Design considerations for array CGH to oligonucleotide arrays.

BACKGROUND: Representational oligonucleotide microarray analysis has been developed for detection of single nucleotide polymorphisms and/or for genome copy number changes. In this process, the intensity of hybridization to oligonucleotides arrays is increased by hybridizing a polymerase chain reaction (PCR)-amplified representation of reduced genomic complexity. However, hybridization to some oligonucleotides is not sufficiently high to allow precise analysis of that portion of the genome. METHODS: In an effort to identify aspects of oligonucleotide hybridization affecting signal intensity, we explored the importance of the PCR product strand to which each oligonucleotide is homologous and the sequence of the array oligonucleotides. We accomplished this by hybridizing multiple PCR-amplified products to oligonucleotide arrays carrying two sense and two antisense 50-mer oligonucleotides for each PCR amplicon. RESULTS: In some cases, hybridization intensity depended more strongly on the PCR amplicon strand (i.e., sense vs. antisense) than on the detection oligonucleotide sequence. In other cases, the oligonucleotide sequence seemed to dominate. CONCLUSION: Oligonucleotide arrays for analysis of DNA copy number or for single nucleotide polymorphism content should be designed to carry probes to sense and antisense strands of each PCR amplicon to ensure sufficient hybridization and signal intensity.     

Oligonucleotide array comparative genomic hybridization refines the structure of 8p23.1, 17q12 and 20q13.2 amplifications in gastric carcinomas

Oligonucleotide array comparative genomic hybridization (aCGH) was applied on fifteen gastric cancer (GCA) samples to reveal information of DNA copy number changes at an exon-level resolution. Twelve of the samples represented the intestinal (IGCA) and three the diffuse (DGCA) type of GCA. The samples had previously been assessed for genetic stability by microsatellite analysis and categorized into microsatellite phenotypes according to the type of alterations. As compared to our previous results obtained using cDNA platforms, the oligonucleotide platforms revealed more aberrations per sample (0-45 vs. 0-22). A total of 22 amplifications were detected by the oligonucleotide arrays. Ten of the amplicons had also been detected on the cDNA platform, but five of them spanned only one or a few cDNA clones, thus resembling apparent outliers. Two tumors showed five or more amplifications by oligonucleotide aCGH, suggesting the presence of an amplifier phenotype. The amplifications occurred irrespective of the microsatellite phenotypes. None of the DGCA tumors showed more than one aberration, whereas the IGCA tumors showed several aberrations. The increased resolution of the oligonucleotide arrays enabled the detection of amplicon boundaries at gene level, allowing, e.g., the determination of the 17q12 core amplicon and interstitial losses within the 8p23.1-->p22 and 20q13.2-->q13.1 amplifications. Previously no losses have been reported within amplified regions in GCA. In addition to novel amplified regions, the oligonucleotide array results describe novel targets for amplicons at 8p11 (SFRP1), 11p12 (LRRC4C), and 19q13.2 (CEACAM6).     

Microarray blob-defect removal improves array analysis

MOTIVATION: New generation Affymetrix oligonucleotide microarrays often have blob-like image defects that will require investigators to either repeat their hybridization assays or analyze their data with the defects left in place. We investigated the effect of analyzing a spike-in experiment on Affymetrix ENCODE tiling arrays in the presence of simulated blobs covering between 1 and 9% of the array area. Using two different ChIP-chip tiling array analysis programs (Affymetrix tiling array software, TAS, and model-based analysis of tiling arrays, MAT), we found that even the smallest blob defects significantly decreased the sensitivity and increased the false discovery rate (FDR) of the spike-in target prediction. RESULTS: We introduced a new software tool, the microarray blob remover (MBR), which allows rapid visualization, detection and removal of various blob defects from the .CEL files of different types of Affymetrix microarrays. It is shown that using MBR significantly improves the sensitivity and FDR of a tiling array analysis compared to leaving the affected probes in the analysis. AVAILABILITY: The MBR software and the sample array .CEL files used in this article are available at: http://liulab.dfci.harvard.edu/Software/MBR/MBR.htm     

Optimized design and assessment of whole genome tiling arrays

MOTIVATION: Recent advances in microarray technologies have made it feasible to interrogate whole genomes with tiling arrays and this technique is rapidly becoming one of the most important high-throughput functional genomics assays. For large mammalian genomes, analyzing oligonucleotide tiling array data is complicated by the presence of non-unique sequences on the array, which increases the overall noise in the data and may lead to false positive results due to cross-hybridization. The ability to create custom microarrays using maskless array synthesis has led us to consider ways to optimize array design characteristics for improving data quality and analysis. We have identified a number of design parameters to be optimized including uniqueness of the probe sequences within the whole genome, melting temperature and self-hybridization potential. RESULTS: We introduce the uniqueness score, U, a novel quality measure for oligonucleotide probes and present a method to quickly compute it. We show that U is equivalent to the number of shortest unique substrings in the probe and describe an efficient greedy algorithm to design mammalian whole genome tiling arrays using probes that maximize U. Using the mouse genome, we demonstrate how several optimizations influence the tiling array design characteristics. With a sensible set of parameters, our designs cover 78% of the mouse genome including many regions previously considered 'untilable' due to the presence of repetitive sequence. Finally, we compare our whole genome tiling array designs with commercially available designs. AVAILABILITY: Source code is available under an open source license from http://www.ebi.ac.uk/~graef/arraydesign/.     

Characterization of cDNA clones encoding rabbit and human serum paraoxonase: the mature protein retains its signal sequence.

Serum paraoxonase hydrolyzes the toxic metabolites of a variety of organophosphorus insecticides. High serum paraoxonase levels appear to protect against the neurotoxic effects of organophosphorus substrates of this enzyme [Costa et al. (1990) Toxicol. Appl. Pharmacol. 103, 66-76]. The amino acid sequence accounting for 42% of rabbit paraoxonase was determined by (1) gas-phase sequencing of the intact protein and (2) peptide fragments from lysine and arginine digests. From these data, two oligonucleotide probes were synthesized and used to screen a rabbit liver cDNA library. A clone was isolated and sequenced, and contained a 1294-bp insert encoding an open reading frame of 359 amino acids. Northern blot hybridization with RNA isolated from various rabbit tissues indicated that paraoxonase mRNA is synthesized predominately, if not exclusively, in the liver. Southern blot experiments suggested that rabbit paraoxonase is coded by a single gene and is not a family member of closely related genes. Human paraoxonase clones were isolated from a liver cDNA library by using the rabbit cDNA as a hybridization probe. Inserts from three of the longest clones were sequenced, and one full-length clone contained an open reading frame encoding 355 amino acids, four less than the rabbit paraoxonase protein. Each of the human clones appeared to be polyadenylated at a different site, consistent with the absence of the canonical polyadenylation signal sequence. Of potential significance with respect to the paraoxonase polymorphism, the derived amino acid sequence from one of the partial human cDNA clones differed at two positions from the full-length clone. Amino-terminal sequences derived from purified rabbit and human paraoxonase proteins suggested that the signal sequence is retained, with the exception of the initiator methionine residue [Furlong et al. (1991) Biochemistry (preceding paper in this issue)]. Characterization of the rabbit and human paraoxonase cDNA clones confirms that the signal sequences are not processed, except for the N-terminal methionine residue. The rabbit and human cDNA clones demonstrate striking nucleotide and deduced amino acid similarities (greater than 85%), suggesting an important metabolic role and constraints on the evolution of this protein.     

Quantitative assessment of filter-based cDNA microarrays: gene expression profiles of human T-lymphoma cell lines

MOTIVATION: While the use of cDNA microarrays for functional genomic analysis has become commonplace, relatively little attention has been placed on false positives, i.e. the likelihood that a change in measured radioactive or fluorescence intensity may reflect a change in gene expression when, in fact, there is none. Since cDNA arrays are being increasingly used to rapidly distinguish biomarkers for disease detection and subsequent assay development (Wellman et al., Blood, 96, 398-404, 2000), the impact of false positives can be significant. For the use of this technology, it is necessary to develop quantitative criteria for reduction of false positives with radioactively-labeled cDNA arrays. RESULTS: We used a single source of RNA (HuT78 T lymphoma cells) to eliminate sample variation and quantitatively examined intensity ratios using radioactively labeled cDNA microarrays. Variation in intensity ratios was reduced by processing microarrays in side-by-side (parallel mode) rather than by using the same microarray for two hybridizations (sequential mode). Based on statistical independence, calculation of the expected number of false positives as a function of threshold showed that a detection limit of [log(2)R] >0.65 with agreement from three replicates could be used to identify up- or down-modulated genes. Using this quantitative criteria, gene expression differences between two related T lymphoma cell lines, HuT78 and H9, were identified. The relevance of these findings to the known functional differences between these cell types is discussed.     

Assessing unmodified 70-mer oligonucleotide probe performance on glass-slide microarrays

Background

Long oligonucleotide microarrays are potentially more cost- and management-efficient than cDNA microarrays, but there is little information on the relative performance of these two probe types. The feasibility of using unmodified oligonucleotides to accurately measure changes in gene expression is also unclear.

Results

Unmodified sense and antisense 70-mer oligonucleotides representing 75 known rat genes and 10 Arabidopsis control genes were synthesized, printed and UV cross-linked onto glass slides. Printed alongside were PCR-amplified cDNA clones corresponding to the same genes, enabling us to compare the two probe types simultaneously. Our study was designed to evaluate the mRNA profiles of heart and brain, along with Arabidopsis cRNA spiked into the labeling reaction at different relative copy number. Hybridization signal intensity did not correlate with probe type but depended on the extent of UV irradiation. To determine the effect of oligonucleotide concentration on hybridization signal, 70-mers were serially diluted. No significant change in gene-expression ratio or loss in hybridization signal was detected, even at the lowest concentration tested (6.25 μm). In many instances, signal intensity actually increased with decreasing concentration. The correlation coefficient between oligonucleotide and cDNA probes for identifying differentially expressed genes was 0.80, with an average coefficient of variation of 13.4%. Approximately 8% of the genes showed discordant results with the two probe types, and in each case the cDNA results were more accurate, as determined by real-time PCR.

Conclusions

Microarrays of UV cross-linked unmodified oligonucleotides provided sensitive and specific measurements for most of the genes studied.     

Preparation of sensitive and specific oligonucleotide probes tailed using terminal transferase and dITP

An oligonucleotide probe tailed with deoxyadenosine-5'-triphosphate or deoxythymine-5'-triphosphate is detectable with high sensitivity, but has a major drawback--the tail co-hybridizes specifically to complementary sequences. This can be a problem when screening cDNA clones that contain poly(dA) sequences. While it is possible to mask the cDNA tail with unlabeled poly(dA) or poly(A) oligonucleotides, false-positive clones are still produced because complete masking of extremely long (dA) tails is difficult. As a result, only cDNA clones that have extremely long poly(dA) sequences are often obtained by hybridization screening using tailed probes. In this report, we describe an oligonucleotide probe tailed with DIG-labeled nucleotide in combination with deoxyinosine-5'-triphosphate that was highly specific and sensitive to cDNAs. Terminal deoxynucleotidyl transferase efficiently adds dI nucleotides to the 3'-end. The dI of the tails did not pair with any nucleotides under stringent hybridization so that the specificity of hybridization assays remained high without affecting the sensitivity of the test. Colony hybridization experiments demonstrated that there were very few (1 of 80 tested) false positives using this technique. Its use may increase the accuracy of cDNA screening.     

Predominant expression of a peroxidase gene in staminate flowers of Mercurialis annua

A direct correlation between the feminizing process of male plants of Mercurialis annua by benzylaminopurine and the disappearance of male-specific peroxidases was established. In order to understand the process of hormonal regulation of the sexual differentiation and peroxidase activity, two peroxidase cDNA clones were selected from a male flower cDNA library with degenerated oligonucleotide probes encoding a peroxidase-specific site. These clones contained three peroxidase-specific sequences and an Eco RI restriction site. The two Eco RI fragments of one cDNA clone were subcloned and used as probes to analyse the expression of corresponding gene(s). The northern blot analysis showed hybridization on a single mRNA band with a predominant expression in male flowers. The feminizing treatment with a cytokinin induced the progressive disappearance of the hybridization signal in correlation with the disappearance of male specific isoperoxidases.     

Isolation and Characterization of Rat and Human cDNAs Encoding a Novel Putative Peroxisomal Enoyl-CoA Hydratase

We have used a PCR-based subtractive hybridization method to identify upregulated cDNAs in the livers of rats treated with a peroxisome proliferator [clofibrate or di(2-ethylhexyl) phthalate]. After four rounds of subtractive hybridization 62 differentially hybridizing clones were partially sequenced and analyzed by sequence homology searching. Of 62, 49 were identical to 14 different upregulated rat sequences in the databank (mostly genes encoding microsomal or peroxisomal enzymes), 4 of 62 were fragments of three previously unknown genes, and 9 of 62 were false positives. Two of the unknown fragments hybridized to a single novel cDNA that was found to be more than 20-fold induced by both peroxisome proliferators. The 36-kDa predicted protein product of this cDNA shows a high degree of sequence homology to enoyl-CoA hydratases of several different species and has a C-terminal peroxisomal targeting sequence. An epitope-tagged protein product of a full-length cDNA was targeted to peroxisomes in a human cell line. We named this gene, which encodes an apparent peroxisomal enoyl-CoA hydratase, ECH1. We have also identified human ECH1 cDNA and mapped its structural gene to 19q13, 3′ to the ryanodine receptor, by hybridization to somatic cell hybrid DNA and chromosome 19-specific cosmid arrays. Possible roles for the ECH1 protein product in peroxisomal β-oxidation are discussed.     

Impact of surface chemistry and blocking strategies on DNA microarrays

The surfaces and immobilization chemistries of DNA microarrays are the foundation for high quality gene expression data. Four surface modification chemistries, poly-L-lysine (PLL), 3-glycidoxypropyltrimethoxysilane (GPS), DAB-AM-poly(propyleminime hexadecaamine) dendrimer (DAB) and 3-aminopropyltrimethoxysilane (APS), were evaluated using cDNA and oligonucleotide sub-arrays. Two un-silanized glass surfaces, RCA-cleaned and immersed in Tris-EDTA buffer were also studied. DNA on amine-modified surfaces was fixed by UV (90 mJ/cm(2)), while DNA on GPS-modified surfaces was immobilized by covalent coupling. Arrays were blocked with either succinic anhydride (SA), bovine serum albumin (BSA) or left unblocked prior to hybridization with labeled PCR product. Quality factors evaluated were surface affinity for cDNA versus oligonucleotides, spot and background intensity, spotting concentration and blocking chemistry. Contact angle measurements and atomic force microscopy were preformed to characterize surface wettability and morphology. The GPS surface exhibited the lowest background intensity regardless of blocking method. Blocking the arrays did not affect raw spot intensity, but affected background intensity on amine surfaces, BSA blocking being the lowest. Oligonucleotides and cDNA on unblocked GPS-modified slides gave the best signal (spot-to-background intensity ratio). Under the conditions evaluated, the unblocked GPS surface along with amine covalent coupling was the most appropriate for both cDNA and oligonucleotide microarrays.     

High-throughput polymorphism screening and genotyping with high-density oligonucleotide arrays

A highly reliable and efficient technology has been developed for high-throughput DNA polymorphism screening and large-scale genotyping. Photolithographic synthesis has been used to generate miniaturized, high-density oligonucleotide arrays. Dedicated instrumentation and software have been developed for array hybridization, fluorescent detection, and data acquisition and analysis. Specific oligonucleotide probe arrays have been designed to rapidly screen human STSs, known genes and full-length cDNAs. This has led to the identification of several thousand biallelic single-nucleotide polymorphisms (SNPs). Meanwhile, a rapid and robust method has been developed for genotyping these SNPs using oligonucleotide arrays. Each allele of an SNP marker is represented on the array by a set of perfect match and mismatch probes. Prototype genotyping chips have been produced to detect 400, 600 and 3000 of these SNPs. Based on the preliminary results, using oligonucleotide arrays to genotype several thousand polymorphic loci simultaneously appears feasible.