Expanding the Dynamic Range of Fluorescence Assays through Single-Molecule Counting and Intensity Calibration

Abstract

Surface capture assays can measure fluorescently labeled analytes across a 1000-fold concentration range and at the sub-nanomolar level, but many biological molecules exhibit 1,000,000-fold variations in abundance down to the femtomolar level. The goal of this work is to expand the dynamic range of fluorescence assays by using imaging to combine molecular counting with single-molecule calibration of ensemble intensities. We evaluate optical limits imposed by surface-captured fluorescent labels, compare performances of different fluorophore classes, and use detector acquisition parameters to span wide ranges of fluorescence irradiance. We find that the fluorescent protein phycoerythrin provides uniquely suitable properties with exceptionally intense and homogeneous single-fluorophore brightness that can overcome arbitrary spot detection threshold biases. Major limitations imposed by nonspecifically bound fluorophores were then overcome using rolling circle amplification to densely label cancer-associated miRNA biomarkers, allowing accurate single-molecule detection and calibration across nearly 5 orders of magnitude of concentration with a detection limit of 29 fM. These imaging and molecular counting strategies can be widely applied to expand the limit of detection and dynamic range of a variety of surface fluorescence assays.

Figure 1.

Optical properties of avidin-conjugated fluorophores. (a) Extinction coefficient (ε) spectra and fluorescence emission spectra for avidin conjugates of four fluorophores. From left to right: dye, QD, PE, and bead. For each class, a schematic depiction shows avidin (gray) and the fluorophore in proportion to their relative sizes. Inset values show quantum yield (Φ) and relative brightness (B). (b) Representative single-molecule images of sparse labels on coverglass. (c) Spot detection counts are shown based on P_FA detection thresholds for bare coverglass (gray) and those with several hundred fluorophores per FOV (black). The red line shows the difference between the two at each threshold value. Data are averages from >10 FOVs, and error bars indicate standard deviation. (d) Slopes of red curves from panel (c) demonstrate the stability of spot counts depending on chosen P_FA threshold. (e) Photostability for the four different fluorophores, showing measured intensity over time of continuous excitation at 561 nm with irradiance of 24.07 W/m². (f) Spot intensities for indicated fluorophores (red) using a 3 × 3 voxel compared with background values (blue). (g) Receiver operating characteristic curve for fluorophore spots measured using a 3 × 3 voxel. For PE and bead measurements, the area under the curve is unity.

Figure 2.

Tuning image acquisition conditions to normalize intensities. (a) Dependence of measured mean pixel intensity (I_p) on detector gain (G) for five different surface densities of QDs, indicated by the calibrated pixel intensities (I_p,c) for G^o = 450. (b) Calibrated intensity per pixel (I_p,c) after applying linear gain factors in eqs 1 and 2. (c) Impact of gain on image signal relative to surfaces without fluorophores. (d−f) Same analysis using integration times to calibrate image intensity. Calibrated pixel intensities are shown for t_int^o = 1000. All data points include standard deviation error bars.

Figure 3.

Single-molecule intensity calibration to extend fluorescence assay range. (a−d) Digital spot counts (red) and calibrated spot counts from analog intensity measurements (blue) are shown as points at the indicated avidin−fluorophore conjugate concentration, showing (a) dyes, (b) QDs, (c) PE, and (d) beads. The black line is a fit connecting analog data points that are above the analog limit of detection (horizontal blue shading) to digital points above the digital limit of detection (horizontal red shading). Dashed lines indicate the theoretical values if all solution fluorophores were detected, and dash-dotted lines indicate spot counts due to nonspecific labeling for surfaces without biotin, extrapolated from data in Supplementary Figure S7. (e) Spot intensity histograms for PE, miRNA labeled with PE through linear extension, and no PE background. (f) Spot intensity histograms for PE, miRNA labeled with PE through rolling circle amplification (RCA) extension, and no PE background. (g) Spot counts based on P_FA thresholds for bare coverslips (gray) and those with several hundred RCA-extended miRNA per FOV (black). Both were labeled with avidin−PE conjugates. The red line shows the difference between the two at each threshold value. Data are averages from >10 FOV, and error bars indicate standard deviation. (h) Digital spot counts and calibrated spot counts from analog intensity measurements for PE-labeled RCA products of miRNA, using the same notations as used in panel (a).

Scheme 1.

Imaging and Analysis Workflow