A novel HPLC method reveals that precipitation of 2′-deoxyadenosine 5′-monophosphate with lithium perchlorate/acetone leads to base depurination

Tissue specimens for pathology examination are commonly preserved by formalin fixation and paraffin embedding (FFPE). FFPE samples are frequently amenable to protein analyses, but are less useful for nucleic acid analyses. Treatment with formaldehyde and organic solvents are believed to cause nucleic acid adduction and fragmentation, which reduce nucleic acid recovery and cause PCR amplification errors [1]. Our group and others have studied the chemical modifications to mononucleotides and short oligonucleotides that occur during the first steps of fixed tissue processing, specifically formalin treatment and ethanol dehydration. Adducts such as methylol (hydroxymethyl) groups and methylene-bridge cross links were shown to be reversible upon heating in formalin-free buffers [2,3]. However, these studies also revealed 1–3% base depurination of formalin-fixed ethanol-dehydrated 2′-deoxyadenosine-5′-monophosphate (dAMP) [2]. This modification is irreversible and represents lost genetic information, which would prevent the identification of nucleic acid changes associated with disease.

The aim of this study was to identify a simple HPLC method to resolve dAMP and its decomposition products and to use this method to identify the source of dAMP base depurination. There are many HPLC procedures available for the separation of nucleotides or for the separation of uncharged nucleosides and bases. Anion exchange systems are common for the former while reversed-phase systems are common for the later. However, few HPLC methods have been described that resolve both charged and uncharged mononucleotides, nucleosides, and bases [4]. These methods typically employ an ion-pairing agent on reversed-phase [5,6] or anion exchange columns [2]. In this study we demonstrate that a versatile nitrile column with an ion pairing agent can successfully resolve dAMP from its charged and neutral decomposition products. This novel HPLC method was used to show that precipitation of dAMP from lithium perchlorate in acetone at −20 °C can cause base depurination of dAMP.

Chromatography was performed with an Agilent 1100 series HPLC system (Santa Clara, CA) using an Agilent Zorbax SB-CN column (5 µm, 4.6 × 150 mm). The nitrile column was run in ion-exchange mode by pre-equilibration with 20 mM triethylammonium acetate (TEAA) at pH 7.4. The column was then eluted over 20 min at 1 mL/min using a gradient of 0 to 10% acetonitrile in 20 mM TEAA, while monitoring absorbance at 260 nm. The following chromatography standards were purchased from Sigma-Aldrich (St. Louis, MO): dAMP, adenosine 5′-monophosphate (AMP), adenosine (Ado), 2′-deoxyadenosine (dAdo), and adenine (Ade). Retention times were as follows: AMP, 5–6 min; dAMP, 7–8 min; Ado, 13.5–14 min; Ade, 15 min; and dAdo, 17–17.5 min (Fig. 1). The corresponding peak widths were typically 0.15 to 0.2 min for AMP, dAMP, and Ado; and 0.2–0.3 min for Ade and dAdo. It was necessary to inject AMP formulated in 20 mM TEAA to avoid earlier eluting artifacts. Serial dilutions indicated that the limit of detection was 0.5% Ade relative to nucleotide (data not shown). This was due to impurities in the chromatographic standards rather than the limits of instrumental detection.

HPLC analysis of adenine-containing species. AMP represents adenosine 5′-monophosphate, Ado is adenosine, Ade abbreviates adenine base, the d- prefix indicates 2′-deoxy analogs, and MeCN is an acronym for acetonitrile. The top chromatogram shows standards, while the bottom two chromatograms show dAMP standards precipitated by different methods. The circled peak of the middle chromatogram was identified as Ade.

This HPLC method was applied to determine the extent of adenosine depurination following formalin fixation and storage under ethanol. To prepare formalin-adducted nucleotide monophosphates (FA-NMP), 5 mM nucleotide in 20 mM potassium phosphate buffer, pH 7, was mixed with an equal amount of methanol-free formalin (Polysciences Inc.; Warrington, PA) and allowed to react for 1 h at room temperature (RT). The phosphate was adjusted to 50 mM and 10 volumes of ice-cold acetone were added. The solution was mixed and left at −20 °C for 15 min and then pelleted at 1400 ×g on an Eppendorf 5810R centrifuge. Pellets were washed twice with cold acetone and dried under 29 mm Hg vacuum at 40 °C for 30 min. Samples were dissolved in antigen retrieval (AR) buffer composed of 10 mM Tris-acetate, 1 mM EDTA, pH 4, and frozen. Some dried pellets were covered with 100% ethanol for 1 h at RT prior to being pelleted, dried, and dissolved in AR buffer as described above. Solutions were heated at 70 °C for 30 min prior to HPLC analysis.

When the procedure above was followed for AMP and dAMP without heating at 70 °C, some 60% of the sample eluted in the void volume of the column (about 2.5 min), consistent with the formation of formalin adducts. Heating at 70 °C for 30 min resulted in peaks that co-eluted with AMP and dAMP. However, peaks consistent with Ade were absent, suggesting that there was no base hydrolysis. This was inconsistent with our previous study, which revealed 1–3 mol% base hydrolysis under the same conditions [2]. The only clear difference between the current study and our earlier work [2] was that here we used neat acetone, rather than 2#x00025; w/v lithium perchlorate (LiClO₄) in acetone, for the precipitation of the formalin adducted nucleotides. LiClO₄/acetone is frequently used to precipitate nucleic acids and nucleotides in order to obtain a hard dehydrated white pellet and to increase sample recovery [7–10]. We compared the adenine base hydrolysis from dAMP standards when precipitated from acetone with, or without, LiClO₄. Fig. 1 demonstrates that base was released in the LiClO₄–containing precipitation, but not for neat acetone. In 20 experiments, we always observed base hydrolysis from dAMP when precipitated in 2% LiClO₄/acetone, however, the level of hydrolysis varied from 1–15%. In contrast, dAMP precipitated in neat acetone did not show any base hydrolysis in 20 parallel experiments. These experiments were repeated with AMP. There was no base hydrolysis following precipitation with acetone, but with 2% LiClO₄/acetone there was 0–2% base hydrolysis. Interestingly, when dAMP was incubated in a solution of 2% LiClO₄ in 10 mM phosphate buffer, pH 7.4, there was no base hydrolysis for time periods up to 14 hr. The degree of base hydrolysis was not affected by ethanol treatment, heating, or changing the AR buffer to water or 10 mM potassium phosphate, pH 7.4. The specific mechanism responsible for adenosine depurination is not clear and is currently under investigation. However, LiClO₄/acetone has been shown to promote similar rearrangements and eliminations in other molecules [11, 12]. It is possible that the variability seen in perchlorate-mediated adenosine depurination arose from differences in the pellet’s size and composition in each tube during centrifugation.

In conclusion, we have demonstrated an HPLC method to separate AMP and dAMP from their corresponding nucleosides and adenine base using a common ion pairing agent and an analytical nitrile HPLC column. Our study has also identified LiClO₄/acetone as the source of the adenosine hase depurination seen in our previous study of formaldehyde adducted dAMP [2]. Thus, despite the advantages of 2% LiClO₄/acetone precipitation of nucleic acids and nucleotides, caution must be employed in its use due to the possibility of base depurination of dAMP.