Freeze-thaw cycles and nucleic acid stability: what’s safe for your samples?

Sept. 2, 2015

The common starting point for most molecular testing is in the preparation of nucleic acid extracts—DNA, RNA, or both—from a tissue sample of interest. In most cases the amount of extract prepared is more than enough needed for requested initial sample testing, so what is done with the excess? The answer to this, of course, depends on individual institutional rules, but a common scenario is to maintain the spare extract for at least short-term storage, and have it available for any additional or repeat testing requested on the original sample. Beyond this short term, the extracts may either be destroyed, or perhaps classified as residual excess sample. In the latter case, the sample is then likely to undergo some form of de-identification, with retention of metadata relating to specimen positivity or negativity for tests employed. Banks of such residual, de-identified samples with known test results for specific markers are extremely valuable as control material in the development and validation of new test procedures.

The question of stability

Whether your laboratory only keeps extracts for the short-term use in additional or re-testing, or longer term as a control substance, it’s a safe assumption that they’re not just stored as aqueous solutions at ambient conditions. Nobody expects DNA (and much less, RNA) to remain intact for very long in that form; oxidative reactions, spontaneous hydrolysis, and even the potential for microbial contamination and subsequent digestion of the sample all would be expected to contribute to sample degradation. The most common approach to increasing longevity of nucleic acid extracts is to freeze the samples to -20C (DNA samples) or -80C (RNA samples).

This approach is essentially universal, and just as common is the anecdotal knowledge passed on to each new member of the laboratory staff that repeated cycles of thawing and re-freezing nucleic acid samples leads to sample degradation. While the most junior member of your molecular testing lab will be able to tell you this, you’ll also probably discover that even the most senior and knowledgeable lab members can’t tell you how much a sample will degrade per cycle, or have any evidence-based justification for what is an acceptable number of times to freeze and thaw a sample.

Instead, there is probably a policy of aliquoting larger samples into smaller portions prior to freezing, and rules that each sample can be used an arbitrary number of times (say, five). This type of policy has likely been applied since Dr. X put it in the protocol for the lab’s first molecular assay in 1998, and its been shown to be valid in some contexts, such as qPCR control standards. “For years, we’ve never noticed any consistent trend in Ct value changes over five freeze-thaw cycles,” the defenders of such a policy might say. And nobody quite knows what to expect if you go to six freeze-thaw cycles; there’s a lurking concern that a Cinderella-at-midnight event might occur.

How degrading: three important mechanisms

If we want to know more than this, we should start by considering what three of the more important mechanisms are by which the freeze-thaw process can degrade nucleic acids. For each, we’ll also simultaneously consider some protective measures.

pH. Nucleic acids are most stable at slightly alkaline conditions, approximately pH 7.5 to 8. If the environment is more acidic (low pH), acid depurination is promoted, and if it’s more alkaline, hydrolysis of the sugar-phosphate backbone increases. Unbuffered water, of course, has a nominal pH of 7 at 25C. As water freezes however, pH microenvironments can transiently occur, particularly if there are dissolved solutes. As the freezing process itself occurs, micro-regions of pure water will crystalize first, as the solutes locally suppress the freezing point. As pure ice crystals form, the solutes are effectively crowded into a smaller remaining liquid volume and their effective concentration increases. With this change in solute concentration can come shifts in pH, impacting nucleic acid strands passing through the liquid zones. Note that this microenvironment effect occurs both during the freezing and thawing steps. Having buffers such as Tris present can help protect against this; while these solutes also undergo transient concentration shifts during freezing and thawing, the ratio of acidic to basic forms of the buffer pair should remain nearly constant and help prevent pH swings in the resulting microenvironments.

Physical shearing by ice crystals. One hypothesized mechanism for nucleic acid damage during freezing is mechanical in nature. It’s thought that the formation of ice crystals can in some cases actually “cut through” nucleic acid strands. One might assume that such an effect is more likely to occur on long nucleic acid molecules, with one end trapped in an already frozen matrix and subjected to distal physical stress from further water crystallization, than short ones, which might be more able to rapidly shift conformation to adapt to the structuring of solvent. Evidence for this form of damage is derived partly from experimental observation that inclusion of cryoprotectants such as glycerol, which break up the extreme rigidity of pure water ice, correlate with better retention of DNA quality after freezing. (See Reference 1 below, for example).

Oxidative damage catalyzed by metal ions. Free radicals—highly reactive species looking to gain or donate single unpaired electrons—can be generated in aqueous environments through processes such as the Fenton reaction. Even highly purified nucleic acid preparations are reported in literature (see Reference 2) to contain potentially dangerous levels of Fe2+ ions. The relevance of this to freeze-thaw cycles is via the same “microenvironment” mechanism as discussed above for pH. The concentrations of iron or other potentially damaging metal ions and their concomitant free radicals can locally spike in trapped aqueous pockets during the freezing and thawing process, exposing nearby nucleic acid strands to damage. Protection from this damage mechanism is probably best afforded by use of chelating agents, which trap the metal ions in a less reactive “shell.”

Taken together, it would appear that storing nucleic acids in incompletely purified state with excessive metal ions, in unbuffered water, and without cryoprotective agents, is probably the worst-case scenario for stability in the face of repeated freeze-thaw cycles. Conversely, a low ionic-strength, slightly alkaline buffered environment (say, Tris-HCl at pH 7.8 to 8), with a chelating agent (say, EDTA) with an added cryoprotectant such as glycerol or DMSO, is probably the safest matrix in which to store a nucleic acid which will be frozen and thawed. This may sound familiar—yes, the common lab standby for DNA solutions, “TE8” at 10mM Tris-HCl, 1 mM EDTA, pH 8.0, is already most of the way there. Add some glycerol or DMSO, and you’ve done about as much as you can to insulate the nucleic acid from injury during freezing and thawing.

How much is too much? It depends.

Of course, even that matrix isn’t perfect, and some damage will still occur. How much damage—and whether it’s “too much” for the intended purpose of the archived nucleic acid—will depend on a number of factors, including purity of the extract and buffer components, endogenous polysaccharides or proteins able to act as cryoprotectants, and rate of freezing. The intended purpose angle is also important, inasmuch as the forms of damage which occur are stochastic in nature; that is, they are more likely to occur across a long section of nucleic acid than a very short one. If the frozen and thawed material is going to be used as a template for a 15,000-base pair (15 kb) long-range PCR, it’s a lot more likely to show evidence of damage than if a 150 bp sub-region is all that’s being detected. (In fact, all else being equal, one would assume a linear 100x difference in chance of damage to each template region).

These differences in nucleic acid effective stability, arising from differences in sample purity that are hard to discern and in the exact parameters of the assay the sample will be used on, mean that it is virtually impossible to make any blanket statements about how many freeze-thaw cycles are “too many.” What’s acceptable, and what’s not, will depend very much on the type of extractions prepared and their intended final use.

If we attempt to search for definitive statements on rates of DNA or RNA decay on exposure to freeze-thaw cycles, this range of possible answers is reflected in the literature. It’s not that definitive statements can’t be found; for example, Reference 3 reports that for long genomic DNAs, regardless of initial sample size distributions, all DNAs fragmented down to an average of 25 kb within 18 freeze-thaw cycles. Reference 4 examined three different qPCR standard materials and showed losses of 17 percent, 45 percent, and 50 percent detectable content respectively after only four freeze-thaw cycles in plain water matrix (with losses increasing to 76 percent, 95 percent, and 96 percent by 16 cycles). Note the template-specific differences in apparent loss here, a clear illustration of our consideration above of assay-specific “intended purpose” issues.

To even further illustrate the range in answers reported in the literature on this topic, consider that Reference 5, which looked at RNA (which one would expect to show less intrinsic stability than DNA) via a bDNA hybridization capture method, showed a very weak statistical trend towards more detectable nucleic acid per cycle—not less! The authors in that case provide “no definitive explanation,” although one might perhaps surmise that bDNA hybridization, as opposed to PCR amplification, might be less susceptible to strand breakage, causing loss of signal. Further reference examples can be found but do little to clarify a generic rule.

So the takeaway is….

Where does all of this leave you and your lab? In the first place, there are both data and mechanisms supporting the belief that in plain water as a matrix, nucleic acid is sequentially degraded by cycles of freezing and thawing. Depending on what your downstream assays will support, buffers, chelating agents, and cryoprotectants in your nucleic acid storage media will help. Beyond that, it makes a good case that if you want to progress beyond “Dr. X’s 1998 recommendations” (or your local version thereof) in your lab, you probably need to put a handful of your representative samples to the test in the context of your particular assay, to know what’s an acceptable number of freeze-thaw cycles for your purpose.


  1. Röder B, Frühwirth K, Vogl C, Wagner M, Rossmanith P. Impact of long-term storage on stability of standard DNA for nucleic acid-based methods. J Clin Microbiol. 2010; 48(11):4260-4262.
  2. Oxford Genome Technologies. DNA Storage and Quality. August 2011. Accessed July 18, 2015.
  3. Shao W, Khin S, Kopp WC. Characterization of effect of repeated freeze and thaw cycles on stability of genomic DNA using pulsed field gel electrophoresis. Biopreservation and Biobanking. 2012;10(1):4-11.
  4. Schaudien D, Baumgärtner W, Herden C. High preservation of DNA standards diluted in 50 % glycerol. Diag Mol Pathol. 2007:16(3):153-157.
  5. Krajden M, Minor JM, Rifkin O, Comanor L. Effect of multiple freeze-thaw cycles on hepatitis B virus DNA and hepatitis C virus RNA quantification as measured with branched-DNA technology. J Clin Microbiol. 1999.37(6):1683-1686.