The usual answer is to flash-freeze tissue samples in liquid nitrogen and then store them at –80°C. Although effective, this technique is problematic and often not practical. Ambion’s RNAlater provides a more convenient method for maintaining tissue for short-term and long-term storage. It prevents RNA degradation and preserves RNA profiles using less extreme temperatures. Tissue samples that are stored at room temperature in RNAlater provide accurate gene expression results after 72 hours of storage, compared to RNA isolated from the same samples either immediately after collection or after –80°C storage for equivalent times (Mutter GL, Zahrieh D, Liu C, Neuberg D, Finkelstein D, Baker HE, Warrington JA. (2004) Comparison of frozen and RNAlater solid tissue storage methods for use in RNA expression microarrays. BMC Genomics 5:88–94.). But RNAlater also provides much longer storage possibilities. When keeping the immersed samples at –20°C, RNAlater usually remains liquid, making the tissue much easier to handle, and the RNA population remains stable and unchanged for more than 2 1⁄2 years.

RNA From Tissue Stored Long-term in RNAlater

A long-term storage test was performed on dissected mouse tissues to examine storage effects on gene expression using microarray analysis. Dual mice (CD-1 albinos, Harlan Bioproducts) were sacrificed and heart and brain were split, with one half quick-frozen and the other placed in RNAlater. Frozen samples were placed at -80°C and those in RNAlater were placed at -20°C after several hours at 4°C (per RNAlater Protocol). Samples were stored for 2 years 7 months.

RNA isolation and assessment. RNA was then prepared using the mirVana™ RNA Isolation Kit. Yields were equivalent for each tissue, whether frozen or stored in RNAlater, with heart yielding less RNA than brain. Analysis on the Agilent 2100 bioanalyzer provided RNA Integrity Numbers (RINs) all greater than 9 (see sidebar, What is RIN?).

Array analysis. For array analysis, RNA (1 µg) from each heart sample (biological replicates) was amplified using the MessageAmp™ II procedure and the cRNA was hybridized to Affymetrix GeneChips (Mouse Genome 430A 2.0). For the brain samples, RNA from only one of the organs was used; duplicate aliquots were processed to provide technical replication of amplification and array analysis.

All eight samples generated Present calls of approximately 60% (Figure 1). There were no consistent trends in the level of Present calls between the two methods of storage. Comparison of concordance between samples, shown in Figure 2, demonstrated a very high correlation of data generated from frozen and RNAlater samples. Actual concordance values are provided in Figure 3. For brain, where all samples were from the same tissue sample, the correlation between frozen and RNAlater replicates (R=0.996 and 0.997, respectively) was virtually the same as the average of that between all the possible frozen-RNAlater comparisons (R=0.994). For the heart samples, where samples 1 and 2 were from biological replicates, the frozen-RNAlater correlations (R=0.975 for heart-1 and 0.981 for heart-2) were similar to the correlation between the frozen (0.976) or RNAlater-treated (0.984) tissues. More telling, as can be seen in Figure 2, the core correlation patterns for each brain series overlap, while the points furthest from the R=1 line do not overlap (i.e., do not trend in same direction), indicating that there is no systematic bias introduced by storage in RNAlater. This result firmly demonstrates that use of RNAlater for extended storage is equal to the established gold-standard of storage at -80°C when considering the quality of RNA expression results. When ease of use and ease of handling of the tissue samples for RNA extraction are considered, RNAlater is truly superior.


Figure 1. Percent Present Calls for RNA From Samples Stored Frozen or in RNAlater. RNA (1 µg) from each heart sample (A. biological replicates) and each brain sample (B. technical replicates) was amplified using the MessageAmp™ II procedure and the cRNA was hybridized to Affymetrix GeneChips. Present calls were determined using Affymetrix software.



Figure 2. Concordance Comparison Between Frozen and RNAlater-stored Samples. Further analyses of the arrays described in Figure 1 show a comparison of concordance between samples by overlap for technical replicates (A) and biological replicates (B).



Figure 3. Correlations For Samples Stored Frozen or in RNAlater.