A simple enzymatic labeling procedure is described to determine spot quality in oligonucleotide microarrays. By using fluorescently labeled dideoxynucleotides or ribonucleotides as substrate for terminal deoxynucleotidyl transferase (TdT), a single fluorophore can be covalently attached at the 3′ end of each oligonucleotide probe molecule in the spot. Fluorescein- 12-ddUTP, Cy™3-ddUTP, Cy5-UTP, and Cy3-UTP were compared as TdT substrates for 3′ end labeling an array of 1273 hexamer probes. Cy5-UTP was found to show minimal bias toward probe base composition and is therefore well suited for quantitative analysis of microarray spots where the oligonucleotide probes are coupled via a 5′ end linkage to the solid phase.

Introduction

Terminal deoxynucleotidyl transferase (TdT) is a template-independent DNA polymerase that can incorporate fluorochrome-labeled nucleotides at the 3′-hydroxyl terminus of DNA strands or oligonucleotides. In the presence of a labeled deoxynucleotide-triphosphate (dNTP), TdT can incorporate 20 to >100 labeled bases per 3′ end (1). Polymerization appears unaffected by the type of base from which the 3′ end extension takes place. However, the efficiency of incorporation is not the same for all nucleotides (2). TUNEL, an assay for apoptosis-induced nuclear DNA fragmentation, with concomitant generation of 3′ ends (3), utilizes fluorescein-labeled dUTP as a substrate for TdT. Like other DNA polymerases, TdT will use a dideoxynucleotide-triphosphate (ddNTP) as substrate for incorporation on the 3′ end of the nascent chain, an event that will stop further chain elongation due to the absence of a 3′-hydroxyl group on the incorporated nucleotide itself. Incorporation of labeled ddNTP is therefore directly proportional to the number of available DNA or oligonucleotide 3′ ends in the reaction (4). Interestingly, TdT can also utilize fluorochrome-labeled ribonucleotide-triphosphate (NTP) as substrate for incorporation at the 3′ end of oligonucleotides (5). Unlike incorporation of dNTPs, incorporation of NTPs is not processive. Thus, most DNA strands acquire only one ribonucleotide at the 3′ end, and very few acquire two or more ribonucleotides. Moreover, the choice of fluorophore in labeled NTPs has a significant effect on incorporation efficiency (5).

In the manufacture of oligonucleotide microarrays using prefabricated oligonucleotides, directional coupling of the oligonucleotide probe to the solid substrate, typically via a 5′ end linker, ensures uniform orientation and presumably functionality (6). However, due to their relatively short length (20- to 70-mers) and single-stranded nature, simple techniques for assessing DNA microarray spot quality, for example, staining with dyes that show enhanced fluorescence upon binding to nucleic acids (7), are of limited usefulness. The blocked 5′ end also precludes the use of T4 polynucleotide kinase for probe labeling by phosphorylation (8). A transcriptional analysis platform was recently described that is based on a universal oligonucleotide microarray. The universal microarray consists of all possible hexamers 4096 arrayed in triplicate on a single slide (9). To ensure adequate quality, we developed a procedure for assessment of spot quality in hexamer microarrays. The procedure can be readily adapted for quality control of any oligonucleotide microarray production process in which probes are directionally coupled to the solid substrate via a 5′ end linkage, regardless of the probe sequence composition.

Materials and Methods

Microarrays were printed by pin deposition (Microarrays, Nashville, TN, USA) on CodeLink™ slides (GE Healthcare, Piscataway, NJ, USA). Probe oligonucleotides (Integrated DNA Technologies, Coralville, IA, USA) consisted of a variable hexamer sequence and a CT mixed position at the 5′ end. The terminal 5′ end was first extended with a C18 spacer, then with a C6 amino-modified linker. Probe concentration was 300 µM in 150 mM sodium phosphate, pH 8.5. The 3′ end labeling reactions were carried out in a Ventana DISCOVERY™ System (Ventana Medical Systems, Tucson, AZ, USA). Microarrays were covered with 280 µL reaction buffer containing 50 mM potassium-acetate, 20 mM Tris-acetate, pH 7.9, 10 mM magnesium-acetate, 1 mM dithiothreitol, 250 µM CoCl₂, 2 U/µL TdT (New England Biolabs, Ipswich, MA, USA), and 100 µM fluorescent nucleotide. Labeling reactions were incubated at 37°C for 1 h. Labeled microarrays were washed according to recommended procedures (Ventana Medical Systems). Fluorescein-12-ddUTP was purchased from Sigma (St. Louis, MO, USA), and Cy™3-ddUTP, Cy5-UTP, and Cy3-UTP were purchased from PerkinElmer Life and Analytical Sciences (Wellesley, MA, USA). Microarrays were scanned using a ScanArray™ Express HT (PerkinElmer Life and Analytical Sciences) at 10 µm resolution. Fluorescent signal intensities (relative fluorescent units or RFU) were quantified using Array-Pro® v4.5 (Media Cybernetics, Silver Spring, MD, USA). A threshold algorithm was used to detect spot edges. Spots contained 186 pixels on average. The sum of all pixel intensities were captured per spot, and averages were calculated from triplicate spots in the array. For a test of microarray probe functionality, ligation reactions were carried out as previously described (9). A 33-mer Cy5-labeled degenerate hairpin oligonucleotide (Integrated DNA Technologies) was used as a surrogate target mixture. The 5′ end-phosphorylated 33-mer is designed to fold into itself to form a 10-bp stem, a 4-base loop, and a single-stranded 9-base 3′ extension containing a degenerate hexamer region with Cy5 blocking the terminal 3′ end. These partial single-stranded target mixtures mimic the annealed targets for ligation to hexamer microarrays in the universal microarray system (UMAS) (9).