Evaluation of the gene-specific dye bias in cDNA microarray experiments

MOTIVATION: In cDNA microarray experiments all samples are labeled with either Cy3 or Cy5. Systematic and gene-specific dye bias effects have been observed in dual-color experiments. In contrast to systematic effects which can be corrected by a normalization method, the gene-specific dye bias is not completely suppressed and may alter the conclusions about the differentially expressed genes. METHODS: The gene-specific dye bias is taken into account using an analysis of variance model. We propose an index, named label bias index, to measure the gene-specific dye bias. It requires at least two self-self hybridization cDNA microarrays. RESULTS: After lowess normalization we have found that the gene-specific dye bias is the major source of experimental variability between replicates. The ratio (R/G) may exceed 2. As a consequence false positive genes may be found in direct comparison without dye-swap. The stability of this artifact and its consequences on gene variance and on direct or indirect comparisons are addressed. AVAILABILITY: http://www.inapg.inra.fr/ens_rech/mathinfo/recherche/mathematique     

Statistical design of reverse dye microarrays

MOTIVATION: In cDNA microarray experiments all samples are labelled with either Cy3 dye or Cy5 dye. Certain genes exhibit dye bias-a tendency to bind more efficiently to one of the dyes. The common reference design avoids the problem of dye bias by running all arrays 'forward', so that the samples being compared are always labelled with the same dye. But comparison of samples labelled with different dyes is sometimes of interest. In these situations, it is necessary to run some arrays 'reverse'-with the dye labelling reversed-in order to correct for the dye bias. The design of these experiments will impact one's ability to identify genes that are differentially expressed in different tissues or conditions. We address the design issue of how many specimens are needed, how many forward and reverse labelled arrays to perform, and how to optimally assign Cy3 and Cy5 labels to the specimens. RESULTS: We consider three types of experiments for which some reverse labelling is needed: paired samples, samples from two predefined groups, and reference design data when comparison with the reference is of interest. We present simple probability models for the data, derive optimal estimators for relative gene expression, and compare the efficiency of the estimators for a range of designs. In each case, we present the optimal design and sample size formulas. We show that reverse labelling of individual arrays is generally not required.     

Comparison of Alexa Fluor<Superscript>®</Superscript> and CyDye<Superscript>™</Superscript> for practical DNA microarray use

Microarrays are a powerful tool for comparison and understanding of gene expression levels in healthy and diseased states. The method relies upon the assumption that signals from microarray features are a reflection of relative gene expression levels of the cell types under investigation. It has previously been reported that the classical fluorescent dyes used for microarray technology, Cy3 and Cy5, are not ideal due to the decreased stability and fluorescence intensity of the Cy5 dye relative to the Cy3, such that dye bias is an accepted phenomena necessitating dye swap experimental protocols and analysis of differential dye affects. The incentive to find new fluorophores is based on alleviating the problem of dye bias through synonymous performance between counterpart dyes. Alexa Fluor 555 and Alexa Fluor 647 are increasingly promoted as replacements for CyDye in microarray experiments. Performance relates to the molecular and steric similarities, which will vary for each new pair of dyes as well as the spectral integrity for the specific application required. Comparative analysis of the performance of these two competitive dye pairs in practical microarray applications is warranted towards this end. The findings of our study showed that both dye pairs were comparable but that conventional CyDye resulted in significantly higher signal intensities (P < 0.05) and signal minus background levels (P < 0.05) with no significant difference in background values (P > 0.05). This translated to greater levels of differential gene expression with CyDye than with the Alexa Fluor counterparts. However, CyDye fluorophores and in particular Cy5, were found to be less photostable over time and following repeated scans in microarray experiments. These results suggest that precautions against potential dye affects will continue to be necessary and that no one dye pair negates this need.     

Possible sources of dye-related signal correlation bias in two-color DNA microarray assays

DNA microarray analyses commonly use two spectrally distinct fluorescent labels to simultaneously compare different mRNA pools. Signal correlation bias currently limits accepted resolution to twofold changes in gene expression. This bias was investigated by (i) examining fluorescence and absorption spectra and changes in relative fluorescence of DNAs labeled with the Cy3, Cy5, Alexa Fluor 555, and Alexa Fluor 647 dyes and by (ii) using homotypic hybridization assays to compare the Cy dye pair with the Alexa Fluor dye pair. Cy3 or Cy5 dye-labeled DNA exhibited reduced fluorescence and absorption anomalies that were eliminated by nuclease treatment, consistent with fluorescence quenching that arises from dye-dye or dye-DNA-dye interactions. Alexa Fluor 555 and Alexa Fluor 647 dye-labeled DNA exhibited little or no such anomalies. In microarray hybridization, the Alexa Fluor dye pair provided higher signal correlation coefficients (R2) than did the Cy dye pair; at the 95% prediction level, a 1.3-fold change in gene expression was significant using the Alexa Fluor dye pair. Lowered signal correlation of the Cy dye pair was associated with high variance in Cy5 dye signals. These results indicate that fluorescence quenching may be a source of signal bias associated with the Cy dye pair.     

Oligonucleotide conjugates with minor groove ligands as probes for hybridization microarray chips

A possibility of using oligonucleotide conjugates with minor groove ligands as probes for hybridization microarray chips was studied. The oligonucleotide conjugates contain a hairpin ligand (MGB) composed of two tripyrrolcarboxamide residues with an aminocaproic acid residue as a linker and bound to the oligonucleotide duplex AT tract in a site-specific manner. We used as (5′-3′)-probes: GACAAGAp, GACAAAAp, GACAAGA-MGB, and GACAAAA-MGB. The oligonucleotides labeled with the Cy3 cyanine dye, Cy3-ACTAATTTTGTC and Cy3-ACTAATCTTGTC, were used as targets. The maximal MGB effect on the fluorescence level of microarray chip spots, which caused its fourfold increase as compared with the initial unmodified duplex, was observed for the duplex containing only AT pairs in the ligand binding site. The presence of AC and GT mutations in the binding site (imperfect duplexes) or a CG pair (perfect duplex) affect the change in fluorescence level to a considerably lesser degree.     

Oligonucleotide conjugates with minor groove ligands as probes for hybridization microarray chips

A possibility of using oligonucleotide conjugates with minor groove ligands as probes for hybridization microarray chips was studied. The oligonucleotide conjugates contain a hairpin ligand (MGB) composed of two tripyrrolcarboxamide residues with an aminocaproic acid residue as a linker and bound to the oligonucleotide duplex AT tract in a site-specific manner. We used as (5'-3') probes GACAAGAp, GACAAAAp, GACAAGA-MGB, and GACAAAA-MGB. The oligonucleotides labeled with Cy3 cyanine dye, Cy3-ACTAATTTTGTC and Cy3-ACTAATCTTGTC, were used as targets. The maximal MGB effect on the fluorescence level of microarray chip spots, which caused its fourfold increase as compared with the initial unmodified duplex, was observed for the duplex containing only AT pairs in the ligand binding site. The presence of A-C and G-T mutations in the binding site (imperfect duplexes) or a C-G pair (perfect duplex) affects the change in fluorescence level to a considerably lesser degree.     

Quantification of sources of variation and accuracy of sequence discrimination in a replicated microarray experiment

cDNA microarray spot variability arises from many sources, and different systems have varying requirements for achieving the desired level of precision. We determined relative contributions to variance and investigated sequence discrimination using a multiple-array experimental design, with arrays subdivided to determine position and pin effect. Related fragments of 67 resistance gene homologs (RGHs) isolated from Theobroma cacao L. and grouped by sequence similarity were spotted onto arrays, using two of the same RGHs in the fluorescent dye channels (Cy3, Cy5) of the hybridization solution in a "dye-flip" design. A comprehensive statistical model accounted for variability well, giving a coefficient of variation (CV) based on experimental error of 2.12%. Although we were able to separate 85% of RGH group means clearly, some groups more similar to the target were indistinguishable due to nonspecific hybridization. Genetic factors together contributed 72.2% of the total variation, while position and pin effects and their interactions contributed 9.8%. Replication effect was statistically significant. Otherwise, no tests for position effects were significant. The results of the analysis indicate that our Genetic Microsystems 417 arrayer and Affymetrix 428 scanner are performing with sufficient precision, and we produced useful information for planning efficient future experiments.     

Correcting for gene-specific dye bias in DNA microarrays using the method of maximum likelihood

MOTIVATION: In two-color microarray experiments, well-known differences exist in the labeling and hybridization efficiency of Cy3 and Cy5 dyes. Previous reports have revealed that these differences can vary on a gene-by-gene basis, an effect termed gene-specific dye bias. If uncorrected, this bias can influence the determination of differentially expressed genes. RESULTS: We show that the magnitude of the bias scales multiplicatively with signal intensity and is dependent on which nucleotide has been conjugated to the fluorescent dye. A method is proposed to account for gene-specific dye bias within a maximum-likelihood error modeling framework. Using two different labeling schemes, we show that correcting for gene-specific dye bias results in the superior identification of differentially expressed genes within this framework. Improvement is also possible in related ANOVA approaches. AVAILABILITY: A software implementation of this procedure is freely available at http://cellcircuits.org/VERA     

Tunable blinking kinetics of cy5 for precise DNA quantification and single-nucleotide difference detection

Fluorescence correlation spectroscopy (FCS) can resolve the intrinsic fast-blinking kinetics (FBKs) of fluorescent molecules that occur on the order of microseconds. These FBKs can be heavily influenced by the microenvironments in which the fluorescent molecules are contained. In this work, FCS is used to monitor the dynamics of fluorescence emission from Cy5 labeled on DNA probes. We found that the FBKs of Cy5 can be tuned by having more or less unpaired guanines (upG) and thymines (upT) around the Cy5 dye. The observed FBKs of Cy5 are found to predominantly originate from the isomerization and back-isomerization processes of Cy5, and Cy5-nucleobase interactions are shown to slow down these processes. These findings lead to a more precise quantification of DNA hybridization using FCS analysis, in which the FBKs play a major role rather than the diffusion kinetics. We further show that the alterations of the FBKs of Cy5 on probe hybridization can be used to differentiate DNA targets with single-nucleotide differences. This discrimination relies on the design of a probe-target-probe DNA three-way-junction, whose basepairing configuration can be altered as a consequence of a single-nucleotide substitution on the target. Reconfiguration of the three-way-junction alters the Cy5-upG or Cy5-upT interactions, therefore resulting in a measurable change in Cy5 FBKs. Detection of single-nucleotide variations within a sequence selected from the Kras gene is carried out to validate the concept of this new method.     

Elimination of laboratory ozone leads to a dramatic improvement in the reproducibility of microarray gene expression measurements

Background  

Environmental ozone can rapidly degrade cyanine 5 (Cy5), a fluorescent dye commonly used in microarray gene expression studies. Cyanine 3 (Cy3) is much less affected by atmospheric ozone. Degradation of the Cy5 signal relative to the Cy3 signal in 2-color microarrays will adversely reduce the Cy5/Cy3 ratio resulting in unreliable microarray data.     

Conditions to ensure competitive hybridization in two-color microarray: a theoretical and experimental analysis

We derived a theoretical model that explains certain biases observed in the two-color microarray hybridization experiments reported in the literature. We show that true competition is achieved only when the hybridization kinetics of the two differentially labeled probes are the same. If the hybridization kinetics of the two differentially labeled probes is different, which can occur when the labeling and hybridization conditions for the two probes are dissimilar, then differential expression observed becomes a function of the amount of the target (i.e., DNA spotted on the slide). We use this model to validate the microarray methodology by determining the differential expression of four select Arabidopsis genes and two human genes (beta-actin and GAPDH) as a function of the amount of target arrayed. We show through both modeling and experiments that the rate constants for Cy5- and Cy3-labeled probes are the same under our exrimental conditions. Therefore, the target concentrations need not greatly exceed the probe concentration. It is obvious from the data presented that a simple treatment of an individual hybridization rate calculation does notfully describe what is occuring in today's complex, multispecies experiments. The method of validation is easily implemented to ensure data reliability by two-color microarray.     

The influence of RNA integrity,purity and cDNA labelling on glass slide cDNA microarray image quality

The accuracy of gene expression measurements generated using cDNA microarrays is dependent on the quality of the image generated following hybridization of fluorescently labelled cDNA. It is not known how this image is influenced by sample preparation factors which such as RNA quality, cDNA synthesis and labelling efficiency. In this study we used a simple metric based on the ratio of the total feature (F) and background (B) fluorescence, which correlates with the visual assessment of 60 microarray images, to determine the influence of sample preparation on image quality. Results indicate that RNA purity (A260/A280) and integrity (18S:28S ratio) do not strongly influence microarray image quality. cDNA having an nucleotide to dye ratio greater than 100 produced poor microarray images, however, cDNA labelled more efficiently was not a guarantee of a better image. The data also indicate that the array image quality is not improved by loading more cDNA into the hybridization mixture however poor image quality did result from a disproportionate amounts of Cy5 and Cy3 labelled cDNA. This study provides insight into the source of variation in microarray image analysis introduced during sample preparation and will assist in the standardisation of cDNA glass slide microarray protocols.     

Statistical methodology for the analysis of dye-switch microarray experiments

Background  

In individually dye-balanced microarray designs, each biological sample is hybridized on two different slides, once with Cy3 and once with Cy5. While this strategy ensures an automatic correction of the gene-specific labelling bias, it also induces dependencies between log-ratio measurements that must be taken into account in the statistical analysis.     

Sequence-dependent fluorescence of cyanine dyes on microarrays

Cy3 and Cy5 are among the most commonly used oligonucleotide labeling molecules. Studies of nucleic acid structure and dynamics use these dyes, and they are ubiquitous in microarray experiments. They are sensitive to their environment and have higher quantum yield when bound to DNA. The fluorescent intensity of terminal cyanine dyes is also known to be significantly dependent on the base sequence of the oligonucleotide. We have developed a very precise and high-throughput method to evaluate the sequence dependence of oligonucleotide labeling dyes using microarrays and have applied the method to Cy3 and Cy5. We used light-directed in-situ synthesis of terminally-labeled microarrays to determine the fluorescence intensity of each dye on all 1024 possible 5'-labeled 5-mers. Their intensity is sensitive to all five bases. Their fluorescence is higher with 5' guanines, and adenines in subsequent positions. Cytosine suppresses fluorescence. Intensity falls by half over the range of all 5-mers for Cy3, and two-thirds for Cy5. Labeling with 5'-biotin-streptavidin-Cy3/-Cy5 gives a completely different sequence dependence and greatly reduces fluorescence compared with direct terminal labeling.     

Simple and effective method for generating single-stranded DNA targets and probes

A simple and efficient PCR method was developed for generating dye- or radiolabeled single-stranded DNA targets or probes used for hybridization studies. The method involved the use of a pair of long primers with high annealing temperatures and a short, labeled primer with a low annealing temperature in a PCR consisting of two cycles at different temperatures. We used this method to generate dye Cy 5-labeled and [32P]-radiolabeled single-stranded DNA targets and probes. These labeled probes were used successfully for the microarray identification of point mutations in Mycobacterium tuberculosis genes and for the Northern blot detection of expression changes of the GATA-2 gene in Pneumocystis carinii-infected rat lungs.     

Comparison of hybridization behavior between double and single strands of targets and the application of asymmetric PCR targets in cDNA microarray

Double stranded targets on the cDNA microarray contain representatives of both the coding and noncoding strands, which will introduce hybridization competition with probes. Here, the effect of double and single strands of targets on the signal intensity and the ratios of Cy5/Cy3 within the same slide were compared. The results show that single stranded targets can increase the hybridization efficiency without changing the Cy5/Cy3 ratio. Based on these results, a new strategy was established by generating cDNA targets with asymmetric PCR, instead of conventional PCR, to increase the sensitivity of the cDNA microarray. Furthermore, the feasibility of this approach was validated. The results indicate that the cDNA microarray system based on asymmetric PCR is more sensitive, with no decrease in the reliability and reproducibility as compared with that based on conventional symmetric PCR.