Multiplexed single molecule immunoassays

Abstract

We have developed a method that enables the multiplexed detection of proteins based on counting single molecules. Paramagnetic beads were labeled with fluorescent dyes to create optically distinct subpopulations of beads, and antibodies to specific proteins were then immobilized to individual subpopulations. Mixtures of subpopulations of beads were then incubated with a sample, and specific proteins were captured on their specific beads; these proteins were then labeled with enzymes via immunocomplex formation. The beads were suspended in enzyme substrate, loaded into arrays of femtoliter wells--or Single Molecule Arrays (Simoa)--that were integrated into a microfluidic device (the Simoa disc). The wells were then sealed with oil, and imaged fluorescently to determine: a) the location and subpopulation identity of individual beads in the femtoliter wells, and b) the presence or absence of a single enzyme associated with each bead. The images were analyzed to determine the average enzyme per bead (AEB) for each bead subpopulation that provide a quantitative parameter for determining the concentration of each protein. We used this approach to simultaneously detect TNF-α, IL-6, IL-1α, and IL-1β in human plasma with single molecule resolution at subfemtomolar concentrations, i.e., 200- to 1000-fold more sensitive than current multiplexed immunoassays. The simultaneous, specific, and sensitive measurement of several proteins using multiplexed digital ELISA could enable more reliable diagnoses of disease.

Figure 1.

Schematic of the multiplexed digital ELISA process.

Figure 2.

Histograms of fluorescence intensity from a sub-region of images of a representative array acquired at excitation/emission wavelengths of: A) 490 nm/530 nm; B) 622 nm/667 nm; and, C) 740 nm/800 nm. The black lines are experimental data; the colored dotted lines are Gaussian fits to these data to identify subpopulations. The vertical solid colored lines are the upper and lower limits for each population identified. For each image, the first peak corresponds to empty wells or wells containing beads not fluorescent in that channel. In image A, the second peak corresponds to beads that are not labeled with AF-488; the third peak corresponds to beads labeled with AF-488. In B, the second peak corresponds to the low level of cy5 labeled beads, and the third peak to the second, higher level of cy5-labeled beads. In C, the second peak corresponds to HF-750-labeled beads.

Figure 3.

Images showing optical scatter that can lead to false positive “on” beads. In this experiment, IL-6 beads labeled with AF-488, and TNF-α, IL-1α, and IL-1β beads that were not labeled with AF-488, were incubated with a high concentration of IL-6 (100 pg/mL) but no other cytokines. Portions are shown of the Simoa images at: A) 490 nm/530 nm ex/em to identify beads; and B) at 574 nm/615 nm ex/em to detect enzyme activity. The bright beads in A correspond to the IL-6 beads; the dim beads correspond to the other cytokines. The red boxes show situations where IL-6 beads associated a high number of enzymes (AEB = 8) were imaged alongside beads not associated with an enzyme. The image in B shows that optical scatter from the high AEB beads can scatter into the wells containing “off” beads. The contrasts in these images were adjusted so the eye can see the optical scatter (~1–2%) that occurs: the scaling was set to 6% of the total dynamic range of the camera (black point = 241; white point = 1,228; maximum camera signal = 16,383).

Figure 4.

Representative images of an array from multiplexed digital ELISA at: A) & E) 574/615 nm ex/em; B) 490/530 nm ex/em; C) 622/667 nm ex/em; D) 740/800 nm ex/em.

Figure 5.

Plots of AEB against protein concentration for 4 beads specific to 4 cytokines measured in bovine serum samples spiked with: A) all 4 cytokines; and B) only TNF-α. The plots for samples spiked with only IL-6, only IL-1α, and only IL-1β are shown in Supplementary Fig. 3.