Introduction

Microarray technology, allowing the quantitative and simultaneous monitoring of the expression of thousands of genes, initialized a new era in gene expression studies (1). It provides a unique tool for the determination of gene expression at the level of messenger RNA (mRNA) and has become a widely utilized technology in laboratories around the world to perform gene expression profiling studies for clinical (2,3,4,5,6) and diagnostical applications (7,8,9,10,11,12) or to determine genetic variation through the detection of single nucleotide polymorphisms (SNPs) (13,14,15). Although the microarray technology has existed for more than a decade, there are a number of effects that remain ill-understood. One such example is the occurrence of doughnut patterns on the microarray scans. A doughnut or ring pattern can be defined as an intensity pattern occurring at the spot level in which the intensity signal is higher at the edges and lower at the center of the probe spot (Figure 1A). The intensity profile across a section of Figure 1B shows that the intensity between the center and the rims of the spot can differ more than 2-fold in intensity.

Figure 1. Typical doughnut hybridization profiles. (Click to enlarge)

The presence of a ring-shaped intensity pattern makes the spots harder to analyze, because it leads to very high in traspot standard deviation values and some of the image analysis tools like GenePix® (Axon Instruments), ImaGene™ (Biodiscovery), QuantArray® (GSI Lumonics), and ScanAlyze do not contain algorithms that are able to detect and analyze spots with these artifacts in a correct manner. The spots are often flagged and left out of the analysis. It is therefore of great importance to get a better insight into the formation of ring-shaped hybridization patterns. Until now, two major causes have been proposed. There have been reports (16,17,18) claiming that doughnuts are created during the spotting process when contact is made between the spotting pins and the microarray surface. However, a good calibration of the microarray spotter can easily circumvent this type of doughnut formation. Another more likely explanation for the occurrence of these patterns is based on the so called coffee-stain effect (19). When a drop of liquid containing solutes dries on a solid surface, it leaves a dense, ring-like deposit over the entire perimeter. The ring formation can be explained as the consequence of a geometrical constraint of the droplet surface. As the free surface is constrained by the pinned contact line, the fluid is squeezed outwards to compensate for evaporative losses (19). Originally, the dissolved DNA material is homogeneously distributed over the entire droplet, but as the solvent evaporates, the DNA molecules flow outwards with the fluid flow. The DNA material is therefore accumulated on the edges of the spot. When spotting DNA microarrays, this effect deposits more probe molecules on the edges of the spot and almost none in the center, resulting in an uneven signal distribution after hybridization.

In order to prevent doughnut patterns during the drying process, different technological adaptations have recently been proposed. A better control of the evaporation process [slower evaporation by using a dimethyl sulfoxide (DMSO) spotting solution, higher humidity in the spotting chamber] and the use of spotting substrates that are ultraphobic instead of the mildly hydrophobized glass slides have reduced the occurrence of doughnut patterns. Because the surface-pinned contact line is the major reason for the coffee-stain effect, a very hydrophobic surface with high contact angles can also be used to reduce the fixation of the contact line (20).

Although the number of ringshaped patterns occurring during microarray analysis has been reduced by the technological adaptations cited above, these artifacts still keep arising in some cases (usually posthybridization). A wider investigation into the phenomenon of doughnuts on microarrays is hence necessary. The present study reports on results obtained from four different approaches that, when combined, show that doughnuts cannot only be formed during the spotting and drying process, but that the hybridization process itself can be considered as an important cause. In the present study, a combination of computer simulations, theoretical, optical, and experimental techniques is used to demonstrate that the occurrence of diffusion-limited conditions during the hybridization process can cause these hybridization patterns. With the term diffusion limitation, we refer to the situation in which the actual hybridization rate is slower than that expected on the basis of the intrinsic reaction kinetics.