Characterization of a Single-Molecule Sensitive Digital Flow Cytometer for Amplification-Free Digital Assays

Abstract

Digital assays such as digital PCR for nucleic acids and digital ELISA for proteins provide absolute quantitation and greater accuracy, sensitivity, and reproducibility than their analogue counterparts (real-time PCR and standard ELISA), but current digital assays involve amplification (e.g., DNA amplification in digital PCR and signal amplification in digital ELISA), which makes high multiplexing difficult, often requires complex and expensive sample compartmentalization, and adds reaction steps. We have developed a single-molecule sensitive flow cytometer, which we termed a digital flow cytometer (dFC). dFC optimizes the sensitivity and efficiency of single-molecule detection by using smaller, planar microfluidic channels, a smaller probe volume, and a shorter working distance/higher numerical aperture objective than used in current commercial high-sensitivity flow cytometers, allowing digital assays via direct single-molecule counting. This paper describes our characterization of the analytical performance of this system when detecting antibody-dye conjugates and demonstrates absolute concentration measurements of commercial antibody-dye conjugates. The dFC exhibited a single-molecule detection efficiency with which over 98% for antibodies conjugated with 18 different small-molecule, phycobiliprotein, and semiconducting polymer dyes were separated from noise, a low false-positive rate, a stable baseline signal, and accurate concentration measurements with a dynamic range spanning 4 orders of magnitude. This system can be used for authenticating antibody-dye conjugates used in flow cytometry and tissue imaging studies and in the development of multiplexed, amplification-free digital assays for nucleic acids and proteins.

Keywords: absolute quantitation; antibody QC/QA; digital assay; digital flow cytometer (dFC); single-molecule counting.

Figure 1.. Single-molecule detection of 18 antibody-dye conjugates.

(A) Raw photon burst traces of 18 antibody (IgG1)-dye conjugates in 12 color channels over a period of 80 ms. Six antibody-dye conjugates (IgG1 labeled with BV421, SBV440, BV480, SBV515, BV605, and SBV710) were excited by a 405-nm laser (top row, purple traces). Four conjugates (IgG1 labeled with AF 488, SBB615, BB700, and SBB700) were excited by a 488-nm laser (lower left, cyan traces). Four conjugates (PE, PE-Cy5, SBY665, and PE-Cy7 were excited by a 561-nm laser (lower middle, yellow traces). Four conjugates (IgG1 labeled with AF647, APC, APC-Cy7, and APC-Fire750) were excited by a 637-nm laser (lower right, red). For each conjugate, emission was collected by using optimal laser power and an acquisition rate of 10 kHz. Each peak represents a single fluorescence emission event from a single molecule. The peak-finding threshold was determined for each dye based on the background signal from the negative control sample and on a log-normal fit of the raw data following background subtraction as described in the Methods. (B) Corresponding SNR distributions. For each conjugate, we analyzed the distribution of SNR at single-molecule level using a log-normal fitting model. The fitting results showed that the SNR followed a log-normal distribution – which is a common characteristic of measured fluorescence from small-molecule dyes, although there can be small deviations for the large polymer dyes and nanoparticle probes (e.g., SBV515) – and the high estimated SMDE values (>98% for all dyes).

Figure 2.. Stability of optical detection and microfluidic flow.

We counted the number of IgG1-AF647 molecules in a 5 pM solution in PBS with 0.1% BSA continuously for one hour and analyzed the average peak height (Avg. P.H.), baseline level, and frequency of detected events per minute. (A) The stable average peak height (red) and baseline level (blue) indicated no change in the efficiency of photon detection and a stable, low background. The event frequency (B) and flow rate (C) did not change significantly over one hour, indicated stable flow in the microfluidic chip.

Figure 3.. Accuracy, dynamic range, and repeatability of concentration measurements using dFC.

We performed serial dilutions to obtain different concentrations of the three analytes, and calculated expected concentrations using stock concentrations provided by the manufacturers. We performed seven independent replicates at each concentration to examine the repeatability of the measurements and overlaid the results (red circles). Linear regression of the serial dilutions indicated excellent concordance between the measured and expected results, with R² values of 0.9997, 0.9973 and 0.9999 for IgG-AF647 (A), IgG-AF488 (B), and 40-nm beads (C), respectively. (D) As an orthogonal validation, we measured the concentration of free R-PE and APC using their known extinction coefficients and absorbance at 561 and 640 nm, respectively, and found the results are highly consistent with those measured using dFC. All measurements were performed with seven independent replicates.

Figure 4.. Concentration measurements of antibody-semiconducting polymer dye and nanoparticle conjugates versus manufacturer values and in a mixture of conjugates.

(A) Measured concentrations of four IgG-Brilliant dye conjugates. The manufacturer-provided concentration of each of these conjugates was 1.3 μM. (B) Measured concentration of four IgG-StarBright conjugates before and after mixing the conjugates, demonstrating that mixing did not affect the measurements. Error bars indicate the standard deviation of the measurements from seven independent replicates.