Microfluidic devices integrating microcavity surface-plasmon-resonance sensors: glucose oxidase binding-activity detection

Abstract

We have developed miniature (approximately 1 microm diameter) microcavity surface-plasmon-resonance sensors (MSPRS), integrated them with microfluidics, and tested their sensitivity to refractive-index changes. We tested their biosensing capability by distinguishing the interaction of glucose oxidase (M(r) 160 kDa) with its natural substrate (beta-D-glucose, M(r) 180 Da) from its interactions with nonspecific substrates (L-glucose, D-mannose, and 2-deoxy-D-glucose). We ran the identical protocol we had used with the MSPRS on a Biacore 3000 instrument using their bare gold chip. Only the MSPRS was able to detect beta-D-glucose binding to glucose oxidase. Each MSPRS can detect the binding to its surface of fewer than 35,000 glucose oxidase molecules (representing 9.6 fg or 60 zmol of protein), about 10(6) times fewer than classical surface-plasmon-resonance biosensors.

Figure 1

(a) Spectra of light emitted by one MSPRS in air (120 ± 20 nm gold-coating thickness on 780 ± 6 nm diameter polystyrene nanospheres) excited using white light in two configurations: (solid line) the MSPRS is illuminated from the top of the sensor and the emitted light collected through the nanoaperture and (dashed line) the MSPRS is illuminated through the nanoaperture and the emitted light collected from the top of the sensor. (b) Spectra of MSPRS-emitted light and neighboring bare flat gold (10 μm to left of the MSPRS) under air (n = 1.000) and under DI-water (n = 1.333). (c) Lorentzian fit of the MSPRS-spectrum under air: (solid line) experiment, (○) multi-peak fit (R² = 0.9964), (△) individual peaks and (dotted line) MSPRS-spectrum under water (experiment). (d) Lorentzian fit of the MSPRS-spectrum under DI-water: (solid line) experiment, (○) multi-peak fit (R² = 0.9968), (△) individual peaks and (dotted line) MSPRS-spectrum under air (experiment).

Figure 2

(a) (solid line) Emitted-light intensity (at 660 nm) of a single MSPRS during two-step sensor functionalization and (○) exponential-decay best-fits: R²(DTSSP) = 0.985, R²(GOx) = 0.965. We used 0.5 nL/s flows at 24.0 ± 0.1°C. *No signal recorded during sample loading. (b) MSPRS spectra under DI-water: (solid line) before and (dashed line) after DTSSP-GOx-complex binding. The formation of a molecular layer (6 ± 1 nm) red-shifted resonance (IV) by 4 nm. Time series in (a) recorded at 660 nm (vertical line). (c) (solid line) The same two-step bare-gold-surface functionalization monitored using a Biacore 3000 and (○) exponential-decay best-fits: R²(DTSSP) = 0.961, R²(GOx) = 0.987. We used 10 μL/s flows at 25°C. *The Biacore 3000 was on standby during sample loading.

Figure 3

Single-MSPRS emitted-light intensity at 660 nm during GOx conformational changes. We injected 1.56 μM GOx freshly dissolved in DI-water into a DTSSP-functionalized chip, let it reach saturation (≈1 h) then washed in PBS for ≈10 h. An exponential fit gave a 106 ± 1 min relaxation time. *No signal recorded during sample loading.

Figure 4

(a) Single-MSPRS signal (at 660 nm) during surface-bound GOx (Figure 2a) interaction with L-Glu, 2Do-Glu, D-Man and D-Glu solutions (all substrates dissolved at 100 mM in PBS). Flow rate 0.5 nL/s. Solution temperature 24.0 ± 0.1 °C. The L-Glu signal serves as a control for the effect of the refractive indices of the substrate solutions vs PBS because it does not bind to GOx. D-Glu produces a fast signal change due to the refractive index of the solution followed by a slower change due to formation of the βD-Glu–GOx complex. D-Glu 100 mM in PBS (50 mM βD-Glu equivalent) interacts with GOx. D-Man and 2Do-Glu interact very weakly with GOx. This experiment used the same MSPRS as in Figure 2a,b. (b) Biacore 3000 signals during an equivalent experiment. Flow rate 10 μL/s. Solution Temperature 25°C. We have subtracted the reference-channel signal from the functionalized-channel signal (Figure 2c). The uniform 25 ± 3 RU signal change for all substrates indicates sensitivity to the solutions’ refractive indices, but no detection of βD-Glu–GOx binding.

Figure 5

Wash-in/wash-out cycles using a single MSPRS showing reproducible GOx–D-Glu interaction detection in different buffers (0.5 nL/s flow rate, 24.0 ± 0.1°C): (a) Differential signal (see Figure S-5) alternating substrate solutions with PBS. The corrected GOx–D-Glu binding signal was 109 ± 22 counts/s. (b) Differential signal alternating substrate solutions with L-Glu. As a second control we ran a four-cycle experiment with 100 mM D-Glu plus 100 mM L-Glu in PBS as the reagent solution and rinsed with 200 mM L-Glu in PBS. Both solutions had the same refractive index. After we subtracted the baseline, the corrected GOx–D-Glu binding signal was 116 ± 21 counts/s. (c) Differential signal alternating substrate solutions with glucose-free DMEM 1x. The corrected GOx–D-Glu-binding signal was 128 ± 29 counts/s. (d) Differential signal alternating substrate solutions with glucose-free 10% HS in glucose-free DMEM 1x. The corrected GOx–D-Glu-binding signal was 112 ± 25 counts/s. Arrows mark the times of reagent or rinsing-buffer injections.