Ambion’s new RecoverAll™ Total Nucleic Acid Isolation Kit (patent pending) is designed for quantitative recovery of RNA and/or DNA from FFPE samples. The nucleic acid from the FFPE samples includes the full complement of microRNA (miRNA). These miRNAs can be used for miRNA expression profiling when used in combination with the flashPAGE™ Fractionator System and mirVana™ miRNA Array Technology.

Isolate microRNA from FFPE Samples

Kidney, liver, and brain tissues from 4 mice were harvested, and half of each tissue was immediately flash-frozen, while the other half was formalin-fixed and paraffin-embedded following a procedure commonly used in hospitals [1]. Total RNA was isolated from frozen tissues using the mirVana miRNA Isolation Kit and from FFPE samples using the RecoverAll Total Nucleic Acid Isolation Kit, which is partly based on mirVana technology. miRNA was isolated and concentrated from each sample using the flashPAGE Fractionator and the flashPAGE Clean-up Kit.

Total RNA (1 µg) from representative tissue samples was also analyzed on a Northern blot hybridized with a probe specific for let-7 (Figure 1). The recovery and quality of the let-7 miRNA in the FFPE and frozen samples were very similar, indicating that the new FFPE RNA isolation procedure was effective in recovering intact miRNA.


Figure 1. let-7 miRNA from FFPE and Flash-frozen Mouse Tissue Samples. Total RNA was isolated from four different mouse tissues that had been frozen (F) or formalin-fixed and paraffin-embedded (Z). This Northern blot (1 µg RNA/sample) was incubated with an oligonucleotide probe specific for let-7 and verified the presence of intact let-7 miRNA.

Obtain Reproducible miRNA Array Results

A pool containing miRNA from 5 different tissues was created as a means of comparing individual samples to each other. Using the mirVana miRNA Labeling Kit, miRNA from the pooled sample was labeled with Cy™3, and miRNA from individual FFPE or frozen tissue samples was labeled with Cy5. Each Cy5-labeled miRNA sample was co-hybridized on a separate miRNA array with a portion of the Cy3-labeled pooled miRNA sample. The relative Cy dye signal intensities were compared for each miRNA, and hierarchical clustering showed that miRNA profiles were nearly identical between FFPE and frozen tissues (Figure 2). Correlations of 93.5%, 95.9%, and 97.8% were obtained for kidney, liver, and brain tissues, respectively, by comparing the average Cy5/Cy3 (R/G) ratio of the fixed samples to that of the frozen samples (Figure 2).


Figure 2. Excellent Correlation of miRNA Expression Profiles from FFPE and Flash-frozen Samples. For each tissue type, the R/G Normalized (Mean) ratio of each set of 4 fixed samples and 4 frozen samples was averaged. The average R/G ratio of the FFPE sample (X axis) was plotted against the average R/G ratio of the frozen sample (Y axis) to determine the correlation between miRNA profiles generated from fixed tissue samples as compared to frozen tissue samples (A. Kidney, B. Liver, C. Brain). D. Representative heat maps from fixed (FFPE, left) and frozen (right) samples from brain.

Detect Differentially Expressed miRNAs

The goal for most miRNA array analyses is to identify the miRNAs that are differentially expressed between samples. To confirm that cross-comparisons of FFPE samples provide the same differential expression data as frozen samples, we compared miRNA array data from both the FFPE and frozen samples for kidney and brain tissues. The expression ratio of the FFPE kidney versus FFPE brain was compared to the expression ratio of the frozen kidney versus frozen brain (Figure 3). The correlation between the FFPE tissue comparison and frozen tissue comparison exceeded 95%. The list of miRNAs that are estimated to be two-fold differentially expressed between kidney and brain samples was very similar for both frozen and FFPE sample comparisons (Figure 3).


Figure 3. Similar Set of miRNAs Were Found to Be Differentially Expressed in Different Tissues in Fixed and Frozen Samples. The R/G Normalized (Mean) ratio of each set (n=4) of fixed and frozen samples was averaged for kidney and brain tissues. Panel A. The kidney to brain expression ratio for both fixed and frozen samples was determined, and the expression ratio of the fixed kidney vs. fixed brain (X axis) was plotted against the expression ratio of the frozen kidney vs. frozen brain (Y axis). Panel B. Representative examples of miRNA differentially express by at least 2-fold in kidney vs. brain are shown.

Examine Global miRNA Expression in FFPE Tissues

Our data show that miRNA profiles are maintained and are not significantly affected by the fixation process. This introduces a significant opportunity for using archived samples to detect changes in miRNA expression levels in various disease states, with ramifications for defining regulatory roles of miRNA in biological processes.

Scientific Contributors

Emily Zeringer, Patricia Powers, Jaclyn Shingara, Kerri Keiger, Tim Barta, Rick Conrad, David Brown • Ambion, Inc.