Direct digital sensing of protein biomarkers in solution

Abstract

The detection of proteins is of central importance to biomolecular analysis and diagnostics. Typical immunosensing assays rely on surface-capture of target molecules, but this constraint can limit specificity, sensitivity, and the ability to obtain information beyond simple concentration measurements. Here we present a surface-free, single-molecule microfluidic sensing platform for direct digital protein biomarker detection in solution, termed digital immunosensor assay (DigitISA). DigitISA is based on microchip electrophoretic separation combined with single-molecule detection and enables absolute number/concentration quantification of proteins in a single, solution-phase step. Applying DigitISA to a range of targets including amyloid aggregates, exosomes, and biomolecular condensates, we demonstrate that the assay provides information beyond stoichiometric interactions, and enables characterization of immunochemistry, binding affinity, and protein biomarker abundance. Taken together, our results suggest a experimental paradigm for the sensing of protein biomarkers, which enables analyses of targets that are challenging to address using conventional immunosensing approaches.

Fig. 1. Working principle of DigitISA and its implementation.

a Schematic illustration of the DigitISA platform and overview of the experimental workflow. DigitISA integrates electrophoretic separation and single-molecule detection in a platform for single-step sensing of target proteins in solution using only a single affinity reagent. The sample including a mixture of the target protein and its fluorescently labeled affinity probe (e.g., aptamer) is injected into a micron-scale electrophoretic separation unit. The application of an electric field allows protein-bound probe molecules to be discriminated from those probe molecules that are not bound to the protein target, owing to a difference in their electrophoretic mobilities. Confocal scanning across the separation chamber is performed and the number of molecules traversing the confocal volume at each of the scanned positions is estimated ‘digitally’ from the recorded photon-count time trace using a combined inter-photon time (IPT) and photon-count threshold burst-search algorithm. From the obtained counts, an electropherogram is created allowing for a discrimination between protein-bound affinity probe and free probe. b Design of the free-flow electrophoresis device. Sample is flown into the microfluidic chip by the central injection port where it is then surrounded by the carrier buffer solution. The electrophoresis chamber is connected to a co-flowing electrolyte solution (3 M KCl) via bridges, which allows for a narrow sheet of electrolyte to flow along the sides of the chamber. An electric field is applied from metal clips at the outlets of the electrolyte channels, which propagates along the electrolyte sheet and enables separation of molecules perpendicular to the flow direction. c Schematic of the confocal microscopy setup used for single-molecule detection. A diode laser is used to excite the sample through an objective, and a single-photon counting avalanche photodiode (APD) is used to register emitted photons from the sample. The confocal volume is moved across the cross-section of the chip in a stepwise manner with the aid of a motorized stage. This allows the flux of the protein-bound probe molecules to be estimated. Details of the setup are described in the Methods section (Data analysis).

Fig. 2. Sensing of a biotin–streptavidin complex using the DigitISA platform.

a Monovalent streptavidin is added to a biotinylated and fluorophore-conjugated DNA strand. Binding of the streptavidin species reduces the electrophoretic mobility of biotinylated DNA probe. b Electropherogram as obtained by scanning the confocal volume across the cross-section of the channel in a stepwise manner for the biotin–streptavidin sample mixture (red line; average of N = 3 repeats, the shaded bands correspond to the standard deviation) and for the control sample (blue line; average of N = 3 repeats, the shaded bands correspond to the standard deviation) at the mid-height of the channel, demonstrating the presence and separation of both streptavidin-bound and nonbound biotinylated DNA molecules in the sample. The region shaded in grey was used to extract the number of streptavidin–biotin complexes that passed the device in a given time, and ultimately, its concentration (see main text). c Exemplary photon-count time traces for the control sample (left panel, blue) and the sample mixture (right panel, red) at the position where the concentration of the complex molecules was the highest as indicated with colored triangles in panel b. The number of molecules was estimated using a burst-search algorithm as detailed in the Methods section (Data analysis). Time traces in the upper panels are zoom-in views of the blue or red shaded areas in the lower panels, with dots indicating detected single-molecule events. The bin time was 1 ms in all traces. Source data are provided as a Source Data file.

Fig. 3. DigitISA immunosensing of IgE with an aptamer probe.

a An IgE-aptamer probe is added to its target IgE. Binding of the antibody reduces the electrophoretic mobility of the probe, allowing for fast electrophoretic separation of the aptamer probe from the immuno-complex for subsequent confocal detection. b Electropherogram as obtained by scanning of the confocal volume across the cross-section of the channel in a stepwise manner for the IgE–aptamer sample (orange line; average of N = 3 repeats) and for the free aptamer probe (green line; average of N = 3 repeats). The shaded bands correspond to the standard deviation. The region shaded in grey was used to quantify the concentration of IgE by monitoring the flux of fluorescent molecules in the region shaded in grey (see main text). c Exemplary photon-count time traces for the control sample (left panel, green) and the sample mixture (right panel, orange) at the position where the concentration of the complex molecules was the highest as indicated with colored triangles in panel b. The number of molecules was estimated using a burst-search algorithm as detailed in the Methods section (Data analysis). Time traces in the upper panels are zoom-in views of the green or orange shaded areas in the lower panels, with dots indicating detected single-molecule events. The bin time was 1 ms in all traces. Source data are provided as a Source Data file.

Fig. 4. Sensing of α-synuclein fibrils at high aptamer probe concentrations.

a Binding of an α-synuclein aptamer to fibrils reduces the electrophoretic mobility of the aptamer probe, allowing for the discrimination between fibril-bound and unbound aptamer species. b Electropherogram (left panel) as obtained by scanning of the confocal volume across the cross-section of the channel in a stepwise manner for the α-synuclein fibrils–aptamer sample (blue line; average of N = 3 repeats) and for the free aptamer probe (purple line; average of N = 3 repeats). The shaded bands correspond to the standard deviation. The right panel shows a zoom-in region of the electropherogram. The shaded region in grey where the concentration of the complex exceeded that of the free probe (1150 µm < x < 2000 µm) was used to estimate the concentration of the fibrils (see main text). Note, the photon arrival frequency in the region where x < 1100 µm (i.e., where the probe elutes) was too high to count molecules one-by-one, hence the detected number of molecules in this region should be viewed as an approximation. c Exemplary photon-count time traces for the control sample (left panel, purple) and the sample mixture (right panel, blue) at the position indicated with colored triangles in panel b (right). The number of molecules was estimated using a burst-search algorithm as detailed in the Methods section (Data analysis). Time traces in the upper panels are zoom-in views of the purple or blue shaded areas in the lower panels, with dots indicating detected single-molecule events. The bin time was 1 ms in all traces. Source data are provided as a Source Data file.

Fig. 5. Sensing and immunochemical analysis of α-synuclein oligomers.

a The binding of an aptamer to wild-type α-synuclein oligomers reduces the electrophoretic mobility of the probe, allowing for the discrimination between oligomer-bound and unbound aptamer species. b Electropherogram as obtained by scanning of the confocal volume across the cross-section of the channel in a stepwise manner for the α-synuclein oligomer–aptamer sample (blue line; average of N = 3 repeats) and for the free aptamer probe (orange line; average of N = 3 repeats). The shaded bands correspond to the standard deviation. The main panel depicts the region where single-molecule counting of the probe or oligomer–probe complex is possible. This region was used to estimate the concentration of oligomers in the sample. The inset shows the full electropherogram and the average photon intensity at each step location, with the region shaded in grey depicting the region shown in the main panel. c Exemplary photon-count time traces for the control sample (orange) and the sample mixture (blue) at the position indicated with colored triangles in panel b. The number of molecules was estimated using a burst-search algorithm as detailed in the Methods section (Data analysis). Time trace in the upper panels is a zoom-in view of the blue shaded area in the lower panel, with dots indicating detected single-molecule events. The bin time was 1 ms in all traces. d Burst intensities of the aptamer–oligomer (blue) and free aptamer (orange) detection events across the single-molecule detection region of the channel. Both the mean and median intensities are reported. Each line is the average of N = 3 repeats, the shaded bands correspond to the standard deviation. e Average burst intensities of the free aptamer (orange) and aptamer–oligomer sample (blue) across the single-molecule detection region of the channel (see panel d). Both the mean and median intensities are reported. Error bars denote the standard deviation of N = 3 repeats of all data points across the single-molecule detection region of the channel (see panel d). Data points are shown. Source data are provided as a Source Data file.

Fig. 6. DigitISA immunosensing of an exosome biomarker.

a Shown is a CD63-specific aptamer bound to CD63 proteins on the surface of the exosomes. The exosome-bound aptamers have a lower electrophoretic mobility than the free aptamers, allowing for the discrimination between exosome-bound and unbound species. b TEM micrographs (left panels) and Western blots (right panels) of exosomes. Exosomes display their characteristic cup-shaped morphology and can be identified by their size (100–200 nm in diameter) and the presence of a lipid bilayer membrane. TEM experiments were repeated three times. Scale bars: 500 nm (top left), 100 nm (bottom left). Western blot (right panels) comparing exosome samples (Exo) and cell lysate sample (CL). CD9 and CD63, both exosome marker proteins, are highly enriched in the exosome sample, while calreticulin (Cal) is excluded from the exosomes, indicating their purity. Western blot experiments were repeated three times. c Electropherograms as obtained by scanning of the confocal volume across the cross-section of the channel in a stepwise manner for the 50 nM exosome–aptamer sample (magenta line; average of N = 3 repeats) and for the free aptamer sample (cyan line; average of N = 3 repeats). The shaded bands correspond to the standard deviation. The light lines with error bands represent the average fluorescence intensity. The solid markers with error bars indicate the number of single-molecule detection events at each coordinate with the error bar representing the standard deviation. The region shaded in grey was used to estimate the concentration of exosomes in the sample. d Exemplary photon-count time traces for the control aptamer-only sample (cyan) and the exosome mixture (magenta) at the position indicated with colored triangles in panel c. Individual exosomes were detected and quantified using a burst-search algorithm as detailed in the Methods section (Data analysis). Dots above the time traces indicate individual exosome detection events. The bin time was 1 ms in all traces. e Mean concentrations of bound exosomes and averaged median binding stoichiometries at three different concentrations of aptamer. Error bars denote the standard deviation of N ≥ 3 repeats. Data points are shown. Source data are provided as a Source Data file.

Fig. 7. DigitISA sensing of protein condensates.

a A low affinity aptamer for FUS protein can be found in three different states: unbound, with high electrophoretic mobility, bound to monomer protein, or partitioned into FUS condensates, the latter two exhibiting lower electrophoretic mobilities. b Fluorescently labelled GGUG aptamer is added to unlabeled FUS, resulting in the formation of biomolecular condensates. Image was taken on a widefield fluorescence microscope. Experiment was repeated three times. Scale bar is 10 µm. c Electropherogram as obtained by scanning of the confocal volume across the cross-section of the channel in a stepwise manner for the FUS–aptamer sample (blue line; average of N = 3 repeats, the shaded bands correspond to the standard deviation) and for the free aptamer probe (magenta line). Note, the ordinate shows the average fluorescence intensity, which was used to calculate the ratio between free and FUS monomer-bound aptamer (see main text). In the top panel, a histogram of the number of condensates detected at each channel position is shown, which was used to calculate condensate number and concentration. Error bars denote the standard deviation of N = 3 repeats. d Exemplary photon-count time traces for the control sample (magenta) and the condensate mixture (blue) at the position indicated with colored triangles in panel c (bottom). The number of condensates was quantified by counting bursts above a relative intensity threshold (shown by the dashed line) unique to each scan position, a process further detailed in the Methods section (Data analysis). Dots above the time traces indicate detected single-condensate events. The bin time was 1 ms in all traces. Source data are provided as a Source Data file.

Fig. 8. Comparison of the DigitISA assay with surface-based sensing methods.

a Schematics depicting the principles of DigitISA and its advantages over surface-based methods. By performing the sensing reaction in solution, high concentrations of the affinity probe can be used, which permits quantitative target binding, even for probes with dissociation constants K_d > 1 nM (c(i)). Rapid removal of nontarget bound probe by electrophoresis prevents the system from re-equilibrating (c(ii)) and sets the basis for quantitative analysis. The assay can be accomplished in a single step and requires only a single affinity reagent. Due to the digital nature of detection, absolute number–concentration quantification is possible, and additional parameters such as immunochemistry, stoichiometry, partitioning ratios and dual affinities can be extracted. DigitISA also allows for the analysis of complex and transient systems. b Schematic of surface-based immunosensor assays and their limitations. Conventional methods are limited to surface-capture probe concentrations (c_probe) in the low nanomolar regime. Under these conditions, a significant amount of the analyte is not bound and thus remains undetected, especially when using affinity probes with K_d > 1 nM (c(i)). The binding equilibrium is disturbed during long washing steps. Hence, dissociation of the complex becomes significant, particularly for low affinity probes with K_d > 1 nM (c(ii)). Conventional assays involve multi-step procedures and typically require additional probes for detection and specificity. Detection modes require calibration to convert the signal to target concentration. The information content on target binding interaction is low and only targets that do not interfere with the surface are assayable. c Speciation curves depicting fraction of probe-bound analyte versus affinity probe concentration (c(i)) and fraction of probe-bound analyte versus time (c(ii)). Curves for probe–analyte affinities with K_d = 0.1–1000 nM (from light to dark blue) are shown. Green and red shaded areas denote operation regimes. d The multimodal detection capabilities of DigitISA augment the information content from sensing experiments beyond what is achievable and assayable with classical techniques. The nature of systems that can be studied are further expanded with DigitISA, enabling the study of transient systems such protein oligomers and liquid condensates.