Given the diversity and inconsistency of methods used to prepare microarrays, adapting a labeling protocol for uniform and reproducible microarray hybridization is a challenge. Arrays are printed on a variety of substrate surfaces (Nylon, poly-L-lysine, aminosilane, and aldehyde, etc.) using any of a number of different commercially available spotters. Different spotting solutions as well as nucleic acids "probe" concentrations have been tried. The probes can consist of either a near full-length cDNA or a specifically chosen oligonucleotide. While it would seem that the protocol for hybridization to oligonucleotide arrays should not be much different from the protocol for cDNA arrays, there are some fundamental differences that must be accommodated to obtain optimal signal. For example, oligonucleotide probes are shorter (25-80 bases long) and offer more gene specificity while cDNAs are longer (average 0.2-2.0kb) and can bind some targets non-specifically. This brief data summary describes the Submicro Oligo kit, which represents the adaptation of Genisphere's Submicro labeling system to oligonucleotide arrays.

Materials and Methods:

OpArray Stress and Aging arrays were purchased from Operon Technologies Inc. These arrays consist of 70 nucleotide long oligonucleotides spotted onto poly-L-lysine coated glass slides. Total RNA from Human Heart and Human Kidney was purchased from Ambion. All other components used were part of the Genisphere Submicro Oligo Kit.

Reverse Transcription:

Hybridization and Washing:

Because of the differences between oligo and cDNA arrays initial experiments on the OpArrays consisted of determining the optimal hybridization conditions. Using Vial 6 hybridization buffer as the standard buffer, the hybridization temperature was varied from 40°C to 60°C at 5°C intervals. Two different total RNA samples, one labeled with Cy™3 and the other Cy^TM5, were used to prepare cDNAs for hybridization and signal intensity, probe specificity and differential magnitude were monitored to determine the appropriate hybridization temperature. Each array was scanned and the data overlaid for analysis (similar to figure 1b). As the temperature of hybridization increased from 40°C to 60°C, the overall signal intensity decreased as expected and the number of elements demonstrating differential increased. Interestingly, we also observed that some features initially demonstrated equal signal (represented by a yellow composite image) at low hybridization temperature but then either changed to a differential (red or green composite image) or completely disappeared. This was probably due to non-specific hybridization of the transcribed cDNAs to the oligos on the array. We also observed that some features that were not initially represented at high temperature yielded a differential at lower temperatures. After performing the negative control experiment, consisting of Cy^TM3 and Cy^TM5 3DNA dendrimers alone (without cDNA) hybridized at each temperature, we determined that the differentials observed at lower temperature at features that do not demonstrate signal at higher temperatures was due to non-specific binding of the 3DNA capture sequence to those elements. The negative control and hybridization temperature titration experiments were repeated using an oligonucleotide that is complementary to the capture sequence as a "capture sequence blocker" (Vial 9) after the prehybridization of cDNA to the 3DNA reagents. The non-specific binding of the 3DNA capture reagents via their capture arms was completely blocked by the "capture sequence blocker" (Vial 9). As a result of these experiments, the optimal hybridization temperature for the tested lot of Stress and Aging OpArrays was determined to be between 50°C and 55°C (data not shown). We also tested a second lot of OpArrays, which demonstrated an optimal hybridization range of 47°C to 52°C. We have concluded that this may be due to differences in either the spotted oligos or age of the array. In general this suggests that for each lot of "ideally" similar arrays, the hybridization conditions should be optimized prior to setting up a complete comparative analysis of critical samples. RNA titration experiments were also conducted and, in general, the best results were obtained using 1-2.5 ug of high quality total RNA.

Figure 3
Click here for close-up of Images

In a parallel set of experiments two different total RNA samples were used in differential expression experiments. Total heart (Cy^TM3) and kidney (Cy^TM5) RNA were reverse transcribed using the appropriate Genisphere primers and the transcribed cDNAs combined into a single hybridization. The composite overlay image is presented in Figure 1, panel b. A similar experiment was done using direct labeling in order to compare differential data. A portion of the data from the composites images of the same sample and different samples (Figure 1, panels a and b, respectively) in each channels is magnified for closer observation and 3 features specifically identified for discussion. It is clear from the differential experiment that certain messages were represented in either one or both RNA samples. Feature 1 which demonstrates a high level expression in kidney (Cy^TM5, red) relative to heart. Observe from the same sample data that little or no signal is visible for feature 1. The differential expression in kidney relative to heart calculates to be 81.6 for this gene (Figure 3). Feature 2 returned a green color after overlay indicating a higher level of expression in heart relative to kidney. This differential expression calculated to be 6.5 heart relative to kidney (Figure 3). Near equal expression (0.83, Figure 3) was observed for feature 3 as indicated by the yellow feature color. Figure 2, panel B represents the scatter plot of the specific signal from each channel. Approximately 1000 elements were included in the calculations after flagging data at less than 3 fold above background and saturation. For comparison purposes the solid lines indicating 3 standard deviations of variance were transposed from the same sample scatter plot. The relative expression, kidney (Cy^TM5) to heart (Cy^TM3), of each feature was calculated and ranged from an 81.6 over expression to a 50 fold under expression in kidney relative to heart. The differential data from the magnified area of Figure 1 is identified in Figure 3 and the corresponding kidney to heart differentials from the direct incorporation experiment included. These data indicate that a range of differential expression was observed and that the Submicro and direct incorporation methods closely agreed in both the direction and relative magnitude of the differentials, although that the differentials for the direct incorporation were somewhat compressed compared to submicro labeling. To confirm that there is minimal channel bias during the reverse transcription, labeling, and hybridization, we "reversed" the RNA samples during the labeling process (Figure 3, panel b). Visually, the features that were green in the forward experiment changed to red in the reverse labeling experiment. The converse is also true. Features demonstrating equal expression in both channels remained unchanged.

Conclusions:

1. The Genisphere Submicro Oligo kit works well with 1-2.5ug of input total RNA.
2. The hybridization conditions will depend on the length of the spotted probe and age of the array. Hybridization temperatures vary from 35-42°C for 30mer oligos and 45-50°C for 50mer oligos to 50-55°C for 70mer oligonucleotides.
3. Since arrays can vary from lot to lot the hybridization conditions need to be optimized for each lot of oligo arrays.
4. Studies in which the same RNA sample is labeled with both Cy™3 and Cy^TM5 indicate that a 40% change in relative expression can be measured with 99.5% confidence.
5. The differential expression of genes can be reproducibly be calculated from arrays hybridized using the Submicro Oligo kit and these data agree with other standard labeling techniques. There is also little or no bias when labeling an RNA sample with Cy^TM3 relative to Cy^TM5.