An electrochemical sensor for the detection of protein-small molecule interactions directly in serum and other complex matrices

Abstract

Here we have demonstrated a general, sensitive, and selective approach for the detection of macromolecules that bind to specific small molecule recognition elements. Our electrochemical approach utilizes a redox-tagged DNA signaling scaffold that is conjugated to a small molecule recognition element and is covalently attached to an interrogating electrode. The binding of a protein to the small molecule recognition element alters the dynamics of the scaffold, increasing or decreasing the efficiency with which the redox tag collides with the electrode and thus altering the observed faradaic current. We optimized the scaffold using a biotin recognition element and streptavidin as a target to determine the variables that define sensor performance before then applying the approach to detection of anti-digoxigenin antibodies using the steroid as the recognition element. We generated streptavidin sensors exhibiting both signal-on (target binding increases the faradaic current) and signal-off behavior, of which only the signal-off approach was generalizable to the detection of antibodies. Sensors for both targets are sensitive (detection limits in the low nanomolar range), rapid (minutes), reusable, and selective enough to function directly in complex matrices including blood serum, soil, and foodstuffs.

Figure 1

Here we demonstrate a novel electrochemical sensing architecture that retains the selectivity and convenience of E-DNA sensors– while expanding its range to the detection of macromolecules that bind to specific small molecules. The new architecture utilizes a largely double-stranded DNA as a rigid-but-dynamic scaffold to support a small-molecule recognition element. One strand of the scaffold, the “anchoring strand”, is attached to the electrode surface at its thiol-modified 5′ terminus and labeled with a redox tag (here methylene blue) at its 3′ terminus. The second strand, the “recognition strand”, is modified either at its 3′ terminus or, as shown, its 5′ terminus with a small-molecule recognition element. (Left) In the unbound state, the scaffold supports efficient electron transfer between the redox label and the electrode. (Center) The binding of the macromolecular target to this recognition element reduces the transfer efficiency, thus significantly reducing the observed faradaic current. (Right) Shown here are representative square wave voltammograms of the free and target-bound sensor (for the detection of 30 nM anti-digoxigenin antibody in 50% blood serum).

Figure 2

Sensor response depends on the flexibility of the DNA scaffold. Using a 27 base anchoring strand and 5′ placement of the biotin recognition element (distal to the electrode), we achieve optimal signaling with a recognition strand of 19 bases (centered on the middle of the anchoring strand). In contrast, 3′ placement of the recognition element produces a sharp transition from signal-off behavior at short lengths to signal-on behavior for recognition strands of 21 or more bases. Double-stranded scaffolds 17 bases in length lacking the small molecule recognition element (Ctrl) do not respond to target. All data represent addition of saturating (50 nM) streptavidin target.

Figure 3

Both the signal-on and signal-off streptavidin sensors achieve subnanomolar detection limits and are able to function in complex samples. Shown on the top are titrations of signal-on (23S17B3) and signal-off (19B5) streptavidin sensors in buffer (the biphasic nature of the curves arises due to the subnanomolar dissociation constant of the streptavidin–biotin interaction). The sensors function comparably in complex samples such as blood serum, soil suspensions, and beer (bottom), yielding similar signals upon addition of saturating (30 nM) streptavidin target. The error bars in this and the following figure represent the standard deviations of measurements conducted using three separately fabricated electrodes. The signal-on construct has large electrode-to-electrode variability in gain, although detection limits are similar for each individual electrode. (Figure S7 presents titration data for individual electrodes.)

Figure 4

The new sensing architecture appears to be general. For example, employing digoxigenin as the recognition element we readily detect anti-digoxigenin antibodies at low nanomolar concentrations in buffer (top; the biphasic curve arises due to the subnanomolar dissociation constant of the antibody–digoxigenin interaction). Antibody detection also functioned in complex sample matrices (bottom), with comparable signals obtained upon addition of saturating (30 nM) antibody target. Unlike the streptavidin sensor, no “signal-on” sensors are observed, suggesting that the “signal-off” approach may be more general.