Detection of viruses by counting single fluorescent genetically biotinylated reporter immunophage using a lateral flow assay

Abstract

We demonstrated a lateral flow immunoassay (LFA) for detection of viruses using fluorescently labeled M13 bacteriophage as reporters and single-reporter counting as the readout. AviTag-biotinylated M13 phage were functionalized with antibodies using avidin-biotin conjugation and fluorescently labeled with AlexaFluor 555. Individual phage bound to target viruses (here MS2 as a model) captured on an LFA membrane strip were imaged using epi-fluorescence microscopy. Using automated image processing, we counted the number of bound phage in micrographs as a function of target concentration. The resultant assay was more sensitive than enzyme-linked immunosorbent assays and traditional colloidal-gold nanoparticle LFAs for direct detection of viruses.

Keywords: bacteriophage; diagnostics; image processing; immunoassay; lateral-flow assay; virus.

Figures

Figure 1. Imaging lateral flow assay with FluorM13 reporters

Anti-M13 and anti-MS2 antibodies were hand-spotted on Fusion 5 strips to generate test and control lines, respectively. The MS2 viruses were dispensed onto distal end of strips that were then washed in buffer. The FluorM13 reporters were spotted on the strips to sandwich MS2 viruses at the test line. Fluorescence micrographs were acquired at the test line, transition, and control line and analyzed using automated-image-processing routines to count the FluorM13 reporters.

Figure 2. Modification of AviTag-M13

(a) AviTag-M13 was modified with AlexaFluor 555 on the p8 major coat proteins. (b) AviTag peptides of AviTag-FluorM13 were biotinylated on the p3 tail protein in vitro using biotin ligase. (c) Neutravidin was conjugated onto biotin of Biotin-FluorM13 phage. (d) Biotinylated anti-MS2 antibody was attached to Neutravidin-FluorM13 through biotin-avidin conjugation.

Figure 3. Characterization of modified FluorM13 reporters

(a) Fluorescence micrograph of FluorM13, demonstrating the ability to resolve individual reporter phage. (b) TMB-ELISA showing absorbance at 450 nm as a function of Biotin-FluorM13 concentration offered to bind on a streptavidin-coated plate. (c) TMB-ELISA showing absorbance at 450 nm as a function of FluorM13 concentration to confirm anti-MS2 antibody retention on FluorM13. Error bars indicate standard deviations from triplicate measurements.

Figure 4. Characterization of modified FluorM13 reporters using magnetic particles

FluorM13 functionalized with biotin, neutravidin, anti-MS2 antibody were captured by anti-M13 antibody-magnetic particles (a, b and c). Neutravidin magnetic particles showed high binding affinity with Biotin-FluorM13 (d). The affinity between neutravidin-magnetic particles and anti-MS2 Ab FluorM13 (f) was ascribed to excess biotins on the biotinylated anti-MS2 antibody based on the result of the HABA assay. The third (g, h and i) and fourth (j, k and l) rows showed that biotin-FluorM13 undergo successful neutravidin and anti-MS2 Ab conjugation, respectively. The bar graphs showed the fraction of fluorescence magnetic particles bearing bound FluorM13, corresponding to the representative micrographs. Error bars were standard deviations from twenty images analyzed for each pair of magnetic particles and modified FluorM13. All images were acquired with identical imaging conditions (camera gain = 8, camera exposure time = 0.3 sec, 100× objective lens).

Figure 5. Automated counting of individual FluorM13 reporters

(a) A fluorescence micrograph was acquired at the LFA test line. (b) Histogram equalization was used to enhance image contrast, allowing FluorM13 reporters to be clearly visible. (c) The Beltrami flow algorithm was used to reduce and smooth background noise and increase the ratio of signal to noise, while preserving the shape of the FluorM13 reporters. (d) A rank-leveling algorithm was used to remove uneven illumination. (e, f) Global and local thresholds were used to segment FluorM13 reporters from the background. (g) Brightfield micrograph acquired at the same focal area as in (a) of the LFA test line. (h) A connected component labeling algorithm was used to automatically count FluorM13 reporters. The segmented M13 phage reporters were overlaid onto the brightfield micrograph. (i) FluorM13 identified by the algorithm were indicated by axis-aligned bounding boxes. The scale bar for all images is 10 μm.

Figure 6. Representative micrographs from imaging LFAs for MS2 virus detection

For each concentration of MS2/strip, two fluorescence micrographs acquired at the control (top; saturated white), transition (middle), and test (bottom) lines are shown. Each micrograph was of dimension 130 × 130 Jm2 and the area of the antibody spotted onto Fusion 5 strips was ~ 1 mm2, so that at most sixty micrographs could be acquired from each antibody spot at a two dimensional plane. In the assay, twenty micrographs were acquired at the control, transition, and test lines of an LFA strip. We observed high fluorescence intensity at the control lines, leading to the bright white images. Although individual FluorM13 reporters cannot be distinguished at the control line, the high fluorescence intensity showed that sufficient FluorM13 reporters migrate along the length of the strip. The number of FluorM13 reporters bound at the test line increased with increasing number of MS2 per strip, and remained invariant at the transition area.

Figure 7. Limit of detection of LFA

Notch boxplot showing the number of FluorM13 reporters per micrograph as a function of number of MS2 phage per strip at LFA test line and transition region (located at 5mm further downstream from the test line). Strips were loaded with 10 JL MS2 virus solution at various concentrations. FluorM13 reporters were counted from twenty micrographs acquired for each MS2 virus concentration using an automated counting program. All test lines of LFA strips were distinguishable from the corresponding backgrounds at a statistical significance level of p < 0.05 using the t-test. The position of the notches indicated the 95% confidence interval; notches for images acquired at the test line for each the nonzero MS2 concentrations did not overlap with that for zero MS2 concentration.