Enhancement of DNA micro-array analysis using a shear-driven micro-channel flow system

Abstract

A very simple micro-channel flow system is used to investigate the potential gain in hybridization rate stemming from the induction of a convective flow past the surface of a DNA micro-array. Reporting on a series of preliminary experiments wherein a two-dimensional convective flow is created past the surface of a conventional micro-array slide, the analysis time could be brought down from overnight waiting (16 h) to some 10 to 30 min. The experiments open the road towards the development of novel, convection-driven hybridization systems yielding shorter analysis times, and/or lower detection limits.

Introduction

In the past decade, DNA micro-array technology has gained wide use in analytical chemistry, with applications in important areas such as gene identification, gene mapping, DNA sequencing and clinical diagnostics [1], [2]. Traditionally, DNA screening assays are effectuated by applying an aliquot of sample between two parallel plates, one carrying a high-density micro-array of different immobilized target DNA spots with known base sequence, while the other plate simply serves to seal off the system and to prevent evaporation of the sample liquid. In such systems, the transport of the sample molecules towards the target spots occurs in a passive mode, i.e., by pure molecular diffusion (cf. the diffusive path in Fig. 1a). As a result, micro-array analyses are very slow, and usually have to occur overnight. Another consequence of the slow diffusive transport is that micro-arrays are highly inefficient in terms of binding efficiency and detection limit. In general, twenty to forty million copies of the same sequence species have to be present in the sample to reach the typical detection limit of 20 000–50 000 hybridized strands per spot. This corresponds to a binding efficiency of less than 1%. Table 1 shows that this poor binding efficiency is due to the fact that, even after overnight incubation, a given target spot can typically only be reached by the probe DNA strands which are present within a distance of less than ℓ=1 mm from the target spot. The latter value is obtained from Einstein’s law of diffusion:

$$\text{ℓ}{}^2\text{=2D}{}_{\mathrm{<mtext>mol</mtext>}}\text{t}$$

using a typical value of D_mol=10⁻¹¹ m²/s to represent the rate of diffusion of DNA strands in a typical micro-array analysis [3]. Comparing the area of a circular region with radius 1 mm to the total surface area of the chip, it can easily be calculated that each spot only “sees” less than 3 mm² out of the total 1875 mm² contacted with the DNA sample in a conventional, microscope slide-sized chip. This corresponds to a very poor maximal binding efficiency of less than 0.2%. Table 1 also shows that if one would try to bring this efficiency up to a level of 2% (requiring to sample the DNA from a region with radius 3 mm), a 6-day analysis would be required, while reaching a 20% binding efficiency would require a dramatically long 2-month analysis time.

To alleviate this diffusion limitation, several solutions have already been proposed: the application of electrically-induced flows [4] and hybridization in flow-through nanopores [5], [6]. Both solutions however only consider the enhancement of the diffusion process in the direction perpendicular to the target spots, and do not solve the problem of the extremely slow lateral (i.e., in the x- and z-directions) transport. In the present contribution, we propose the use of shear-driven flows to generate a lateral convective transport across the micro-array surface.

Shear-driven flows have already been proposed as a solution for the pressure-drop limitation in on-chip liquid chromatography [7], [8], [9], and have the unique property that they can transport extremely thin fluid layers at very high velocities, without the aid of an electrical field and much faster than what is possible with a pumping system. The flow driving principle is very simple, and relies on the dragging action exerted by a moving surface on an adjacent fluid layer. The resulting velocity profile is linear (u=0 near the wall carrying the target spots, u=u_wall near the moving wall), such that, due to the rapid diffusive radial equilibration, all liquid molecules travel at a mean velocity equaling one half of the moving wall velocity, independently of the channel thickness. In its most extreme limit, it could even be possible to use flow channels which are only a minimal number of times larger than the probe molecules, thereby nearly completely eliminating the time needed for radial diffusion (cf. y-direction in Fig. 1b), whereas the transport in the lateral direction (i.e., from spot to spot in the x-direction, see Fig. 1b) occurs by rapid convection (literally guiding the sample DNA strands from spot to spot).

In the present paper, the emphasis is on the potential increase of the hybridization rate which can be obtained by inducing a lateral convective flow past the chip surface. In a later stage, the further increase of the hybridization rate which can be expected from the use of thinner, and hence kinetically more advantageous channels will be investigated. The latter is however less important, because the diffusion limitation in conventional DNA chips is much more pronounced in the lateral direction (where the diffusion distances are of the order of centimeters) than in the radial direction (where the diffusion distances are only of the order of 50 μm). Furthermore, by decreasing the channel height, the amount of sample which is contacted with the target spots is reduced as well, unless more concentrated samples would be used in order to keep the total amount of DNA constant.