Electrostatic readout of DNA microarrays with charged microspheres

Abstract

DNA microarrays are used for gene-expression profiling, single-nucleotide polymorphism detection and disease diagnosis. A persistent challenge in this area is the lack of microarray screening technology suitable for integration into routine clinical care. Here, we describe a method for sensitive and label-free electrostatic readout of DNA or RNA hybridization on microarrays. The electrostatic properties of the microarray are measured from the position and motion of charged microspheres randomly dispersed over the surface. We demonstrate nondestructive electrostatic imaging with 10-mum lateral resolution over centimeter-length scales, which is four-orders of magnitude larger than that achievable with conventional scanning electrostatic force microscopy. Changes in surface charge density as a result of specific hybridization can be detected and quantified with 50-pM sensitivity, single base-pair mismatch selectivity and in the presence of complex background. Because the naked eye is sufficient to read out hybridization, this approach may facilitate broad application of multiplexed assays.

Figure 1

Electrostatic microarray readout using microparticle probes. (a) A suspension of negatively charged silica microspheres is gravitationally sedimented over a microarray surface. The positions and motions of a population of microspheres are used to image the surface charge of the microarray and detect hybridization. This is because areas displaying double-stranded DNA are highly negatively charged compared to areas displaying ssDNA, and both contrast with the positively charged background. (b) Typical epifluorescence image of a microspot displaying DNA A after hybridizing with 50 nM Cy3-labeled A′ (20 min, 1× SSC). (c) Brightfield image after 5.6-mm diameter silica microspheres are allowed to sediment gravitationally for 20 min. The dashed line indicates the spot's perimeter as determined by fluorescence. (d) Representative RICM image of 5.6-μm diameter silica spheres. Such interferograms are used to measure the height of microspheres, and, consequently, the magnitude of electrostatic repulsion over the surface. (e) The charge density map, as compiled from RICM observations of microspheres (black dots). (f) Over negatively charged areas, microspheres are laterally mobile as indicated by the variance of brightfield intensity over time (pixel size is 3 × 3 μm). (g) Microspheres that remain adhered after agitating the surface can be observed by darkfield microscopy to identify positively charged regions. Images b–g correspond to the same spot and were collected in 100 μM NaCl.

Figure 2

Electrostatic response to DNA surface density. (a) A graded DNA density was generated by printing spots with a mixture of specific, A, and control, B, DNA while maintaining a constant total DNA concentration (5 or 6 μM, as indicated). Charge density and fluorescence images of the same array are shown after hybridization with 50 nM A′. (b) Plot of the average charge density and fluorescence intensity in spots along the 5-μM lane in a. The dashed lines are linear fits. The charge density roughly doubles as the molar fraction of A increases from 0 to 1. This is consistent with the expectation that complementary DNA binding should double the ssDNA charge density. (c) Electrostatic response of a 2 × 2 array of A and B DNA to specific hybridization with target DNA A′, B′ or both A′ and B′ strands. (d) Image of A₁₂ (5′-TACCACATCATC-3′) and A_12m (5′-TACCAAATCATC-3′) spots before and after hybridization with 50 nM A′ for 20 min, which indicates that electrostatic imaging can resolve a single base-pair mismatch. (e) Epifluorescence and electrostatic images of A and B spots after overnight hybridization with 100 pM A′ DNA. All hybridizations were performed in 1× SSC and images were collected in 100 μM NaCl.

Figure 3

Simplified readout using charged microparticles. (a) A series of microarray spots are printed with a gradient of ssDNA densities to titrate the surface charge from net positive to net negative. After hybridization (in 1× SSC), complementary spots become more negatively charged. In each series, the change in DNA density can be identified by a shift in the number of negatively biased spots relative to a control series. (b) Schematic and experimental data demonstrate this concept. Images of the variance in brightfield intensity over 30 s indicate where sedimented 2.34-μm diameter silica spheres remain mobile. Negatively charged areas appear bright due to the lateral motion of microspheres repelled by the surface. Relative to the control DNA series, B, two additional spots in the AA′ row appear negatively biased, indicating a specific change in charge density due to hybridization. (c) The observed shift is dependent on the concentration of target A′. The plot compares this label-free readout with fluorescence data obtained on the same substrate under identical conditions. Inset points were hybridized overnight, and all others were performed for 20 min. (d) Darkfield and epifluorescence (inverse contrast) micrographs of a representative area from a 7,000-spot microarray hybridized (20 min, 1× SSC, 50 nM A′). This suggests that this assay is compatible with conventional microarrays that cover cm² areas. (e) Photograph of a side-illuminated microarray after hybridization and development with 2.34-μm diameter silica spheres. On average, there are 200–300 microparticles/spot. Inset, right, shows a digitally magnified region of the array (inverse contrast with subtracted background). Bright areas indicate regions of high DNA density. All images collected in 100 μM NaCl.

Figure 4

Label-free expression profiling with primary mRNA. (a) Scheme of procedure used to measure mRNA expression in breast adenocarcinoma MCF-7 cells. (b) A brightfield intensity variance image of 2.34-μm silica microspheres shows the differential expression of human aldolase A gene (ALD) and human methionine-tRNA synthetase (MARS) gene in a 4 × 4 array of spots (in 100 μM NaCl). This indicates that MARS is more highly expressed compared to ALD in this sample of cells.