Measurement of Postreplicative DNA Metabolism and Damage in the Rodent Brain

Abstract

The DNA of all organisms is metabolically active due to persistent endogenous DNA damage, repair, and enzyme-mediated base modification pathways important for epigenetic reprogramming and antibody diversity. The free bases released from DNA either spontaneously or by base excision repair pathways constitute DNA metabolites in living tissues. In this study, we have synthesized and characterized the stable-isotope standards for a series of pyrimidines derived from the normal DNA bases by oxidation and deamination. We have used these standards to measure free bases in small molecule extracts from rat brain. Free bases are observed in extracts, consistent with both endogenous DNA damage and 5-methylcytosine demethylation pathways. The most abundant free base observed is uracil, and the potential sources of uracil are discussed. The free bases measured in tissue extracts constitute the end product of DNA metabolism and could be used to reveal metabolic disturbances in human disease.

Graphical Abstract

INTRODUCTION

The DNA of higher organisms is composed predominantly of the canonical bases adenine, thymine, guanine, and cytosine. In 1948, 5-methylcytosine (5mC) was identified in calf thymus DNA.¹ It is formed by enzymatic methylation of a cytosine residue in DNA following DNA replication.^2,3 Subsequent studies established that 5mC codes as cytosine during DNA replication,⁴ but the presence of 5mC profoundly alters DNA–protein interactions.^5,6 The location of 5mC residues in DNA establishes a cytosine methylation pattern that is one of the key elements of the epigenetic programming of gene transcription in eukaryotes.^7,8

Enzymatic demethylation of 5mC in DNA has been proposed by multiple groups.⁹ However, evidence in favor of such a pathway was recently supported with the identification of 5-hydroxymethylcytosine (5hmC) in the DNA of rodent cerebellar Purkinje neurons.¹⁰ Members of the TET family of α-ketoglutarate-dependent dioxygenases have been shown to convert 5mC to 5hmC in DNA.¹¹ Further details of this pathway are less clear, but evidence has been reported^12–14 indicating further oxidation of 5hmC to 5-formylcytosine (5foC) and 5-carboxycytosine (5caC), as well as deamination to the corresponding uracil analogues. The uracil analogues include 5-hydroxymethyluracil (5hmU), 5-formyluracil (5foU), and 5-carboxyuracil (5caU), as shown in Figure 1.

Figure 1.

Potential reactions in the 5-methylcytosine oxidation/deamination pathways. Free base products released by known glycosylases are shown in bold. Glycosylases known to remove these bases are shown adjacent to the hollow arrows.

DNA bases can also be modified by chemical reactions including both oxidation and deamination.¹⁵ One of the more frequent endogenous DNA damage events is the hydrolytic deamination of cytosine, which results in the formation of uracil.¹⁶ Recent studies have also demonstrated that cytosine residues in DNA can be enzymatically deaminated by activation-induced deaminases (AID) and apolipoprotein B mRNA editing enzyme catalytic polypeptide 1 (APOBEC-1).^17–19 The enzymatic conversion of cytosine to uracil facilitates somatic hypermutation in immunoglobulin variable genes in mammals, which results in increased antibody diversity. While uracil is a normal component of RNA, it is recognized as a damaged base when it is in DNA and is cleaved by members of the uracil-DNA glycosylase family.^20–24 Products of enzymatic modification of 5mC, with the exception of 5hmC,²⁵ such as 5foC, 5caC, and the deamination products 5hmU, 5foU, and 5caU, can also be removed from DNA by glycosylases of the base excision repair (BER) pathway (Figure 1).

While DNA has been considered historically to be a repository of genetic information that can be passed unmodified to progeny cells, recent evidence suggests that DNA can undergo multiple enzymatic modifications that can provide epigenetic regulation of gene transcription, diversity in immunological cells, and, potentially, other functions. The emerging pattern is that DNA can be metabolically active with the generation of modified bases that are subject to removal by glycosylases of the BER pathway. Multiple methods have been reported for measuring DNA modifications as the corresponding 2′-deoxynucleosides;^26,27 however, substantially fewer studies have been devoted to examining modified free bases. As the free bases can be considered to be the end products of multiple DNA metabolic pathways, examination of free bases in human cell extracts and tissues could help to reveal further details of these pathways in normal tissues as well as defective pathways associated with human disease. In this article, we describe the synthesis and characterization of the stable-isotope analogues of the potential enzymatic products of the DNA pyrimidines. Furthermore, we present initial studies in which these analogues have been used to examine DNA metabolites in the rat brain.

MATERIALS AND METHODS

Materials.

Triethyl orthoformate, ethyl cyanoacetate, diethyl malonate, sodium metal, lithium aluminum hydride solution, triethylamine, potassium persulfate, silver nitrate, and ¹⁵N₂-urea were obtained from Sigma-Aldrich (St. Louis, MO). Potassium perruthenate, paraformaldehyde, N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide + 1% tert-butyldimetheylchlorosilane (MTBSTFA + 1% TBDMCS), reagent grade solvents, and anhydrous solvents were purchased from Fisher Scientific (Pittsburgh, PA). ¹⁵N₂-Orotic acid, ²H₄-thymine, DMSO-d₆, and methanol-d₄ were purchased from Cambridge Isotope Laboratory (Cambridge, MA). ²H₄-5-methylcytosine and ²H₂-cytosine were obtained from CDN Isotopes (Quebec, Canada). ¹⁵N₂-uracil was purchased from Euriso-top (Saint-Aubin, France).

Analytical Methods.

NMR spectra were acquired with a Bruker UltraShield 300 MHz spectrometer (Billerica, MA) using DMSO-d₆ or methanol-d₄ as the solvent.

UV spectra were obtained with a Varian Cary 300 Bio UV/vis spectrophotometer (Santa Clara, CA).

HPLC analysis was performed with a ThermoFinnigan Surveyor HPLC system with a photodiode array detector (Waltham, MA) using a reverse-phase HPLC column (SUPELCOSIL-LC-18-S, 15 cm × 4.6 mm, 5 μm). The column was equilibrated with 10 mM ammonium acetate buffer (pH 6.8; flow rate, 1 mL/min). Mobile phases consisted of (A) 10 mM ammonium acetate at pH 6.8 and (B) methanol. The method was run isocratic for the first 10 min with mobile phase (A), and then a 20 min linear gradient was applied running from 0 to 50% B followed by column re-equilibration with 0% B at a flow rate of 1 mL/min. Analysis was conducted using a photodiode array detector (wavelength, 200–800 nm).

GC-MS data for analytical characterization of the free bases was obtained with an Agilent Technologies 5975C inert XL EI Triple-Axis detector coupled to an Agilent Technologies 7890A GC. For each free base standard, 50 μL of a 2.5 mM solution was placed into a 12 × 32 mm vial containing a 250 μL fused silica insert (Thermo Scientific; Rockwood, TN), and the solvent was evaporated under reduced pressure. The dried samples were reconstituted with 20 μL of acetonitrile and 20 μL of MTBSTFA + 1% TBDMCS, sealed, and heated at 140 °C for 40 min to make their tert-butyl-dimethylsilyl (TBDMS) derivatives. Each silylated sample was injected (1 μL) onto a GC-MS equipped with a Hewlett-Packard silica capillary column (30 m × 0.25 mm) coated with cross-linked 5% phenyl/95% methylpolysiloxane (film thickness, 0.25 μm). The following temperatures were used: 250 °C for the injector and 260 °C for the detector interface. The initial oven temperature was 100 °C for 2 min and was ramped to 180 °C at 10 °C/min and then to 260 °C at 30 °C/min and held for 7 min.

High-resolution GC-MS spectra, of the TBDMS derivatives, were obtained with a Waters Micromass GCT (Milford, MA) coupled to an Agilent 6890 GC. Ultra-high-resolution microelectrospray mass spectra of the underivitized standards were obtained with a Bruker Solarix 12 T FT-ICR MS. Samples were constantly infused and ionized by positive ion microelectrospray.²⁸

Synthetic Procedures.

Cytosine Analogues.

¹⁵N₂-Ethyl Ureidomethylenecyanoacetate (1).

To a 25 mL dry round-bottomed flask were added the following: ¹⁵N₂-urea (1 g, 16.66 mmol, 98% enriched), triethyl orthoformate (4.5 mL, 26.66 mmol), and ethyl cyanoacetate (1.7 mL, 16.66 mmol). The neat reaction mixture was refluxed for 10 h. The solid was filtered and washed with acetone to obtain ¹⁵N₂-ethyl ureidomethylenecyanoacetate (1.7 g, 55% yield). ¹H NMR (DMSO-d₆, 300 MHz) δ (ppm): 10.54–10.58 (d, 1H), 8.07–8.11 (d, 1H), 7.61 (br s, 1H), 7.38 (br s, 1H), 4.20–4.27 (q, 2H), 1.24–1.29 (t, 3H).

¹⁵N₂-5-Carboxyethylcytosine (2).

To the solution of sodium ethoxide, prepared by mixing 15 mL absolute ethanol and sodium (230 mg, 10 mmol), was added ¹⁵N₂-ethyl ureidomethylenecyanoacetate (1.7 g, 9 mmol), and the reaction mixture was refluxed for 4 h. The alcohol was removed by filtration, and the collected solid was dissolved in 100 mL of water. This solution was filtered and acidified with glacial acetic acid to pH 5. The precipitate was collected on a Buchner funnel and washed with alcohol and then with ether to obtain ¹⁵N₂-5-carboxyethylcytosine (0.8 g, 50% yield). ¹H NMR (DMSO-d₆) δ (ppm): 11.38 (s, 1H), 8.20 (s, 1H), 7.83 (br s, 1H), 7.63 (br s, 1H), 4.18–4.25 (q, 2H), 1.24–1.29 (t, 3H).

¹⁵N₂-5-Hydroxymethylcytosine (3) and 15N₂-5-Carboxycytosine (4).

To a 10 mL solution of LiAlH₄ (1 M in THF) was added ¹⁵N₂-5-carboxyethylcytosine (370 mg, 2 mmol) at room temperature. The reaction mixture was warmed to 40 °C and stirred for 2 h. After cooling the mixture to room temperature, water (5 mL) was added dropwise to quench excess LiAlH₄. The solid was removed by filtration and washed with water (50 mL). The water extracts were combined and concentrated to a volume of 10 mL. GC-MS analysis of this crude mixture identified the presence of ¹⁵N₂-5-hydroxymethylcytosine and ¹⁵N₂-5-carboxycytosine. The crude product was purified by HPLC using a Hypersil prep HPLC column, which was eluted with up to 50% methanol in 10 mM ammonium acetate (pH 5.5) to obtain ¹⁵N₂-5-hydroxymethylcytosine (84 mg, measured by UV with ε₂₆₉ = 5700 M⁻¹ cm⁻¹) and ¹⁵N₂-5-carboxycytosine (21 mg, measured by UV with ε₂₇₅ = 5800 M⁻¹ cm⁻¹). ¹H NMR for ¹⁵N₂-5-hydroxymethylcytosine (DMSO-d₆) δ (ppm): 10.40 (s, 1H), 7.26 (s, 1H), 6.47 (br s, 2H), 4.92–4.95 (br t, 1H), 4.14–4.16 (d, 2H). HRMS (ESI+): m/z calcd for C₅H₇N¹⁵N₂O₂ [M + H]⁺, 144.05508; found, 144.05518. ¹H NMR for ¹⁵N₂-5-carboxycytosine (CD₃OD) δ (ppm): 12.34 (s, 1H), 8.38 (s, 1H). HRMS (ESI+): m/z calcd for C₅H₅N¹⁵N₂O₃ [M + H]⁺, 158.03438; found, 158.03448.

¹⁵N₂-5-Formylcytosine (5).

A solution of ¹⁵N₂-5-hydroxymethylcytosine (50 mg, 0.35 mmol) was prepared in 5 mL of 50 mM sodium hydroxide and cooled to 0 °C. To this solution was added potassium perruthenate (71 mg, 0.35 mmol), and the reaction was allowed to stir at 0 °C for 2 h and then at room temperature for 10 h. The fine black particles were filtered through a 3 mL syringe packed with 1 mL of silica and 0.5 mL of sand and then through a 0.45 μm syringe filter [From bottom to top: filter, 1 mL of silica, and 0.5 mL of sand]. The collected filtrate was concentrated and HPLC purified using a Hypersil prep HPLC column, which was eluted with up to 50% methanol in 10 mM ammonium acetate (pH 5.5) to obtain ¹⁵N₂-5-formylcytosine (15.3 mg, measured by UV with ε₂₇₆ = 6600 M⁻¹ cm⁻¹). ¹H NMR (DMSO-d₆) δ (ppm): 9.41 (s, 1H), 8.38 (s, 1H), 7.84–7.86 (br m, 3H). HRMS (ESI+): m/z calcd for C₅H₅N¹⁵N₂O₂ [M + H]⁺, 142.03948; found, 142.03952.

Uracil Analogues.

¹⁵N₂-Ethyl Ureidomethylenemalonate (6).

To 25 mL dry round-bottomed flask were added the following: ¹⁵N₂-urea (1 g, 16 mmol), triethyl orthoformate (4.5 mL, 26.66 mmol), and diethyl malonate (3.5 mL, 21 mmol). The neat reaction mixture was refluxed for 4 h. The solid was filtered and washed with acetone to obtain ¹⁵N₂-ethyl ureidomethylenemalonate (1 g, 27% yield). ¹H NMR (DMSO-d₆) δ (ppm): 11.41 (s, 1H), 8.14 (s, 1H), 7.55 (br s, 1H), 7.30 (br s, 1H), 4.17–4.24 (q, 4H), 1.23–1.27 (t, 6H).

¹⁵N₂-5-Carboxyethyluracil (7).

To a solution of sodium ethoxide, prepared by mixing 15 mL absolute ethanol and sodium (100 mg, 4.3 mmol), was added ¹⁵N₂-ethyl ureidomethylenemalonate (0.5 g, 2.15 mmol), and the reaction mixture was refluxed for 3 h. The alcohol was removed by filtration, and the collected solid was dissolved in 10 mL of water. This solution was acidified with glacial acetic acid to pH 5 to precipitate the desired product. The precipitate was collected on a Buchner funnel and washed with alcohol and then with ether to obtain ¹⁵N₂-5-carboxyethyluracil (0.353 g, 93% yield). ¹H NMR (DMSO-d₆) δ (ppm): 11.80 (s, 1H), 11.29 (s, 1H), 8.09 (s, 1H), 4.11–4.18 (q, 2H), 1.19–1.24 (t, 3H).

¹⁵N₂-5-Carboxyuracil (8).

A solution of ¹⁵N₂-5-carboxyethyluracil (100 mg, 0.53 mmol) in 10 mL of 1 M NaOH was heated at 60 °C for 12 h. At the end of reaction, the mixture was allowed to cool at room temperature, and pH of the solution was adjusted to 4.5 using dilute HCl. The solution was concentrated to the volume of 5 mL and allowed to stand at 4 °C overnight to precipitate 5-carboxyuracil. The precipitates were collected via filtration and washed with acetone (10 mL) to obtain ¹⁵N₂-5-carboxyuracil (75 mg, 90% yield). ¹H NMR (DMSO-d₆) δ (ppm): 12.73 (s, 1H), 11.99 (s, 2H), 8.26 (s, 1H). HRMS (ESI+): m/z calcd for C₅H₄¹⁵N₂O₄ [M + H]⁺, 159.01838; found, 159.01849.

¹⁵N₂-5-Hydroxymethyluracil (9).

Triethylamine (0.5 mL, 3.3 mmol) was added to a suspension of ¹⁵N₂-uracil (250 mg, 2.2 mmol) and paraformaldehyde (200 mg, 6.6 mmol) in 10 mL of water. The solution turned clear when heated at 65 °C and was allowed to stir for 12 h at 65 °C. The water was reduced to a volume of 5 mL, and 5 mL of 95% ethanol was added to the mixture. The mixture was allowed to stand at 4 °C overnight to precipitate ¹⁵N₂-5-hydroxymethyluracil as a white solid (250 mg, 80% yield). ¹H NMR (DMSO-d₆) δ (ppm): 10.79–11.18 (br m, 2H), 7.22–7.24 (d, 1H), 4.83 (br s, 1H), 4.10–4.11 (m, 1H). HRMS (ESI+): m/z calcd for C₅H₆¹⁵N₂O₃ [M + H]⁺, 145.03908; found, 145.03920.

¹⁵N₂-5-Formyluracil (10).

A solution of ¹⁵N₂-5-hydroxymethyluracil (100 mg, 0.69 mmol) in 5 mL of water was heated at approximately 90 °C. The solution was then cooled to 45 °C, and K₂S₂O₈ (187 mg, 0.69 mmol) and AgNO₃ (3.5 mg, 0.02 mmol) were added. The product began to slowly precipitate. The reaction was stirred for 20 min at 40 °C and was then cooled to room temperature over 15 min while stirring was continued. The suspension was filtered, and the collected solid was rinsed with 2 mL of cold water to yield ¹⁵N₂-5-formyluracil (70 mg, 71% yield). ¹H NMR (DMSO-d₆) δ (ppm): 11.89 (s, 1H), 11.50 (s, 1H), 9.74 (s, 1H), 8.14 (s, 1H). HRMS (ESI+): m/z calcd for C₅H₄¹⁵N₂O₃ [M + H]⁺, 143.02348; found, 143.02356.

Animals.

Animal studies were conducted in a facility approved by the American Association for the Accreditation of Laboratory Animal Care (AAALAC), and all experiments were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee (protocol no. 1312056) of the University of Texas Medical Branch (UTMB). All animals were housed at the UTMB animal care facility and maintained according to U.S. Department of Agriculture standards (12 h light/dark cycle, food and water ad libitum). Male Sprague–Dawley (Charles Rivers, Wilmington, MA) rats (400–500 g) were anesthetized and humanely euthanized, and then their brains were collected and immediately frozen.

Metabolite Extraction from Rat Brain Tissue.

Whole brain tissue (1.75 g) from a male Sprague–Dawley rat was combined with 4 mL of deionized water and homogenized using a tissue homogenizer (POLYTRON PT 3100) for 45 s (15 s × 3) on an ice bath.²⁹ The homogenized tissue was pelleted by centrifugation for 30 min at 14 000 rpm at 4 °C. The clear supernatant (3.5 mL) was aspirated, and 1 mL of 80:20 methanol/water at dry ice temperature (−75 °C) was added to the pellets and mixed. After 15 min at −75 °C, the sample was centrifuged at 14 000 rpm for 10 min at 4 °C, and the soluble extract (supernatant) was removed and mixed with the previously aspirated fraction. The pellets were then resuspended in 1 mL of 80:20 methanol/water and placed on dry ice for 15 min. The solution was centrifuged at 14 000 rpm for 10 min at 4 °C to yield a second clear supernatant, which was aspirated and combined with the previous extracts. The pellets were resuspended in 1 mL of 80:20 methanol/water, and the resulting suspension was sonicated in an ice bath for 45 s (15 s × 3) using a Branson Sonifier 450. The suspension was then centrifuged at 14 000 rpm for 10 min at 4 °C to yield a third clear supernatant, which was aspirated and combined with previous fractions to give a total extract volume of 6 mL. The extract was then filtered through a 3000 Da cutoff centrifugal spin filter (Millipore UFC900324) at 14 000g for 30 min. The clear filtrate (5.5 mL) was collected and mixed with isotope-labeled internal standards for HPLC purification and GC-MS analysis.

HPLC Isolation and GC-MS/MS Quantification of Metabolites.

A portion of the rat brain tissue extract (2 mL) was mixed with an internal standard mixture (50 μL, 1.13 × 10⁻⁵ M, 11 labeled standards) and purified by HPLC over a SUPELCOSIL-LC-18-S (15 cm × 4.6 mm, 5 μm) column, which was isocratically eluted with 10 mM ammonium acetate, pH 6.8, and 50% methanol for 10 min at a flow rate of 1 mL/min. In 2 mL microfuge tubes, HPLC fractions (0.6–1.5 mL each) were collected based on the HPLC retention times of the free base standards. To ensure separation of 5caU and 6caU, fractions were collected between 2.8–3.4 and 3.5–4.1 min, respectively. Each collected fraction was transferred to an autosampler vial and dried, and pyrimidines were converted to their TBDMS derivatives as described earlier. GC-MS/MS was used for the quantification of pyrimidine free base metabolites in the extracted rat brain tissue. For quantification, 1 μL injections were made on an Agilent Technologies 7000C GC-MS Triple-Quad detector coupled to an Agilent Technologies 7890B GC. The following temperatures were used: 250 °C for the injector and 260 °C for the detector interface. The initial oven temperature was 100 °C for 2 min and was ramped to 260 °C at 30 °C/min and held for 10 min. Data were collected in multiple reaction monitoring mode (MRM) using transitions determined using commercially available and synthetic standards.

RESULTS AND DISCUSSION

Synthesis and Characterization of Stable-Isotope-Enriched Pyrimidine Analogues.

Cytosine Analogues.

Previously, Whitehead³⁰ demonstrated that urea, triethylorthoformate, and ethyl cyanoacetate could be condensed to form ethyl ureidomethylene cyanoacetate (Figure 2) followed by ring closure to form 5-carboxyethylcytosine. We utilized the same strategy, but substituted ¹⁵N₂-enriched urea, to yield ¹⁵N₂-enriched 5-carboxyethylcytosine. The product obtained was pure, as indicated by ¹H NMR, and no other pyrimidines were observed by GC-MS. Using this procedure, approximately 1 g of ¹⁵N₂-enriched urea can generate 0.8 g of enriched 5-carboxyethylcytosine.

Figure 2.

Synthetic pathway for cytosine analogs. The bold N denotes heavy isotope labeling (¹⁵N).

Acid hydrolysis of the ethyl ester of 5-carboxyethylcytosine yielding 5-carboxycytosine did not go to completion. However, alkaline hydrolysis in 1 N NaOH provided ¹⁵N₂-5-carboxycytosine as the only pyrimidine product.

The reduction of the ethyl ester of 5-carboxyethylcytosine using LiAlH₄ to form 5-hydroxymethylcytosine has been previously reported.³¹ In our hands, both ¹⁵N₂-5-hydroxymethylcytosine and the hydrolysis product ¹⁵N₂-5-carboxycytosine were generated in approximately equal amounts. We therefore used this method to generate the ¹⁵N₂-analogues of the 5-hydroxymethyl and 5-carboxy analogues of cytosine, which were subsequently purified by HPLC. Product identity and purity were verified by GC-MS analysis.

The last product in the cytosine series, 5-formylcytosine, was generated by oxidation of enriched 5-hydroxymethylcytosine with potassium perruthenate.³² The product was purified by HPLC and characterized by GC-MS.

Uracil Analogues.

In the uracil series, the analogous condensation of urea, triethyl orothoformate, and diethylmalonate has been shown to yield ethylureidomethylene malonate, with ring closure generating 5-carboxyethyluracil,³³ as shown in Figure 3. Using this procedure, approximately 1 g of ¹⁵N₂-enriched urea can be converted to 350 mg of ¹⁵N₂-enriched 5-carboxyethyluracil.

Figure 3.

Synthetic pathway for uracil analogs. The bold N denotes heavy isotope labeling (¹⁵N).

The acid hydrolysis of 5-carboxyethyluracil yields 5-carboxyluracil; however, substantial amounts of uracil were also observed. Alternatively, alkaline hydrolysis of enriched 5-carboxyethyluracil generates enriched 5-carboxyluracil as the only pyrimidine product. The reduction of 5-carboxyethyluracil to 5-hydroxymethyluracil with LiAlH₄, unlike the reduction of 5-carboxyethylcytosine described above, was unsuccessful. Similarly, reduction of 5-carboxyethyluracil with sodium borohydride did not proceed with significant yield.³⁴ An alternative approach was needed, therefore, to generate ¹⁵N₂-enriched 5-hydroxymethyluracil and 5-formyluracil.

Previously, it has been established that the condensation of ¹⁵N₂-enriched urea with propiolic acid in polyphosphoric acid yields ¹⁵N₂-enriched uracil.^35,36 The conversion of uracil to 5-hydroxymethyluracil has been reported to proceed in high yield with aqueous formaldehyde and triethylamine.³⁷ We confirmed the analogous formation of ¹⁵N₂-5-hydroxymethyluracil conversion using enriched uracil.

The oxidation of 5-hydroxymethyluracil with silver nitrate and potassium persulfate in water yields 5-formyluracil,³⁷ completing the synthesis of the uracil series. The ¹⁵N₂-enriched uracil analogues were purified by HPLC and characterized by GC-MS, NMR, and UV spectroscopies as described below.

Characterization of the Pyrimidine Analogues.

Synthetic analogues could be separated from one another in sufficient quantity by HPLC, as shown in Figure 4. HPLC-purified analogues were characterized by GC-MS, as shown in Figure 5. The corresponding UV spectral characteristics and pK_a values are shown in Table 1. GC-MS mass and retention times are provided in Table 2.

Figure 4.

HPLC chromatogram of oxidized pyrimidine free bases obtained with UV absorbance. A mixture of oxidized pyrimidine free bases was separated by HPLC using a SUPELCOSIL-LC-18-S column eluted with 50% methanol in 10 mM ammonium acetate at pH 6.8 (flow rate, 1 mL/min).

Figure 5.

GC-MS chromatogram of free bases mixture. A mixture of oxidized pyrimidine free bases was derivatized to the TBDMS derivatives and injected onto an Agilent Technologies 7890A GC system that was coupled to an Agilent 5975C MSD with Triple-Axis detector. Data was acquired in the selected ion mode and is presented as the total ion chromatogram.

Table 1.

UV λ_max, ε, pK_a Values, and H6 Proton Chemical Shift for Pyrimidine Free Bases Examined in This Study^a

compound	λmax at pH 7 (nm)	ε at pH 7 (M−1 cm−1)	pKa	H6 proton chemical shift (ppm)
thymine	265	7900b	9.9, >13.0	7.25
5-hydroxymethyluracil	261	8000b	9.4, ~14.0	7.23
5-formyluracil	293	12 000	6.8c, >9.4	8.14
5-carboxyuracil	268	8300	4.1d, 9.0c	8.26
uracil	2b0	8200b	9.5, >13.0	7.38
5-methylcytosine	273	6200b	4.6, 12.4	7.69
5-hydroxymethylcytosine	269	5700b	4.3, ~13.0	7.26
5-formylcytosine	276	6600	2.6e, 9.1	8.38
5-carboxycytosine	275	5800	2.5f, 4.7f, >9.1	8.38
cytosine	267	6100b	4.5, 12.2	7.30
6-carboxyuracil	279	7700b	2.4, 9.5	6.00

^aValues reported in this table were measured in this study for the free base analogues except where otherwise noted.

^bValues for λ_max, ε, and pK_a are from ref 40. Values for ε correspond to λ_max.

^cThe pK_a values for 5-formyluracil and 5-carboxyuracil are from ref 41.

^dThe pK_a value reported correspond to the 2′-deoxynucleoside analogue from ref 42.

^eThe reported pK_a value corresponds to the 2′-deoxynucleoside analogue from ref 43.

^fThe pK_a value for 5-carboxycytosine is from ref 44.

Table 2.

HR-MS (GC-ToF-MS) Data, GC, and HPLC Retention Times of Stable Isotope Enriched Pyrimidine Free Bases Examined in This Study

compound	chemical formula TBDMS derivatives	theoretical M-[tbutyloxy]	experimentala M-[tbutyloxy]	GCa tR (min)	HPLCb tR (min)
2H4-thymine	C17H302H4N2O2Si2	301.17056	301.1697	12.13	15.63
15N2-5-hydroxymethyluracil	C23H4815N2O3Si3	429.22091	429.2229	14.47	6.22
15N2-5-formyluracil	C17H3215N2O3Si2	313.11879	313.1196	13.04	6.93
15N2-5-carboxyuracil	C23H4615N2O4Si3	443.20018	443.2021	15.08	3.01
15N2-uracil	C16H3215N2O2Si2	285.12387	285.1239	11.71	6.47
2H4-5-methylcytosine	C17H312H4N3OSi2	300.18654	300.1856	12.81	9.69
15N2-5-hydroxymethylcytosine	C23H49N15N2O2Si3	428.23690	428.2385	15.14	5.15
15N2-5-formylcytosine	C17H33N15N2O2Si2	312.13477	312.1355	13.35	9.36
15N2-5-carboxycytosine	C23H47N15N2O3Si3	442.21616	442.2178	15.58	3.24
2H2-cytosine	C16H312H2N3OSi2	284.15834	284.1571	12.71	4.84
15N2-6-carboxyuracil	C23H4615N2O4Si3	443.20018	443.2007	15.08	3.79

^aMass and retention times for TBDMS derivatives were measured on a high-resolution single quadrupole GC-ToF-MS.

^bHPLC retention time for underivatized free bases (t_R).

All of the DNA pyrimidine metabolites can be separated by both HPLC and GC, facilitating identification and quantification. In principle, a mixture of analytes and standards could be quantified without separation by examining appropriate mass transitions. Within the series of metabolites examined here, the nominal mass of the unenriched 5-hydroxymethyl analogues is the same as that of the ¹⁵N₂-enriched 5-formyl analogue (isobars) in both the uracil and cytosine series. However, at higher mass resolution, the analogues can be distinguished without separation, as shown in Figure 6 and Table 3, based on their exact masses.

Figure 6.

Ultra-high-resolution 12 T FT-ICR MS positive ion microelectrospray ionization spectrum of a mixture of ¹⁵N₂-5foC and 5hmC showing a zoomed spectrum over a 60 mDa window. FT-ICR MS easily baseline resolves the biologically active 5hmC that differs in mass by only ~22 mDa from the isotope-labeled standard ¹⁵N₂-5foC. Mass accuracy was ~350 ppb for ¹⁵N₂-5foC and ~210 ppb for 5hmC. Spectra were acquired on a 12 T Bruker Solarix Fourier transform ion cyclotron resonance mass spectrometer (FT-ICR MS).

Table 3.

Ultra-High-Resolution FT-ICR MS Mass Measurements for Pyrimidine Free Bases Examined in This Study

compound	formula	[M + H]+ (theoretical)	[M + H]+ (experimental)
thymine	C5H6N2O2	127.05018	127.05018
5-hydroxymethyluracil	C5H6N2O3	143.04508	143.04516
5-formyluracil	C5H4N2O3	141.02948	141.02950
5-carboxyuracil	C5H4N2O4	157.02438	157.02442
uracil	C4H4N2O2	113.03458	113.03450
5-methylcytosine	C5H7N3O	126.06618	126.06613
5-hydroxymethylcytosine	C5H7N3O2	142.06108	142.06108
5-formylcytosine	C5H5N3O2	140.04548	140.04543
5-carboxycytosine	C5H5N3O3	156.04038	156.04037
cytosine	C4H5N3O	112.05048	112.05044
15N2-5-hydroxymethyluracil	C5H615N2O3	145.03908	145.03920
15N2-5-formyluracil	C5H415N2O3	143.02348	143.02356
15N2-5-carboxyuracil	C5H415N2O4	159.01838	159.01849
15N2-5-hydroxymethylcytosine	C5H7N15N2O2	144.05508	144.05518
15N2-5-formylcytosine	C5H5N15N2O2	142.03948	142.03952
15N2-5-carboxycytosine	C5H5N15N2O3	158.03438	158.03448

Development of a Method To Measure Pyrimidine Frees Bases from Tissue Extracts.

Our initial attempts to measure pyrimidine standards by GC-MS revealed that 10 of the 11 pyrimidines could be resolved by GC alone (Figure 5). Only the isomeric 5- and 6-carboxyuracil analogues co-eluted. When we first attempted to measure the pyrimidines in tissue extracts by GC-MS, we observed substantial background noise. Peaks expected for stable-isotope analogues that were added into the extracts also displayed coeluting contaminants. Additionally, overall metabolite recovery from the tissue extracts was much lower than expected.

Our initial GC-MS results suggested that additional sample cleanup was necessary, so we turned to HPLC separation as an initial cleanup step. Our work with the synthesis of the stable-isotope analogues revealed that we could obtain substantial resolution of many of the pyrimidines by HPLC (Figure 4). Most importantly, we were able to separate the 5- and 6-carboxyuracil isomers that cannot be resolved by GC-MS.

Although all of the pyrimidines expected from DNA metabolism can be distinguished by exact mass, the isomer of 5-carboxyuracil, 6-carboxyuracil (orotic acid), is a normal metabolite in the de novo synthesis of pyrimidines^38,39 and therefore its presence in biological extracts is expected. The 5- and 6-carboxyuracil analogues are inseparable by GC; however, they are easily separated by HPLC at pH 6.8 (Figure 4). The pK_a values for this pyrimidine series^40–44 are listed in Table 1. We note that the pK_a values for the carboxyl groups of 5- and 6-carboxyuracil differ by 1 pH unit.

The next step in our method development was to combine the HPLC prepurification with GC-MS analysis, as shown in Figure 7. Freshly obtained animal tissue is homogenized, and cell debris is removed by centrifugation. The aqueous extract is then combined with water and methanol, chilled, and cleared by centrifugation, as described in Materials and Methods. The clear filtrate is then mixed with stable-isotope-enriched standards, and a portion of this mixture is injected onto the HPLC. Fractions are obtained containing one or more of each of the pyrimidine analytes, and these fractions are dried under reduced pressure, derivatized, and analyzed by GC-MS.

Figure 7.

Workflow diagram for the isolation and analysis of pyrimidine free bases from a rat brain.

Our initial workup used GC separation and a single-quad mass spectrometer. Although selected ion monitoring on the single quad can be used for analyte identification and quantification, we observed substantial background peaks coeluting with the pyrimidines of interest. We therefore used a triple-quad mass spectrometer, which is able to substantially reduce background noise by trapping ions of interest and measuring selected fragments. Using this method, the results shown in Figure 8 are obtained. The level of each pyrimidine is then determined by measuring the size of the analyte peak relative to the size of the isotope standard peak. The limit of detection for this method is approximately 1 × 10⁻¹⁵ moles of pyrimidine injected. The volume of the derivatized sample is 40 μL, of which 1 μL is injected in splitless mode. Therefore, approximately 10⁻¹³ moles per sample can be detected with this method.

Figure 8.

GC-MS/MS analysis of HPLC-purified pyrimidines obtained from a rat brain. Red, isotope-enriched internal standards; Black, unenriched metabolites from rat brain extract.

Measurement of Metabolites in Normal Rat Brain Extract.

The DNA of all living cells is persistently damaged by oxidation and hydrolysis.¹⁵ In addition to these chemically mediated pathways, DNA bases undergo enzyme-mediated modifications, including methylation, demethylation, and deamination of cytosine residues.^10–14 Many of the chemically damaged or enzymatically modified bases are removed by proteins of the base excision repair (BER) pathway, and the damaged DNA is restored by repair synthesis.^20–25 Collectively, these modifications constitute postreplicative DNA metabolism, and the end products of these pathways are the free base metabolites that are released into the cytoplasm and subsequently reutilized, excreted, or degraded.

Although substantial work from many laboratories has described the measurement of damaged bases in DNA, relatively little work has focused on measuring the free bases that are spontaneously or enzymatically released. In this work, we have prepared a series of isotopically enriched standards for a series of pyrimidine analogues known to be part of both endogenous DNA damage and enzyme-mediated DNA modification pathways. The isotope standards have been characterized by NMR and high-resolution mass spectrometry. We demonstrate here the feasibility of measuring the free bases in tissue extracts using stable-isotope-enriched standards. Following tissue homogenization and methanol–water extraction, isotope standards are added and small molecules are isolated using a spin filter. Free bases are separated by HPLC, and selected fractions are dried, derivatized, and examined by GC-MS/MS.

Nine of the 11 pyrimidines studied here are measurable in the rat brain extract by this method (Figure 8 and Table 4). Neither the potential metabolite 5foU nor its isotope standard are observed by GC-MS/MS, suggesting that 5foU undergoes chemical modification in the procedure described here. Potentially, the activated aldehyde of 5foU could condense with amines or sulfhydryl groups present in the biological extract. In contrast, the isotope standard of 5caC is observed. However, the level of unenriched 5caC in the tissue extract is below the limit of detection (~1 × 10⁻¹³ mol/g tissue).

Table 4.

Quantities (mol/g of Tissue) of Normal and Oxidized Pyrimidines in a Rat Brain Extract Measured Using Isotope-Labeled Standards^a

compound	moles of free bases/g of tissue
uracil	1.20 ± 0.11 × 10−08
thymine	9.01 ± 0.46 × 10−10
cytosine	1.18 ± 0.15 × 10−09
5-methylcytosine	2.90 ± 0.13 × 10−11
5-formylcytosine	1.21 ± 0.15 × 10−09
5-hydroxymethyluracil	6.58 ± 0.37 × 10−11
5-carboxyuracil	8.76 ± 0.05 × 10−10
6-carboxyuracil	7.11 ± 0.08 × 10−10
5-hydroxymethylcytosine	7.77 ± 0.10 × 10−11
5-carboxycytosine	<1.14 × 10−13

^aThe data represent the average and standard deviation (SD) values from three different injections.

The amount of free thymine measured by this method is 9 × 10⁻¹⁰ mol/g, which corresponds to approximately 0.1% of the thymine found in DNA extracted from a tissue of similar size. In rat DNA, the ratio of thymine to cytosine is approximately 1.34,⁴⁵ and the rate of spontaneous depyrimidination of thymine to cytosine is 1.28.⁴⁶ Therefore, we would have expected more thymine than cytosine in the extracts resulting from depyrimidination. Surprisingly, the amount of thymine found was only about 80% that of cytosine.

Other potential mechanisms could also modulate thymine levels. Perhaps the deamination of 5mC to thymine, followed by glycosylase removal, could also contribute to the thymine levels;⁴⁷ however, the ratio of 5mC to thymine in rat DNA is small, only about 1:30. Therefore, this pathway would not be expected to contribute substantially to the free thymine levels. DNA from dead and dying cells can be degraded to generate 5′-monophosphates that can be either reutilized for DNA synthesis or dephosphorylated to 2′-deoxynucleosides that can by degraded to free pyrimidines by phosphorylases. Uracil and thymine free bases can be generated and utilized by thymidine phosphorylase.⁴⁸ In addition, uracil and thymine free bases analogues can be metabolized by dihydropyrimidine dehydrogenase and excreted.⁴⁹ For example, the chemo-therapeutic agent 5-fluorouracil can be incorporated into the nucleotide pool by thymidine phosphorylase,⁵⁰ which suggests a pathway for reducing thymine levels in the brain extracts. The observed levels of thymine in the extracts, therefore, most likely reflect thymine generated by spontaneous hydrolysis minus that which is salvaged or degraded.

Both cytosine and 5mC are observed in the tissue extracts. This result was unexpected, as most glycosylases do not target cytosine derivatives, although 5mC glycosylases have been reported in some species.^51–53 In rat brain, approximately 4% of the cytosine residues are methylated.⁵⁴ The ratio of 5mC to cytosine observed in our rat brain extracts was approximately 2.5%. Both cytosine and 5mC can be lost from the DNA by spontaneous hydrolysis of the glycosidic linkage, and the rates of depyrimidination for both are similar.⁴⁶ Cytosine and 5-substituted cytosine bases are not enzymatically deaminated in animal cells, although the corresponding nucleoside and nucleotide analogues are rapidly deaminated in animal cells.⁵⁵ For example, 5-fluorocytosine serves as a prodrug of 5-fluorouracil and is activated in mammalian tissues only by deamination in the presence of microbial cytosine deaminase.⁵⁶ Unlike the uracil analogues discussed below, cytosine is not reutilized by salvage pathways, and it is not degraded by dihydropyrimidine dehydrogenase. Therefore, the cytosine and 5mC free bases observed in these extracts most likely reflect endogenous depyrimidination.

Among the oxidized bases, both 5hmU and 5hmC can arise by chemical oxidation of the corresponding thymine and 5mC methyl groups.⁵⁷ An enzymatic pathway has also been identified for the conversion of 5mC to 5hmC in DNA.^11–13 In mouse brain, the ratio of 5mC to 5hmC is approximately 10.⁵⁸ The amount of thymine in rat DNA exceeds the amount of 5mC by a factor of 30. Therefore, the amount of 5hmU in the extracts would be expected to exceed that of 5hmC if only endogenous chemical pathways were considered. Furthermore, 5hmU is removed from DNA by the BER pathway, but 5hmC is not a substrate for TDG glycosylase, and no activities have been observed that remove 5hmC from DNA.²⁵ The observation of similar levels of 5hmU and 5hmC in the extracts is, therefore, surprising and might suggest an as yet unidentified pathway for the removal of 5hmC from DNA.

Both 5hmU and 5hmC can be further oxidized to the corresponding 5-formyl analogues 5foU and 5foC.⁴⁵ The enzymatic oxidation of 5hmC can also generate 5foC in DNA.⁵⁹ Unlike 5hmC, 5foC paired with guanine is a preferred substrate for the TDG glycosylase,²² and 5foU is repaired by multiple members of the uracil-DNA glycosylase family.⁶⁰ Neither 5foC nor 5foU would be expected to remain in DNA for long due to the BER pathway. In this study, neither 5foU nor the isotopomer standard was observed. It is known that 5foU can react with both amines and thiol groups, likely explaining its disappearance in this study.^61,62 In contrast, substantial amounts of 5foC are observed. The observation of 5foC is consistent with enzymatic 5mC demethylation, where 5mC is first oxidized to 5hmC and then to 5foC and released from DNA by the BER pathway.

Further oxidation of 5foU and 5foC can generate the 5-carboxy analogues 5caU and 5caC.²² The human single-strand-selective mispaired uracil DNA-glycosylase (SMUG1) can remove 5caU from DNA,⁴² and human thymine-DNA glycosylase (hTDG) can remove 5caC.²² Although 5caU is observed in the extracts, 5caC is not observed. Unlike 5foU discussed above, the internal standard 5caC added to the extracts is measurable. Since the level of 5caC free base in the rat brain is below the level of detection in this study, it is unlikely to be a significant DNA metabolite.

The measurement of 5caU created an analytical challenge, as 6-carboxyuracil (6caU, orotic acid) is a normal metabolite in the de novo synthesis of pyrimidines.^38,39 The isomeric 5- and 6-carboxyuracil analogues are inseparable by GC-MS; however, they can be separated by HPLC and subsequently measured in independent GC-MS/MS runs. Similar levels of both 5caU and 6caU are observed in this study.

The observed level of 5caU exceeds that of 5hmU by a factor of 10. As the analogues 5hmU, 5foU, and 5caU are generated by sequential oxidation, it is unlikely that the observed 5caU arose by endogenous oxidation. It has been proposed, however, that enzymatic oxidation of 5mC followed by enzymatic deamination to the corresponding uracil analogue is an additional pathway for 5mC demethylation.²⁴ As 5foU is not observable by this method, the sequence of transformations leading from 5mC to 5caU cannot be established. However, the simultaneous observation of 5foC and 5caU suggests that multiple pathways exist for the modification of 5mC residues in DNA.

The most abundant pyrimidine observed in the tissue extract is uracil. As with thymine, uracil can be both a product of and substrate for thymidine phosphorylase,⁴⁸ and both are degraded by dihydropyrimidine dehydrogenase.⁴⁹ In addition, uracil and its analogues are both products and substrates of uridine phosphorylase.⁶³ For example, exogenous 5-fluorouracil is incorporated into both DNA and RNA. Therefore, the phosphorylases would tend to decrease uracil levels, as opposed to increasing the level of uracil over that of thymine in the extracts.

There are two mechanisms by which uracil can be incorporated into DNA. In the first pathway, dUTP is incorporated by DNA polymerases during DNA replication. In the second pathway, cytosine residues in DNA are hydrolytically deaminated to uracil.⁶⁴ In both pathways, the uracil in DNA is removed by members of the uracil-DNA glycosylase family, generating free uracil in the extract. The dUTP misincorporation pathway would predominate in replicating cells, whereas the deamination pathway would predominate in postmitotic cells. Both replicating and postmitotic cells would be present in the tissue extracts observed here. If the deamination pathway predominated, then we would expect uracil levels to be similar to cytosine levels in the extracts since the rates of DNA depyrimidination and cytosine deamination are similar. However, the level of uracil is more than an order of magnitude greater than that of cytosine. This suggests that uracil transiting through the DNA due to dUTP misincorporation might account for more of the released uracil. Because dUTP levels can be modulated by manipulation of one carbon metabolism, future studies using the methods presented here could resolve this question.^65,66

CONCLUSIONS

While DNA is the repository of genetic information that is passed onto progeny cells, it is metabolically active in all cells due to persistent endogenous damage and repair. It is also subject to enzymatic modifications in cells that undergo differentiation, and both damage and enzymatic modification can occur simultaneously. When enzymatically removed from DNA, the modified free bases are released into the cytoplasm and are either excreted, further metabolized, or reutilized. Free bases can be recovered from cell and tissue extracts and analyzed. In this article, we describe the synthesis and characterization of a complete series of ¹⁵N₂-enriched analogues of the possible base metabolites described to date. These analogues can be used to study DNA synthesis, modification, and base turnover in both normal tissues and cells with metabolic defects. Such studies should shed additional light on these increasingly important DNA metabolic pathways.

ABBREVIATIONS

ε

molar absorptivity

λ_max

lambda max

AAALAC

American Association for the Accreditation of Laboratory Animal Care

AID

activation-induced deaminase

APOBEC-1

apolipoprotein B mRNA editing enzyme catalytic polypeptide 1

BER

base excision repair

dUTP

2′-deoxyuridine 5′-triphosphate

ESI⁺

electrospray ionization in positive ion mode

FT-ICR MS

Fourier transform ion cyclotron resonance mass spectrometer

GC-MS/MS

gas chromatography/tandem mass spectrometry

GC-ToF-MS

gas chromatography time-of-flight mass spectrometry

HRMS

high-resolution mass spectrometry

hTDG

human thymine DNA glycosylase

[M + H]⁺

protonated molecule

MRM

multiple reaction monitoring mode

MTBSTFA

N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide

TBDMCS

tert-butyldimetheylchlorosilane

TBDMS

tert-butyl-dimethlylsilyl

TDG

thymine DNA glycosylase

TET

ten-eleven translocation