Simultaneous quantification of cytidine, methylcytidine, and hydroxymethylcytidine by isotope-dilution LC–MS/MS with application to mouse liver samples

Abstract

DNA methylation and hydroxymethylation are important epigenetic modifications that play key roles in cancer development and aging processes by regulating gene expression and genome stability. Traditionally, bisulfite conversion-based or antibody-based enzyme-linked immunosorbent assays are used to find DNA methylation. These tests are non specific, tedious, and not able to differentiate the difference between methylation and hydroxymethylation. To address these issues, we developed a sensitive, reproducible, and specific LC–MS/MS method for simultaneous quantification of two major DNA methylation products, 5-methyl-2′-deoxycytidine (5-mdC) and 5-hydroxymethyl-2′-deoxycytidine (5-hmdC), as well as 2′-deoxycytidine (2-dC), using corresponding stable isotope-labeled internal standards: 5-methyl-2′-deoxycytidine-d₃, 5-(hydroxymethyl)-2′-deoxycytidine-d₃, and 2′-deoxycytidine-¹³C,¹⁵N₂. We purified DNA samples from mouse liver tissue, broke them down with enzymes, filtered them, added internal standards, and then run them through a SCIEX 6500+ Triple Quad LC–MS/MS system with an Atlantis T3 C18 column under a binary gradient. The method showed great chromatographic separation and specificity, with MRM transitions of m/z 228.154 to 112.1 for 2-dC, 242.143 to 126.2 for 5-mdC, and 258.135 to 142.1 for 5-hmdC.Peak area ratio of analyte to internal standard exhibited linearity across calibration ranges of 5–5000 ng /mL for 2-dC, 0.5–500 ng/ mL for 5-mdC, and 0.05–10 ng/mL for 5-hmdC (R² > 0.999), using 2 μL injection and a total runtime of 9 min. The 5-hmdC level in female mouse liver significantly increased with aging from two to sixteen months old (0.0958 % to 0.1984 %; P < 0.001), whereas 5-mdC remained unchanged (3.47 % to 3.56 %; n.s.). These data confirm the accurate and reproducible quantification of DNA methylation and hydroxymethylation in tissue samples using the developed LC-MS/MS assay and indicate a broad application to cell culture and clinical biomarker studies.

1. Introduction

Epigenetic modifications are crucial systems that govern gene expression, maintain genome integrity, and influence cellular function without changing the fundamental DNA sequence. These chemical alterations are essential in processes including embryonic development, cell differentiation, and tissue-specific gene control. Aberrant epigenetic modifications have been associated with various human diseases, including cancer, neurological disorders, autoimmune diseases, and age-related ailments [1–3]. DNA methylation and hydroxymethylation are two of the many epigenetic changes that have gotten a lot of attention because they are important for normal physiology and are strongly linked to disease progression [4–6].

At the molecular level, 2′-deoxycytidine (2-dC) denotes the unmodified cytosine nucleoside in DNA and serves as a reference for evaluating cytosine alterations. The two principal derivatives, 5-methyl-2′-deoxycytidine (5-mdC) and 5-hydroxymethyl-2′-deoxycytidine (5-hmdC) represent direct products of DNA methylation and hydroxymethylation, respectively [6] (Fig. 1). Measuring these three nucleosides provides essential information regarding the degree and equilibrium of epigenetic changes within biological systems. Because there is a strong link between abnormal levels of 5-mdC and 5-hmdC and the progression of cancer and other diseases, it is important to measure these nucleosides accurately for mechanistic studies and to find clinical biomarkers [5,7,8].

Fig. 1.

Mechanistic pathway of cytosine modifications in DNA. The unmodified nucleoside 2-dC is converted to 5-mdC through enzymatic methylation catalyzed by DNA methyltransferases (DNMTs). 5-mdC is further transformed to 5-hmdC by Ten-eleven translocation (TET) enzymes via hydroxymethylation [6].

Conventional analytical methods for detecting DNA methylation and hydroxymethylation, despite their importance, possess significant limitations. Bisulfite conversion based approaches and sequencing methods are widely used; however, they are time-consuming and costly can introduce chemical artifacts, and cannot distinguish between methylated and hydroxymethylated cytosines [1,6]. Antibody-based enzyme--linked immunosorbent assays (ELISA) often lack sufficient specificity or repeatability, limiting their applicability for quantitative studies [8]. These problems show how important it is to have more strict, sensitive, and accurate methods for measuring cytosine nucleosides in biological samples [5,9].

Liquid chromatography coupled with tandem mass spectrometry (LC–MS/MS) has effectively addressed these challenges [2,3,5,7,9–11]. This method has higher sensitivity, selectivity, and repeatability, which makes it easier to measure 2-dC, 5-mdC, and 5-hmdC at the same time and with certainty in complex biological samples [10,11]. The utilization of stable isotope-labeled internal standards improves accuracy by compensating for variability in sample preparation and instrumental analysis [10]. Published LC–MS/MS assays have enabled sensitive and specific quantification of 2-dC, 5-mdC, and 5-hmdC in biological samples [5–11]; however, many of these methods required large DNA input, lacked complete sets of isotope-labeled internal standards, relied on long chromatographic runtimes, or did not fully validate matrix effects, carryover, and stability—limitations consistently noted across previous studies [5–12]. These gaps highlight the need for a more rigorous, low-input, and high-throughput LC–MS/MS method, which forms the rationale for the present work.

We developed and validated a sensitive, specific, and reproducible LC–MS/MS method for the simultaneous quantification of 2-dC, 5-mdC, and 5-hmdC in enzymatically digested DNA samples. The method was also used to look at DNA taken from mouse liver tissue. It showed great linearity, precision, and accuracy, and it worked effectively for high-throughput analysis. This method overcomes the constraints of traditional assays and gives researchers a reliable way to study DNA methylation and hydroxymethylation for basic and applied research [5–12].

To provide a proof-of-principle application of the developed assay in biosamples, liver tissues from young and aged mice were selected for the following reasons. First, liver is a relatively homogeneous organ, which facilitates robust benchmarking of assay performance and reproducibility without the confounding factor of cellular heterogeneity [13,14]. Second, hepatic DNA has been shown to undergo age-dependent epigenetic remodeling, making it a suitable biological system to evaluate in vivo changes in DNA methylation and hydroxymethylation [15, 16]. Finally, one of our research focuses is on the molecular regulation of liver biology and liver-associated diseases [17,18]. Establishing assay performance in mouse liver samples provides a relevant technological foundation for future mechanistic and disease-oriented research. Unlike many other matrices such as blood or brain, the liver is directly involved in metabolic homeostasis, detoxification, and age-associated molecular remodeling, making it one of the most responsive tissues for detecting biologically meaningful changes in global and modified cytosine levels.

2. Experimental

2.1. Materials and reagents

2′-Deoxycytidine (purity ≥99 %) was procured from Sigma-Aldrich Inc. (St. Louis, MO, USA). We got 5-(hydroxymethyl)-2′-deoxycytidine (purity ≥98 %) and 5-methyl-2′-deoxycytidine (purity ≥98 %) from Cayman Chemical in Ann Arbor, Michigan, USA. LC-MS grade methanol and water were obtained from VWR Chemicals BDH^® (Chicago, IL, USA). We got the stable isotope internal standards 2′-deoxycytidine-¹³C,¹⁵N₂ (isotopic purity 98.6 %), 5-(hydroxymethyl)-2′-deoxycytidine-d₃ (isotopic purity 99.6 %), and 5-methyl-2′-deoxycytidine-d₃ (isotopic purity 99.7 %) from Toronto Research Chemicals in Toronto, Canada. The rest of the chemicals were acquired from commercial vendors.

2.2. LC–MS/MS instrumentation and analytical conditions

The analysis was performed using a 6500+ Triple Quad LC–MS/MS System (SCIEX, CA, USA) fitted with a Waters^® Atlantis^™ T3 C18 column (2.1 × 100 mm, 3 μm). A binary gradient method was used to separate the chromatographic mixtures. The mobile phase A was made up of 2 mM ammonium acetate and 0.04 % acetic acid in water, while the mobile phase B was made up of 2 mM ammonium acetate and 0.04 % acetic acid in methanol (pH 3.5). The flow rate was sustained at 0.4 mL/ min, with an injection volume of 2 μL. The assay duration was 9 min, with data gathering occurring between 2.5 and 4.0 min. The gradient program proceeded as follows: mobile phase B was initially set at 0 % for 1.0 min, escalated to 15 % at 2.5 min, surged to 70 % at 4.5 min, maintained for 1.0 min, decreased to 5 % in 0.1 min, elevated to 80 % for 0.5 min, and ultimately reverted to 0 % for 1.5 min. R0 had 10 % methanol, R2 had 90 % methanol, and R3 had 50 % methanol. The autosampler temperature stayed at 4°C, and the column oven was set to 30°C. We got mass spectrometry data by using electrospray ionization (ESI) in positive ion detection mode with multiple reaction monitoring (MRM). The ion spray voltage was established at 5500 V, accompanied by a source temperature of 450°C. The curtain gas was regulated at 45 psi; gas 1 (nebulizer) and gas 2 (warmer) were calibrated to 65 psi and 50 psi, respectively; and the collision gas (CAD) was established at 9. Data acquisition was conducted using Analyst software version 1.6.3. The parameters for compound-specific MS/MS for the analytes and internal standards are detailed in (Table 1).

Table 1.

Compound dependent parameters for 2-dC, 5-mdC, 5-hmdC and (IS) 2’-Deoxy Cytidine- ¹³C,¹⁵N₂, 5-Methyl-2’-deoxy Cytidine-d3, 5-(Hydroxymethyl)-2′-deoxycytidine-d3.

Compound Name	Q1 (m/Z) (Da)	Q3 (m/Z) (Da)	DP (V)	EP (V)	CE (V)	CXP (V)
2-dC13C,15N2	231.11	115	46	10	19	10
5-mdC_d3	245.12	129.1	46	10	21	6
5-hmdC_ d3	261.11	145.1	41	10	17	8
2-dC	228.15	112.1	56	10	21	10
5-mdC	242.14	126.2	51	10	21	8
5-hmdC	258.13	142.1	20	10	17	8

Q1: precursor ion, Q3: fragmentation, DP: declustering potential; EP: entrance potential; CE: collision energy, CXP: collision cell exit potential.

2.3. Standard and quality control (QC) samples

Stock solutions of 2-dC, 5-mdC, and 5-hmdC were formulated in water at a concentration of 1 mg/mL, respectively, and preserved at −20°C. Native nucleoside standards were prepared and stored in water due to their high aqueous solubility and stability at the stock-solution concentration. Stock solutions of the internal standard (IS, 2′-Deoxycytidine-¹³C,¹⁵N₂, 5-(Hydroxymethyl)-2′-deoxycytidine-d₃, and 5-methyl-2′-deoxycytidine-d₃) were formulated at a concentration of 1 mg/mL for 2-dC ¹³C,¹⁵N₂ in a 50 % methanol-water mixture, for 5-Methyl-2′-deoxycytidine-d₃ in methanol with 5 % DMSO, and for 5-(Hydroxymethyl)-2′-deoxycytidine-d₃ in a 50 % methanol-water solution containing 5 % DMSO. . These solvent compositions were selected in accordance with manufacturer recommendations and to ensure complete dissolution and long-term stability of the isotopically labeled standards at high stock concentrations. A mixed internal standard working solution was formulated in 50 % methanol with final concentrations of 2,000 ng/mL for 2-dC-¹³C,¹⁵N₂, 100 ng/mL for 5-mdC-d₃, and 80 ng/mL for 5-hmdC-d₃ and was added at a constant volume to all samples. The use of organic cosolvents was limited to the stock and internal standard working solutions.

Intermediate working solutions were generated from the primary analyte stocks and subsequently diluted in water to provide mixed working standards over a wide concentration range. All working standards and QC samples were prepared by appropriate dilution into aqueous matrices prior to analysis. The ultimate calibration standards were 50/5/0.5, 100/10/1, 500/50/2.5, 2500/250/5, 10000/1000/10, 25000/2500/50, and 50000/5000/100 ng/mL for 2-dC, 5-mdC, and 5-hmdC, respectively. Calibration standards were established by introducing these working solutions into a blank matrix with an internal standard, encompassing standard curve ranges of 5–5000 ng/mL for 2-dC, 0.5-500 ng/mL for 5-mdC, and 0.05–10 ng/mL for 5-hmdC.

The quality control samples were made individually by diluting stock solutions of each analyte with a blank matrix. The nominal concentrations were 5 ng/mL (LLOQ), 15 ng/mL (LQC), 500 ng/mL (MQC), and 4000 ng/mL (HQC) for 2-dC; 0.5 ng/mL (LLOQ), 1.5 ng/mL (LQC), 50 ng/mL (MQC), and 400 ng/mL (HQC) for 5-mdC; and 0.05 ng/mL (LLOQ), 0.15 ng/mL (LQC), 1 ng/mL (MQC), and 8 ng/mL (HQC) for 5-hmdC.

2.4. DNA sample preparation and processing

Liver samples were collected from C57BL/6J female mice at two months old (F1-4) and sixteen months old (F10-13) by Dr. Xiaoqin Liu at the Texas A&M University Health Science Center in Houston. Animal usage was reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) (approval number: 2021-0264-IBT) and performed in accordance and compliance with all relevant regulatory and institutional guidelines. Genomic DNAs were purified using genomic DNA extraction kit (TaKaRa Bio. 740952.250). Four micrograms of mouse liver DNA were enzymatically digested using a combination of Benzonase (Sigma E1014-5KU, ≥ 250 U/μL), phosphodiesterase (Sigma P3243), and alkaline phosphatase (Sigma P7923-2KU) in a Tris buffer solution (100 mM NaCl, 20 mM MgCl₂, 20 mM Tris, pH 7.9). The samples were incubated at 37°C for 6 h, and each reaction had 108 μL of buffer and 2 μL of enzyme mix. Negative control experiments were conducted concurrently with the identical enzyme mixture devoid of DNA. After digestion, all materials were preserved at − 80°C until additional processing.

Amicon Ultra centrifugal filters (0.5 mL, 10K NMWL, regenerated cellulose) and the centrifuge were pre-chilled to 4°C before utilization. Samples were processed in filters and centrifuged at 14,000 rcf for 30 min at 4°C. A 45 μL portion of the filtrate was subsequently transferred to a microcentrifuge tube and augmented with 2.5 μL ofa mixed internal standard solution containing 2 μg/mL 2-dC-¹³C,¹⁵N₂, 100 ng/mL 5-mdC-d₃, and 80 ng/mL 5-hmdC-d₃. The mixture was completely vortexed and subsequently put into vials for LC–MS/MS analysis.

To prepare calibration standards and QC samples, 45 μL of filtered negative control matrix was augmented with 5 μL of a mixed working standard solution comprising 2-dC, 5-mdC, and 5-hmdC, followed by the incorporation of 2.5 μL of the mixed internal standard as illustrated in (Fig. 2). The solution was vortexed and subsequently placed into vials for LC–MS/MS analysis.

Fig. 2.

Schematic reflecting sample preparation and analysis.

2.5. Method validation

The method was validated in accordance with the US Food and Drug Administration (FDA) Bioanalytical Method Validation Guidelines for Industry (Food and Drug Administration, 2018) [19].

2.5.1. Calibration curve and carry-over

Calibration curves were constructed by plotting the analyte-to–internal-standard peak-area ratio against nominal concentration and were generated using weighted (1/x) least-squares linear regression. The slope, intercept, and coefficient of determination (r²) were used to evaluate linearity across the validated concentration ranges. Carryover was assessed by injecting six consecutive upper-limit-of-quantification (ULOQ) samples—50,000 ng mL⁻ for 2-dC, 5,000 ng mL⁻¹ for 5-mdC, and 100 ng mL⁻¹ for 5-hmdC—followed by duplicate blank injections to confirm the absence of residual analyte signal.

2.5.2. Accuracy and precision

Accuracy and precision were assessed using replicate analyses of quality-control (QC) samples: 2-dC LLOQ 5, LQC 15, MQC 500, and HQC 4,000 ng mL⁻¹; 5-mdC LLOQ 0.5, LQC 1.5, MQC 50, and HQC 400 ng mL⁻¹; and 5-hmdC LLOQ 0.05, LQC 0.15, MQC 1, and HQC 8 ng mL⁻¹. Replicates were analyzed on a single validation day (intra-day, n = 6) and across three validation days (inter-day, n = 18). Accuracy (RE %) and precision (CV %) were calculated using standard bioanalytical definitions, where C‾_measured and C_nominal represent the mean measured and theoretical concentrations, respectively. Acceptance limits were ±15 % for QC samples and ±20 % for LLOQ, consistent with FDA criteria

2.5.3. Matrix effect

Matrix effects were evaluated for 2-dC, 5-mdC, 5-hmdC, and the internal standards at three quality-control (QC) concentration levels. The potential influence of endogenous matrix components on ionization efficiency was assessed by comparing the peak area of the analyte spiked into the matrix after preparation (A_spiked) with that of the same analyte concentration prepared in neat solvent (A_solvent). The matrix effect (ME %) was calculated using standard bioanalytical definitions, where A_spiked and A_solvent represent the respective peak areas of the analyte in matrix and solvent. Because the digested DNA samples were injected directly after the addition of internal standards and no extraction or cleanup procedure was performed, evaluation of extraction recovery was not applicable, as all analytes remained fully retained in solution throughout the preparation workflow.

2.5.4. Stability

The bench top and long-term stabilities were assessed by positioning spiked QC samples on the bench top at ambient temperature for 6 h and in a freezer at −80°C for 10 days, respectively. The stability of the autosampler was evaluated by maintaining spiked QC samples in the autosampler tray at 4°C for 24 h. The freeze–thaw stability was determined by subjecting spiked QC samples to three freeze–thaw cycles, thawing at 4°C for 1–2 h, followed by re-freezing at −80°C for 12–24 h. Accuracy and precision for all stability tests were expressed as RE (%) and CV (%), calculated using standard bioanalytical definitions.

3. Results and discussion

3.1. Method development

3.1.1. MS/MS optimization

Electrospray ionization in the positive ion mode yielded maximal peak intensity for all analytes. (Fig. 3) illustrates the MS2 spectra of 2-dC, 5-mdC, and 5-hmdC, while (Fig. 4) presents the matching stable isotope-labeled internal standards. Collision-induced fragmentation produced numerous stable product ions, and the optimal MRM transitions were identified: m/z 228.154 to 112.1 for 2-dC, 242.143 to126.2 for 5-mdC, and 258.135 to 142.1 for 5-hmdC. The internal standards were assessed at m/z 231.109 to 115.0 (2-dC-¹³C,¹⁵N2), 245.122 to 129.1 (5-mdC-d₃), and 261.112 to 145.1 (5-hmdC-d₃). Source conditions were refined to attain elevated sensitivity and a consistent signal response.

Fig. 3.

MS2 spectra of (A) 2-dC, (B) 5-mdC, and (C) 5-hmdC on a Triple Quad LC–MS/MS for the identification of nucleoside standards.

Fig. 4.

MS2 spectra of internal standards: (A) 2-dC-¹³C,¹⁵N₂, (B) 5-mdC-d₃, and (C) 5-hmdC-d₃ on a Triple Quad LC–MS/MS for the identification of isotope-labeled nucleoside internal standards.

Previous LC–MS/MS studies often relied on a single internal standard or lacked isotope-labeled analogs for each nucleoside, contributing to quantification variability [5–8]. In contrast, the present method employs three structurally matched isotope-labeled internal standards, providing improved correction for matrix effects and injection-related variability, in accordance with current recommendations for isotope-dilution LC–MS/MS–based nucleoside quantification [9–11]. To evaluate the robustness and reproducibility of the enzymatic DNA digestion process, a DNA input titration experiment was performed in which genomic DNA amounts were varied from 0.1 to 8 μg while maintaining identical digestion conditions, including fixed enzyme composition and volume, buffer composition, incubation time, and temperature. Under these controlled conditions, the measured concentrations of digestion products (2-dC, 5-mdC, and 5-hmdC) increased proportionally with DNA input, and the relative composition metrics (5-mdC % and 5-hmdC %) remained stable across the working range, indicating reproducible and effectively complete hydrolysis without DNA-input–dependent bias. Increased variability for 5-hmdC was observed only at very low DNA inputs, consistent with proximity to the method’s sensitivity limit. Based on these results, a DNA input of 4 μg with 2 μL enzyme mixture was selected as the standard digestion condition for all subsequent experiments. Because the isotopically labeled internal standards are introduced at the nucleoside level after digestion, they are intended to correct analytical variability arising downstream of hydrolysis rather than digestion efficiency per se. To further mitigate the impact of any minor variation in absolute digestion yield, modified cytosines were quantified as percentages normalized to total cytosines.

3.1.2. Column and mobile-phase optimization

In the development of the LC–MS/MS method, we methodically refined the mobile phase composition to ensure dependable separation, consistent ionization, and low interference among 2-dC, 5-mdC, and 5-hmdC. Preliminary experiments utilizing 2 mM ammonium acetate and methanol led to premature elution of analytes and significant matrix interference. The use of 0.05–0.1 % acetic acid improved peak morphology and marginally increased retention but failed to adequately distinguish 2-dC from baseline noise. Substituting ammonium acetate with 0.1 % formic acid enhanced ionization efficiency, resulting in sharper and more symmetrical peaks; still, 2-dC persisted in eluting prematurely. A further improved mobile phase, 2 mM ammonium acetate with 0.04 % acetic acid in water (A) and methanol (B), provided an optimal equilibrium between ionization stability and analyte retention. This greatly reduced matrix effects and improved the overall effectiveness of the chromatography.

Column screening was conducted to improve peak resolution and retention. The CORTECS^® UPLC T3 column (2.1 × 50 mm, 1.6 μm) yielded satisfactory peak morphologies; however, it exhibited co-elution and slight interference. The Agilent ZORBAX Eclipse Plus C18 (2.1 × 100 mm, 1.8 μm) enhanced retention and diminished baseline noise but displayed variable reproducibility. The Waters^® Atlantis^™ T3 C18 column (2.1 × 100 mm, 3 μm) has a polar-endcapped stationary phase that works well in water and gave the most stable baseline and perfect separation of 2-dC, 5-mdC, and 5-hmdC. This column was chosen because it is reliable, exhibits reduce matrix interference, and is suitable for high-throughput analysis.

Most prior published LC–MS/MS assays used C18 or HILIC columns but often reported constraints such as peak tailing, long equilibration, or inefficient retention of 5-hmdC [5–7,10]. UHPLC–MS approaches using core-shell particles [3] provided faster runs but were more susceptible to matrix effects. The polar-endcapped Atlantis T3 column used here offers enhanced aqueous compatibility, aligning with modern approaches for low-retention nucleosides [9,11] and improving upon older methods that required ion-pairing reagents [1], which complicate MS detection.

3.1.3. Gradient optimization

The gradient was optimized to achieve complete separation within a short run time while maintaining robustness for high-throughput operation. After elution of all target nucleosides at 70 % mobile phase B, the gradient was rapidly reduced to 5 % B within 0.5 min to wash the stationary phase under highly aqueous conditions. This step improves retention-time stability and minimizes gradient-history effects that are particularly pronounced for highly polar nucleosides. Subsequently, the gradient was increased to 80 % B to re-equilibrate the column and remove strongly retained matrix components generated during enzymatic DNA digestion, thereby reducing carryover and improving injection-to-injection reproducibility. This gradient produced sharp, symmetrical peaks, minimal matrix interference, and a total runtime of 9 min, establishing a high-throughput workflow.

Many earlier published LC–MS/MS workflows required 15–30 min runs [5–8] or relied on nano-LC systems with longer equilibration times and reduced robustness [10,11]. The current method achieves equivalent or better sensitivity with significantly shorter run times, minimizing instrument downtime and improving batch throughput.

In addition, most previously published LC–MS/MS assays required ≥10 µg genomic DNA, multi-step digestion or SPE cleanup, or long gradients exceeding 15 min [5–8,10]. The present method quantifies 2-dC, 5-mdC, and 5-hmdC from ~4 μg DNA samples, using a simple single-tube digestion and direct injection. Sensitivity of this assay (LOD ranged from 0.02 to 2 ng/mL) exceeds earlier LC–MS/MS values and approaches nano-flow LC–MS/MS performance reported by Qiu et al [11], but without the need for nano-flow instrumentation. Many previous studies lacked comprehensive LC-MS/MS assay validation of matrix effects, carryover, or stability [5–8,10,19], whereas the present workflow integrates full FDA-style validation with matched isotope-labeled internal standards. Collectively, this places the current method among the most analytically rigorous and high-throughput LC–MS/MS approaches for cytosine-modification profiling.

3.2. Method validation

3.2.1. Linearity, sensitivity, and lower limits

We developed calibration standards by using a negative-control digest matrix that had the same enzyme mix and Tris buffer as the DNA digestion, but no genomic DNA. The enzymes were Benzonase, phosphodiesterase, alkaline phosphatase, and Tris buffer (100 mM NaCl, 20 mM MgCl₂, 20 mM Tris, pH 7.9). This blank digest had a background that matched the matrix and did not have any detectable endogenous 2-dC, 5-mdC, or 5-hmdC. This confirmed that it was good for making calibration curves. Although this digest does not contain endogenous nucleosides, its consistent composition and absence of background signals made it suitable for calibration. Calibration curves prepared in the negative control digest matrix demonstrated excellent linearity for all analytes, with correlation coefficients exceeding (r² > 0.999) within the validated ranges of 5–5000 ng/mL for 2-dC, 0.5–500 ng/mL for 5-mdC, and 0.05–10 ng/mL for 5-hmdC .

The calibration curves were generated using weighted (1/x) least-squares linear regression and met the acceptance criteria for bioanalytical method validation, confirming reliable quantitative performance across the evaluated concentration ranges. Chromatograms exhibited clear baselines in blanks and distinctly defined peaks in both standards and samples. The quantities of 2-dC, 5-mdC, and 5-hmdC in two groups of mouse liver DNA samples (F1–F4 and F10–F13) were 711–1010, 22.7–40.1, and 0.688–2.15 ng/mL, respectively, with no discernible signals in negative controls (Table 2).

Table 2.

Concentrations of 2-dC, 5-mdC, and 5-hmdC in mouse liver DNA at two months old (F1–F4) and sixteen months old (F10–F13).

Sample No.	Sample Description	Concentration (ng/ml)
		2-dC	5-mdC	5-hmdC
1	F1_mouse liver	713	27.2	0.69
2	F2_mouse liver	824	33.7	0.78
3	F3_mouse liver	838	31.3	0.88
4	F4_mouse liver	817	22.7	0.83
5	F10_mouse liver	821	26.9	1.57
6	F11_mouse liver	1010	40.1	1.70
7	F12_mouse liver	711	30	1.65
8	F13_mouse liver	923	30.6	2.15
9	Negative Control	N.D	N.D	N.D

^*N.D= Not detected.

The lowest limits of quantification (LLOQs) were 5 ng/mL for 2-dC, 0.5 ng/mL for 5-mdC, and 0.05 ng/mL for 5-hmdC. These were based on a signal-to-noise ratio of 5 or higher. The lower limits of detection (LLODs), which are the lowest concentrations of an analyte that can be reliably distinguished from baseline noise at a signal-to-noise ratio of ≥3, were estimated to be about 2 ng/mL for 2-dC, 0.2 ng/mL for 5-mdC, and 0.02 ng/mL for 5-hmdC. While matrix-induced ion suppression may influence the apparent LLOD, the LLOQs remain linear and reliable due to isotope-dilution–based internal standard correction, which effectively cancels matrix-related bias and preserves quantitative accuracy.

Previously published LC–MS/MS methods for global cytosine modification analysis report a range of detection and quantification limits for 5-mdC and 5-hmdC. Earlier studies reported sensitivity in the femtomole range; for example, Le et al [5] reported detection limits of approximately 0.5 fmol per sample for both 5-mdC and 5-hmdC using conventional-flow LC–MS/MS, while Chen et al [7] reported LLOD/LLOQ values of 0.06/0.20 fmol for 5-mdC and 0.19/0.64 fmol for 5-hmdC using capillary LC–MS/MS. More recently, Qiu et al [11] reported LLOD/LLOQ values of 4/12 nM for 5-mdC and 3/10 nM for 5-hmdC using nano-LC–MS/MS with sample stacking, but this approach required nano-flow instrumentation, longer equilibration times, and higher DNA input. In contrast, the present method achieved LLOQs of 5 ng/mL for 2-dC, 0.5 ng/mL for 5-mdC, and 0.05 ng/mL for 5-hmdC, with corresponding LLODs of approximately 2, 0.2, and 0.02 ng/mL, respectively, using approximately 4 μg of genomic DNA, a 2 μL injection volume, and a 9-min chromatographic run under conventional-flow LC–MS/MS conditions. This shows how sensitive the method is. The sensitivity levels made it possible to accurately find both common and rare nucleosides, making the method a reliable and accurate way to find methylation and hydroxymethylation of cytidine nucleosides in biological samples.

3.2.2. Selectivity and carryover

Double blanks, IS-spiked blanks, and QC samples showed no endogenous interference at analyte or internal-standard retention times (Fig. 5). Post-run blanks confirmed no detectable carryover, indicating that the injection system effectively prevented analyte accumulation. Earlier published LC–MS/MS workflows frequently reported ion suppression or co-eluting matrix background [5–8], but the optimized chromatographic retention and use of three isotope-matched standards ensure strong selectivity in the present method.

Fig. 5.

Representative MRM chromatograms of nucleosides: (A) double blank negative control; (B) blank spiked with internal standards; (C–E) negative control spiked with 2-dC, 5-mdC, and 5-hmdC with their internal standards; (F–H) enzymatically digested mouse liver DNA showing endogenous 2-dC, 5-mdC, and 5-hmdC with internal standards.

3.2.3. Accuracy and precision

Accuracy and precision outcomes for all analytes are summarized in (Table 3), with every QC level meeting U.S. FDA acceptance criteria for bioanalytical methods [19]. For 2-dC, intra-day accuracy ranged from −5.63 % to 0.90 %, with corresponding precision values of 1.27–5.69 %. For 5-mdC, intra-day accuracy varied between −5.20 % and 6.67 %, and precision ranged from 1.19–8.62 %. The more oxidized nucleoside, 5-hmdC, exhibited intra-day accuracy between −6.05 % and 3.00 %, with precision values of 5.69–9.25 %.

Table 3.

Intra-day and inter-day accuracy and precision of 2-dC, 5-mdC, and 5-hmdC in negative control.

Compound Name	QC	Nominal Concentration (ng/mL)	Intra-day (n=6)			Inter-day (n=18)
			Observed Concentration (Mean ± SD)	Accuracy (RE %)	Precision (CV %)	Observed Concentration (Mean ± SD)	Accuracy (RE %)	Precision (CV %)
2-dC	LLOQ	5	5.0 ± 0.2	0.9	4.62	4.9 ± 0.23	−2.68	4.76
	LQC	15	14.6 ± 0.2	−2.44	1.27	14.5 ± 0.44	−3.52	3.02
	MQC	500	486.5 ± 10.8	−2.7	2.21	500.2 ± 18.64	0.04	3.70
	HQC	4000	3775 ± 214.6	−5.63	5.69	3917 ± 167.2	−2.08	4.27
5-mdC	LLOQ	0.5	0.5 ± 0.03	0.67	6.50	0.47 ± 0.03	−5.73	6.77
	LQC	1.5	1.60 ± 0.02	6.67	1.19	1.50 ± 0.07	−0.22	4.47
	MQC	50	47.40 ± 1.5	−5.20	3.14	50.37 ± 1.99	0.76	3.93
	HQC	400	382.8 ± 33.01	−4.29	8.62	391.8 ± 22.47	−2.04	5.74
5-hmdC	LLOQ	0.05	0.05 ± 0.004	−5.03	9.25	0.05 ± 0.003	−8.81	7.77
	LQC	0.15	0.15 ± 0.009	3.00	5.69	0.14 ± 0.01	−3.98	6.83
	MQC	1	0.94 ± 0.058	−6.05	6.15	0.98 ± 0.06	−1.46	5.63
	HQC	8	7.59 ± 0.448	−5.13	5.91	7.40 ± 0.35	−7.50	4.81

Inter-day variability across all QC levels remained low for each analyte, reflecting strong reproducibility of both the chromatographic separation and the instrument response over multiple days. This level of inter-day, multi-concentration validation is more comprehensive than what has been reported in several earlier LC–MS/MS studies [5–8], many of which presented only same-day precision data or lacked full QC-tier reporting. In contrast, the present method includes a complete assessment aligned with FDA bioanalytical method-validation requirements, demonstrating robust and reliable quantification across independent analytical runs.

3.2.4. Matrix effects

Matrix effects were evaluated using a negative-control digest matrix prepared from the enzyme mix and buffer without genomic DNA. This digest, which contained only the digestion enzymes and buffer but no nucleosides, was used to prepare standard curve samples by spiking known concentration of nucleosides. It served as a surrogate matrix because it mimics the chemical conditions of the digestion process while providing a clean background for preparing calibration standards. Although it does not contain the endogenous nucleosides present in real DNA samples, it offered a consistent and interference-free baseline for assessing analyte responses.

When the analyte-to–internal-standard peak-area ratios obtained from this surrogate digest were compared with those from DNA-containing samples, the matrix-effect values summarized in (Table 4) showed that the DNA digest produced stronger ion suppression. To better understand this difference, we compared how the isotope-labeled internal standards behaved in both matrices. The internal standards consistently showed lower peak areas in DNA digests than in the surrogate matrix, indicating that components released during DNA digestion contributed to additional ion suppression.

Table 4.

Matrix effect of analyte in negative control.

Compound Name	QC	Nominal concentration(ng/mL)	Matrix Effect (%) (n=6)
2-dC	LQC	15	88
	MQC	500	87
	HQC	4000	96
	IS	2000	85
5-mdC	LQC	1.5	84
	MQC	50	84
	HQC	400	90
	IS	100	71
5-hmdC	LQC	0.15	100
	MQC	1	92
	HQC	8	86
	IS	80	86

To determine whether this suppression affected quantification, we evaluated the analytes along with their corresponding isotope-labeled internal standards in both matrices. When known amounts of each analyte were spiked directly into DNA-containing samples, both the analyte and the internal standard exhibited lower absolute peak areas compared with the surrogate digest. However, the analyte-to-internal-standard ratios remained nearly unchanged between the two matrices. This demonstrates that the isotope-labeled internal standards experienced suppression proportional to their analytes, preserving the relative response needed for accurate quantification.

Since isotope-dilution LC–MS/MS relies on analyte-to-IS ratios rather than absolute signal intensity, these results confirm that calibration curves prepared in the surrogate digest matrix can be reliably applied to DNA-containing samples. Overall, despite the stronger ion suppression in the biological matrix, the use of isotope-labeled internal standards effectively compensates for matrix differences and supports accurate measurement of 2-dC, 5-mdC, and 5-hmdC in DNA samples.

Although the surrogate digest does not contain endogenous matrix components, the close agreement in peak-area ratios demonstrates that the assay maintains consistent ionization conditions across sample types. In contrast, many earlier LC–MS/MS workflows for global DNA methylation reported substantial ion suppression or required solidphase extraction to reduce matrix interference [5–8,10]. The present method achieves acceptable matrix performance without additional cleanup steps, supporting reliable and high-throughput quantification.

3.2.5. Stability

Stability studies confirmed that all three nucleosides remained stable under typical laboratory handling and storage conditions (Table 5). Short-term stability at room temperature for 6 h produced recoveries between 85–115 % of nominal concentrations. The observed decrease after bench-top exposure reflects limited chemical stability under ambient conditions and the presence of the enzymatic digestion matrix; however, these changes remain acceptable according to FDA bio-analytical method validation guidance and do not compromise method reliability. Processed samples kept in the autosampler at 4°C for 24 h remained within FDA-accepted limits. Following three freeze–thaw cycles, variability stayed within ±15 %, indicating that repeated handling did not compromise analyte integrity. Long-term stability testing demonstrated that samples stored at −80°C for up to 10 days retained concentrations within 10 % of their initial values.

Table 5.

Stability data for 2-dC, 5-mdC and 5-hmdC in negative Control.

Compound Name	Stability Test	Nominal Concentration (ng/ml)	Observed Concentration (Mean ± SD)	Accuracy (RE %)	Precision (CV %)
2-dC	Auto-Sampler (4°C, 24h)	15	14.03 ± 0.66	−6.5	4.71
		500	494 ± 12	−1.2	2.43
		4000	3990 ± 264.56	−0.25	6.63
	Bench top (6 h)	15	12.98 ± 0.74	−13.5	5.71
		500	447 ± 16.52	−13.5	5.71
		4000	3673 ± 61.10	−8.17	1.66
	Freeze and thaw (−80°C to 4°C, 3 cycles	15	15.65 ± 0.37	4.33	2.36
		500	511.3 ± 9.02	2.27	1.76
		4000	4280 ± 51.96	7	1.21
	Long-term (−80°C, 10 days)	15	16.02 ± 0.46	6.83	2.85
		500	516 ± 10.15	3.20	1.97
		4000	4260 ± 175.7	6.50	4.13
5-mdC	Auto-Sampler (4°C, 24 h)	1.5	1.39 ± 0.04	−7.16	3.18
		50	51.97 ± 0.51	3.93	0.99
		400	409.7 ± 30.62	2.42	7.47
	Bench top (6 h)	1.5	1.29 ± 0.11	−14	8.39
		50	42.86 ± 1.83	−14.27	4.28
		400	345 ± 2.65	−13.75	0.77
	Freeze and thaw (−80°C to 4°C, 3 cycles	1.5	1.59 ± 0.03	5.67	1.59
		50	52.5 ± 2.23	5.00	4.26
		400	395 ± 5.20	−1.25	1.32
	Long-term (−80°C, 10 days)	1.5	1.63 ± 0.09	8.3	5.38
		50	51.57 ± 2.64	3.1	5.12
		400	414 ± 19.97	3.5	4.82
5-hmdC	Auto-Sampler (4°C, 24 h)	0.15	0.15 ± 0.01	−1.7	9.33
		1	0.99 ± 0.05	−1.5	5.16
		8	7.79 ± 0.55	−2.6	7.06
	Bench top (6 h)	0.15	0.13 ± 0.02	−16.22	12.34
		1	0.87 ± 0.04	−13.00	5.01
		8	7.2 ± 0.16	−10.04	2.21
	Freeze and thaw (−80°C to 4°C, 3 cycles	0.15	0.15 ± 0.01	0.67	4.59
		1	1.10 ± 0.05	10.33	4.56
		8	9.10 ± 0.10	13.79	1.04
	Long-term (−80°C, 10 days)	0.15	0.17 ± 0.02	12	10.14
		1	0.96 ± 0.02	−3.8	1.73
		8	8.57 ± 0.34	7.08	3.91

Most previously published LC–MS/MS assays for global DNA methylation did not report complete stability assessments—particularly freeze–thaw cycling, autosampler stability, or multi-day long-term storage [5–8,10]. In contrast, the present method includes full stability evaluation consistent with FDA bioanalytical guidelines [19], ensuring reliable quantification across routine and extended workflows.

3.3. DNA methylation and hydroxymethylation in aging livers

Changes in DNA methylation occur in many biological processes in multiple tissues, including liver aging, regeneration, and fibrosis progression [18,20]. Here, we applied this LC–MS/MS method to quantify the global changes in DNA methylation and hydroxymethylation in young (two months old) and old (16 months old) mouse livers. 5-mdC and 5-hmdC levels were calculated as percentages of the total amount of cytosines, which include 2-dC, 5-mdC, and 5-hmdC (Fig. 6). Our results demonstrated that young and old livers show no significant difference in their 5-mdC levels, making up 3.47 % and 3.56 % of their deoxycytidines, respectively(n.s.). Notably, the 5-hmdC level was significantly increased in the aged liver (0.198 %) compared to the young liver (0.096 %), with a mean difference of 0.103 % and this increase corresponded with the statistical result shown in (Fig. 6) (P < 0.001). The observed age-associated elevation of 5-hmdC, in the absence of detectable changes in 5-mdC, aligns with previously reported LC–MS/MS data from mouse liver tissue [21]. It should be noted that the number of biological replicates analyzed in each age group was limited (n = 4), and therefore these observations should be interpreted as exploratory and descriptive rather than as evidence of biologically definitive age-associated changes.

Fig. 6.

The abundance of 5-mdC and 5-hmdC in two and sixteen months old (mo) mouse livers. (A) The percentages of endogenous 5-mdC in young (2 mo) and old (16 mo) mouse livers. (B) The percentages of endogenous 5-hmdC in young (2 mo) and old (16 mo) mouse livers. n=4, ***P < 0.001, n.s. not significant.

To place these findings in context, earlier studies have consistently shown that 5-hmdC exhibits dynamic variation during development, differentiation, and aging. Bachman et al. (2015) [22] showed that oxidized cytosine derivatives, like 5-hmdC, build up over time as tissues mature. This suggests that hydroxymethylation is a stable epigenetic mark and not just a step in the process of demethylation. Occean et al. (2024) [21] similarly found that hepatic 5-hmdC rises with age in mice across various tissue types, whereas 5-mdC remains mostly stable, indicating selective modulation of cytosine oxidation pathways throughout aging.

The trend in our dataset—stable 5-mdC and high 5-hmdC—shows a biological pattern that has been preserved and supports the idea that aging has a bigger effect on TET-mediated oxidation than on DNMT-mediated methylation. Our method’s ability to identify these small but biologically significant changes shows that it is a good choice for epigenetic applications where it is important to tell the difference between closely related cytosine modifications.

Collectively, these advancements render the current methodology a viable, reproducible, and regulatory-compliant framework for high-throughput analysis of DNA methylation and hydroxymethylation under conditions of limited sample availability. This LC–MS/MS assay addresses the primary limitations of earlier LC–MS and immunoassay techniques by integrating low sample demand, single-tube preparation, isotope-corrected quantitation, a brief nine-minute run, and comprehensive FDA-style validation, thereby offering a reliable analytical instrument for epigenetic and translational research.

4. Conclusion

This paper reports the development and validation of a quick, sensitive, and highly specific LC–MS/MS approach for the simultaneous quantification of 2-dC, 5-mdC, and 5-hmdC in biological samples. This method exhibits exceptional linearity, accuracy, precision, and stability, with negligible matrix interference, hence affirming its suitability for routine bioanalytical applications. The assay addresses significant limitations of bisulfite conversion and antibody-based approaches by offering direct and reproducible measurements of DNA methylation and hydroxymethylation, which frequently suffer from issues of specificity or reproducibility. This work creates a dependable analytical platform that enhances both fundamental epigenetic research and translational applications, presenting considerable potential for biomarker discovery and clinical implementation.

Data availability

Data will be made available on request.