentration of $3.3\times {10}^{11}\mathrm{virions}∕\mathrm{ml}$ , and was diluted in phosphate buffered saline (PBS) containing 0.1% tween-20 (PBS-T) to working concentrations immediately prior to the experiments. Anti-M13 monoclonal ${\mathrm{IgG}}_{2\mathrm{a}}$ (anti-M13) and anti-E tag monoclonal ${\mathrm{IgG}}_{2\mathrm{a}}$ (anti-E) were obtained from Amersham Biosciences (Piscataway, New Jersey). Anti-M13 was used both as a capture antibody and a detecting antibody as described later. Anti-E was used as a nonspecific control for virus capture. Anti-M13 detecting antibody was made by fluorescently labeling anti-M13 antibody with Alexa Fluor 647 (AF647, Molecular Probes, Eugene, Oregon). Labeling procedures were performed according to the manufacturers’ instructions. The concentration of the labeled antibody along with the degree of labeling was calculated from the absorbance at $280\mathrm{nm}$ (IgG molar extinction coefficient: $213,000{\mathrm{cm}}^{-1}{\mathrm{M}}^{-1}$ ) and the peak label absorbance ( $650\mathrm{nm}$ , $239,000{\mathrm{cm}}^{-1}{\mathrm{M}}^{-1}$ ). Working solutions of both labeled and unlabeled antibody were stored at $4°\mathrm{C}$ , and for long-term storage aliquots were stored at $-20°\mathrm{C}$ .

2.2.

Antibody Immobilization on Filament

Unlabeled anti-M13 and anti-E capture antibodies were passively adsorbed to a clear polyester monofilament with a diameter of $120\mathrm{\mu }\mathrm{m}$ (Sulky Invisible, Punta Gorda, Florida). This monofilament was selected because of its low autofluorescence properties and its nonporous surface. This latter property permits rapid processing while avoiding filament swelling. The monofilament was wound around the ends of a PhastGel sample applicator (Amersham Biosciences) and placed within the concave teeth as described previously.¹³ The comb/filament apparatus was washed in 70% ethanol, rinsed in water, washed in 10% $\mathrm{HCl}$ , and rinsed again. The filament was dried and $0.75\mathrm{\mu }\mathrm{l}$ of capture antibody solution $(500\mathrm{\mu }\mathrm{g}∕\mathrm{ml})$ was pipetted onto each tooth of the comb. The comb and filament were incubated in a humidified box for $45\min$ to allow time for adequate adsorption of the capture antibody to the filament. Following incubation, the comb and filament were rinsed in PBS-T to remove unbound capture antibody from the filament. Filaments were used within one hour of the final rinse.

2.3.

Filament Processing

One-quarter-inch (OD) thick-walled glass capillary tubing with either 2-mm ID or 0.75-mm ID was cut into $75\mathrm{mm}$ lengths, and the ends were flared outward to facilitate smooth movement of the filament through the chambers. Table 1 summarizes the dimensions, contents, and function of each chamber. Chamber solutions were loaded manually using a pipette. Each glass chamber was housed in an aluminum holder with holes for two positioning bolts, as shown in the inset of Fig. 1. A horizontal aluminum stage with a matrix of predrilled holes was used to align the chambers. The spacing between chambers was approximately $1\mathrm{cm}$ . After a filament was threaded through the chambers and detector, it was attached to a rotating spindle atop the rotary stage. A small weight on the opposite end kept the filament under low tension and centered within the middle of each chamber (Fig. 1).

Table 1

Contents and description of each chamber.

Filament movement through the five chambers and detector was performed using a rotary stage and control system (Sigma II Servo System, Yaskawa, Waukegan, Illinois). The rotary stage was controlled through a LabView Virtual Instrument (National Instruments, Austin, Texas) that positioned the filament regions containing capture antibody within each chamber. The LabView interface controlled the speed of the filament through the chambers, oscillatory movements within the chambers, and the total residence time within each chamber. Overall distance from the first chamber through the detector was approximately $50\mathrm{cm}$ .

2.4.

Fluorescence Detection

After bioprocessing, a laser excitation source was used to excite the fluorescently labeled detecting antibody associated with antibody-virus complexes on the filament. The detection system incorporated two diode lasers at a $90\text{-}\mathrm{deg}$ orientation to the filament. The dual laser design allowed for flexibility when choosing appropriate fluorescent labels for detection. The filament was threaded through a detection chamber (Newport-Oriel, Stratford, Connecticut) that allowed easy coupling of two lasers and two photomultiplier tubes (PMT). Figure 1 shows a diagram of the virus detection scheme and its position relative to the bioprocessing chambers. Laser 1 (638-nm, 25-mW diode laser, Coherent, Santa Clara, California) was used to excite the AF647 fluorescent dye. Laser 2 (532-nm, 20-mW diode-pumped, solid state laser, B&W Tek, Incorporated, Newark, Delaware) was used to excite AF555 fluorescent dye, which was used primarily in preliminary experiments. The detector’s dual laser and dual PMT design permits two fluorescent labels to be used concurrently, adding flexibility to this technique. The lasers were attached to the detection chamber on either side using custom adaptors to fit the ports of the detection chamber. A polarizer was placed within the adaptor to reduce the 638-nm laser power to approximately $5\mathrm{mW}$ .

The PMTs were also attached via custom adaptors to the top and bottom of the chamber. Reflectance of the laser light from the filament was a significant source of noise, so the choice of emission filters was critical to achieving a high signal to noise ratio (SNR). For the AF647 channel, long-pass filters with cutoffs at $685\mathrm{nm}$ (Chroma, Rockingham, Verment) and $665\mathrm{nm}$ (Melles Griot, Rochester, New York) were combined to filter out reflected laser light. The AF555 channel combined a bandpass filter centered at $565\mathrm{nm}$ (30-nm bandwidth, Chroma) with a long-pass filter (570-nm cutoff, Melles Griot) to filter out reflected light. Filters were placed between the sample chamber and the PMTs as shown in Fig. 2 .

Fig. 2

Optical path of virus detection system. Using a $638\text{-}\mathrm{nm}$ laser excitation source, detection of filament fluorescence is possible using the appropriate emission filters. Exposure area on the filament is reduced by an excitation slit.

Figure 2 shows the optical path for the detector. A custom brass slit ( $\sim 1\mathrm{mm}$ wide) was placed in the laser path to minimize the area on the filament illuminated by the laser. After exciting bound detecting antibody, fluorescence emission from the detecting antibody passed through a pinhole that reduced much of the reflected laser light. The light then passed through a pair of emission filters, which removed reflected excitation light while allowing emitted light to pass through to the PMT (R928, Hamamatsu). Each PMT was powered by an 800-V signal (Pacific Instruments, Concord, California), and the resulting current was converted to voltage and amplified by a factor of ${10}^5$ by a transimpedance amplifier (Model 101C, UDT Instruments, Baltimore, Maryland). The amplified voltage (0 to $14\mathrm{V}$ ) was sampled by a digital acquisition board (BNC 2120, National Instruments) controlled through a LabView virtual instrument. Filament movement and signal sampling were synchronized using a single LabView interface.

2.5.

Experimental Parameters

Filament movement parameters were held constant throughout all experiments. Filament speed between chambers was maintained at $2\mathrm{cm}∕\mathrm{s}$ , and filament regions were oscillated a distance of $2\mathrm{cm}$ at $1\mathrm{cm}∕\mathrm{s}$ within each chamber for the requisite length of time. All filaments were blocked in chamber 1 for $15\min$ before experiments were started. All experiments used M13K07 virus ( $3.3\times {10}^{10}\mathrm{virions}∕\mathrm{ml}$ , unless otherwise noted) and used AF647 anti-M13 $(20\mathrm{mg}∕\mathrm{ml})$ for detection. Other bioprocessing components were the same in all experiments. SNR was defined as the average peak height of anti-M13 regions divided by the peak height of a negative control anti-E region. An unpaired t-test was used to compare average peak values with average anti-E values to determine statistical significance. P values less than 0.05 signified positive virus detection.

Basic system tests were performed to show the effectiveness of the optical system and to determine some standard parameters that helped optimize signal to noise (Table 2 ). These tests were designed to determine FARA SNR and to determine if laser exposure significantly bleached the label on the detecting antibody. Preliminary observations using the laser at full power $\left(25\mathrm{mW}\right)$ showed that laser exposure may decrease fluorescence on repeated scans. To show that this effect was minimized with a less powerful laser beam, multiple laser scans of two filaments were performed after a typical virus detection experiment. One set of scans used the full power 25-mW laser, while the other used a reduced power of approximately $5\mathrm{mW}$ . The change in SNR with each scan was then calculated. Incubation parameters for these system tests are summarized in Table 2.

Table 2

Incubation parameters for all experiments.

Once basic parameters were established, incubation times were shortened to determine the shortest assay time that resulted in positive virus detection. Virus and detecting antibody incubation times were shortened to one minute, and wash times were shortened to fifteen seconds. When this shortened assay time was established, experiments were performed to determine the limit of sensitivity of this technique and the minimum assay times required to detect lower virus concentrations. Virus concentrations from $3.3\times {10}^7$ to $3.3\times {10}^{10}$ were used for these experiments. Virus incubation time was increased until detection was achieved. A summary of these parameters is shown in Table 2.

Important aspects of this assay design are the online detection of fluorescence associated with virus capture and the continuous control of filament position. This combination suggests that feedback from the detector signal could be used to direct subsequent filament motion. In this assay, there are two methods to increase the fluorescence signal associated with virus capture. One approach is to increase the number of detecting antibodies associated with each virus-antibody complex on the filament. Assuming that the initial residence time within the detecting antibody chamber is initially suboptimal, increasing this time should increase the signal. In the first experimental design to evaluate this capability, a filament containing three capture antibody regions and one negative control region was incubated for $30\sec$ in the detecting antibody chamber and scanned. The capture-antibody region of this filament was repositioned in the detecting antibody chamber for an additional $30\sec$ and rescanned. This cycle was repeated four times.

The second approach to increase signal is to increase the number of antibody-virus complexes associated with the filament. This can potentially be done by increasing the residence time in the virus chamber. Therefore, in the second experimental test, the filament was repositioned back to the virus chamber. In this test, the first cycle of each experiment used a one-minute virus incubation that yielded only a weak signal. Following this initial measurement, the capture antibody region of the filament was repositioned within the virus chamber for additional incubation. After each additional virus incubation, processing within chambers 3, 4, and 5 was repeated. The filament was cycled through this process three times to show the capability of using on-line detection and filament motion to enhance the fluorescence signal. Table 2 summarizes incubation parameters for both repositioning experiments.

2.6.

Results

As described later, repositioning the regions of the filament containing the capture antibodies through the reaction chambers and detector in an effort to increase the signal was an important aspect of many experiments. One concern with this approach is the bleaching effect of laser illumination on the fluorescent molecules and on the antigen-antibody interactions. Therefore, we first verified that consecutive fluorescence measurements produced similar intensities. Effects of laser illumination were investigated by taking multiple scans of two filaments following two virus detection experiments. Each filament was prepared with three anti-M13 regions and processed as summarized in Table 2. As Fig. 3 illustrates, full-power laser exposure decreased the observed signal during repeated scanning. After five successive scans, the signal was reduced to $\sim 8\%$ of its initial value (black bars). We reduced the laser power to determine if this would prevent the drop in signal intensity for subsequent scans. A polarizer was used to reduce laser power from its full strength of $25\mathrm{mW}$ to approximately $5\mathrm{mW}$ . To further reduce laser exposure of the filament, the filament scanning speed was increased to $4\mathrm{cm}∕\sec$ and an excitation slit was placed in front of the laser. As shown by the gray bars in Fig. 3, a reduced laser power had a much lower effect. After five successive scans, fluorescence was still above 80% of the initial value. All subsequent results were obtained using the polarizer and excitation slit.

Fig. 3

Effects of laser power on scan repeatability. Repeated laser scans of virus detection filaments result in a significant signal drop for filaments using full laser power (black bars) but not with reduced laser power (gray bars).

Figure 4 shows that regions of the filament containing the capture antibody (anti-M13) produced strong fluorescence signals. In these experiments each $\sim 3\text{-}\mathrm{cm}$ length of the filament included one negative control anti-E region and three anti-M13 regions. After processing through the five chambers, strong fluorescence signal (SNR $51\pm 4.5$ ) was observed in regions of the filament containing immobilized anti-M13 capture antibody, indicating successful virus detection. The fluorescence observed in the region containing immobilized anti-E capture antibody is indistinguishable from the background. The lack of fluorescence in the anti-E region of the filament indicates that, as expected, antigen/antibody binding is a highly specific process and that no virus was attached to this region. Scanning a filament with both the integrated detector and using the previous method of a flatbed microarray scanner,¹³ we found that SNR increased by nearly a factor of 5.

Fig. 4

Representative filament scan showing photomultiplier output as a function of filament position. Laser scanning of the filament detects virus in all three capture antibody regions (anti-M13) with a SNR of approximately $51\pm 4.5$ . Negative control antibody region (anti-E) is not distinguishable from background. Arrows indicate the location of each immobilized antibody region along the filament.

In this system, higher virus concentrations could be detected with shorter virus incubation times. Figure 5 shows the virus incubation time required to detect virus of different concentrations along with the SNR achieved in each experiment. Detection of $3.3\times {10}^8\mathrm{virions}∕\mathrm{ml}$ was achieved in $40\min$ with a SNR of more than 3. Raising the virus concentration to $3.3\times {10}^9\mathrm{virions}∕\mathrm{ml}$ reduced the required virus incubation time to $10\min$ , while raising the SNR to more than 6. Concentrations of $3.3\times {10}^{10}\mathrm{virions}∕\mathrm{ml}$ , the highest concentration tested, reduced the required time even further to only $1\min$ and resulted in a high SNR of almost 10. All other reaction parameters remained constant for these experiments (Table 2), so the total assay times for these trials ranged from $41.5\min$ for the lowest concentration to only $2.5\min$ for the highest concentration. Since filament blocking can be performed ahead of time, blocking times were not included in these estimates.

Fig. 5

Virus incubation times required to detect different virus concentrations. A one-minute virus incubation detected $3.3\times 1010\mathrm{virions}∕\mathrm{ml}$ with a SNR of nearly 10. Concentrations below $3.3\times 108\mathrm{virions}∕\mathrm{ml}$ were not detectable with a 75-min virus incubation time. Asterisks indicate that a difference in fluorescence signal between capture antibody regions and regions without antibodies is statistically significant $(p<0.05)$ .

Online detection and control of filament position are unique aspects of FARA. In this approach, a weak signal could potentially be amplified by reprocessing the immobilized capture antibody regions through the reaction chambers. This aspect could be very important for a new antibody/antigen pair, for which optimal incubation times are not yet known, or for rapid detection applications in which time is critical. In this approach, a rapid detection protocol might be used to detect a high virus titer, but if the initial screen is negative, then a higher-sensitivity, slower detection protocol could be used. As an initial test of this strategy, experiments were performed to test the effects on the fluorescence of repositioning the capture antibody regions of the filament within the detecting antibody chamber. The top panel of Fig. 6 shows the increase in signal after the filament was cycled through the detecting antibody chamber three times for additional incubation. In this assay, signal strength increases by almost a factor of 4 as the filament region containing the capture antibodies is reincubated within the virus test solution.

Fig. 6

Effects of reprocessing on fluorescence. Signal increases as the capture antibody region of the filament is cycled back to the detecting antibody chamber (top) or the virus chamber (bottom) for additional incubation. Arrows indicate regions of immobilized antibodies. The lower sampling rate in the top panel resulted in a smoother curve. Virus concentration was $3.3\times {10}^{10}\mathrm{virions}∕\mathrm{ml}$ .

Next, a second test of this strategy was performed by repositioning the capture regions within the virus chamber for increasing incubation times. In these experiments, a short initial virus incubation time was used for virus detection, so that fluorescence intensity was initially low. After the initial processing was performed and the filament scanned, the filament was repositioned within the virus chamber for an additional incubation, followed by the standard processing steps in the subsequent chambers. The bottom of Fig. 6 shows the increases in signal for cumulative incubation times of 1, 5, and $10\min$ .

The filament used for this experiment contained an anti-E negative control region adjacent to three anti-M13 regions. Fluorescence increases approximately 4 to 5 fold for the anti-M13 regions from cycle 1 to 3, but there is almost no change in the control antibody region. In addition, a negative control experiment was performed to ensure that increases in fluorescence were a result of specific antigen/antibody interactions. This experiment used an equivalent filament and experimental parameters, but no virus was used for this assay. The lower panel of Fig. 6 (“no virus” traces) shows that after three cycles through the incubation chambers, fluorescence does not increase above background.

As a final test of the filament repositioning strategy, we compared the signal intensities obtained with a single pass through the reaction chambers to the intensities obtained with filaments cycled through three successive processing cycles. As Fig. 7 indicates, the reincubation strategy produced signal intensities very similar to intensities of filaments with continuous incubation times when cumulative virus incubation times were matched. This indicates that filament motion and additional processing steps do not reduce the observed signal. Furthermore, it suggests that by using this system, flexible adaptive processing strategies utilizing repositioning of the filament after fluorescence measurements can be developed.

Fig. 7

Comparison of multiple short virus incubations (black bars) to one long virus incubation (gray bars). Virus cumulative incubation times of 1, 5, or $10\min$ resulted in similar fluorescence for a single filament exposed multiple times or for separate filaments each exposed for the same cumulative time indicated. Virus concentration was $3.3\times {10}^{10}\mathrm{virions}∕\mathrm{ml}$ .