Tracking Decitabine Incorporation into Malignant Myeloid Cell DNA in vitro and in vivo by LC-MS/MS with Enzymatic Digestion

The DNA hypomethylating agents decitabine and 5-azacytidine are the only two drugs approved for treatment of all subtypes of the myeloid malignancy myelodysplastic syndromes (MDS). The key to drug activity is incorporation into target cell DNA, however, a practical method to measure this incorporation is un-available. Here, we report a sensitive and specific LC-MS/MS method to simultaneously measure decitabine incorporation and DNA hypomethylation. A stable heavy isotope of 2′-deoxycytidine was used as an internal standard and one-step multi-enzyme digestion was used to release the DNA bound drug. Enzyme-released decitabine along with other mononucleosides were separated by a reverse-phase C18 column and quantified by mass spectrometry using multiple-reaction-monitoring (MRM) mode, with a lower limit of quantitation at 1.00 nM. In vitro studies demonstrated dosage and time-dependent incorporation of decitabine into myeloid leukemia cell DNA that correlated with extent of DNA hypomethylation. When applied to clinical samples serially collected from MDS patients treated with decitabine, the method again demonstrated correlation between decitabine DNA-incorporation and DNA hypomethylation. This novel assay to measure the intended molecular pharmacodynamic effect of decitabine therapy can therefore potentially provide insights into mechanisms underlying sensitivity versus resistance to therapy.

into cellular DNA can potentially guide individualization of therapy towards achievement of intended molecular pharmacodynamic effects, and provide insights into sensitivity versus resistance of malignant cells, which could be from failure to produce these intended molecular pharmacodynamic effects. Although decitabine incorporation into DNA has previously been measured using radioisotopic and liquid scintillation counting assays 11,12 , these previous methods had either limited sensitivity and used radioactive isotope labeled decitabine, which is not practical for use in clinical settings. LC-MS/MS method have been developed to quantify decitabine concentration in plasma 13 however, this method is not applicable to measure decitabine incorporation in DNA and is less sensitive to use in clinical application. Another LC-MS/MS method has been developed to quantify total intracellular decitabine nucleotide concentrations, as well as DNA incorporation levels of decitabine 14 , but is less sensitive and correlation of DNA incorporation to hypomethylation was not demonstrated.
To address this unmet need, here we developed a sensitive LC-MS/MS method to simultaneously quantify incorporation of decitabine into DNA and DNA methylation, by measuring dG, dC, and 5mdC. The method was applied both in vitro and in vivo. The results demonstrated that this method can quantify the incorporation of decitabine into DNA, enabling quantification of the variation in the incorporation and response to treatment in different models and in vivo. This assay could be a useful tool for the purpose of understanding treatment sensitivity versus resistance and provide important guidance towards an overall goal of individualizing and optimizing therapy with this exclusive class of agent.

Results
enzymatic hydrolysis of DNA. In this work, the DNA samples were hydrolyzed by a one-step tetra-enzyme reaction (Fig. 1). This one-step enzyme reaction had been optimized and validated previously to retain 100% digestion efficiency in comparison to an optimized stepwise enzyme reaction 15 . Among the four enzymes employed, DNase I is an endonuclease that splits phosphodiester bonds of DNA and yields oligonucleotides with a free 3′-end hydroxyl group and a free 5′-end phosphate group; PDE I is a 3′ to 5′ exonuclease that successively hydrolyzes an oligonucleotide from 5′-end to 3′-end and produce deoxynucleoside 5′-phosphate; NP1 is a 5′ to 3′ exonuclease that acts in opposite direction to PDE I, and completely hydrolyzes an oligonucleotide from 3′-end to 5′-end to produce deoxynucleoside 5′-phosphate; and ALP is a hydrolase that hydrolyzes phosphate groups of deoxynucleotides to deoxynucleosides. Since the electrospray ionization tandem mass spectrometer (ESI-MS/ MS) used for this work had much lower limits of quantitation for deoxynucleosides than those of deoxynucleotides; therefore, deoxynucleosides were the sought-after products for the enzyme digestion.
Enzyme digestion efficiency was evaluated by comparing the DNA concentration calculated from the amount of dG produced from DNA hydrolysis with the DNA concentration measured using UV spectroscopy. The rationale for the study is that DNA concentrations can be either measured by UV spectroscopy or calculated from the amount of a nucleoside released from hydrolysis and measured by LC-MS/MS 15 . In this work, the concentration of dG released from hydrolysis of calf DNA (250 mg/L measured by UV spectroscopy) of was used to calculate the concentration of DNA using eq. 1 (see Materials and Methods section) to evaluate the enzyme digestion efficiency. The DNA concentration (257 ± 4 mg/L, n = 5) calculated from the quantity of dG was similar to that measured by UV (250 ± 6 mg/L, n = 5), indicating that the enzymatic hydrolysis procedure used can completely hydrolyze DNA into individual nucleosides.
Measurement of decitabine, dG, dC and 5mdC by LC-MS/MS. The mass spectrometer was tuned by infusion of a mixture of decitabine, dC, 5mdC, dG and IS at each concentration of 500 ng/mL in the mobile phase to optimize both compound-dependent and source-dependent parameters. Since these deoxynucleoside were more easily to form protonated ions than deprotonated species in the acidic mobile phase, the positive ESI was used for the analytes' identification and quantification. As shown in Fig. 2A  www.nature.com/scientificreports www.nature.com/scientificreports/ As shown in Fig. 3A, decitabine, dC, 5mdC and dG released by the enzyme digestion along with IS were separated and detected by our LC-MS/MS method in less than 6 min. The control studies showed that there were no contaminations observed from free nucleosides or unincorporated decitabine when DNA was incubated with digestion buffer without enzymes (Fig. 3B). This method has a lower limit of quantification (LLOQ) of 1.00 nM for each analyte, which was defined by the lowest calibrator of a calibration plot; and a linear calibration range from LLOQ to 2.00 × 10 3 nM (Fig. 4). Validation of the method was done according to the US-FDA guidance for industry on bioanalytical method validation 16 . The inter-day and intra-day accuracy and precision of the three mixed QC concentrations were ≤±6% and ≤8%, respectively ( Table 1). The accuracy and precision of mixed calibrators over six calibration plots were ≤±6% and ≤11%, respectively ( Table 2). These data were well within the acceptance ranges of US-FDA.   www.nature.com/scientificreports www.nature.com/scientificreports/ The studies of matrix effect and analyte recovery on all analytes were conducted using mixed QC at three concentrations (Table 3). For all analytes, the absolute matrix factors ranged 0.933-1.03 and the IS normalized matrix factors ranged 0.955-1.03; and the absolute recoveries ranged 90.3-100% and the IS normalized recoveries ranged 96-102%. These data indicated there was neither significant matrix effect from enzyme digestion buffer nor significant difference in recoveries of analytes between enzyme digestion buffer and the mobile phase of LC separation, and the preparation of mixed calibrators of decitabine, dC, dG and 5mdC in either enzyme digestion buffer or the LC mobile phase makes no difference in the analytical results. Therefore, the mixed calibrators of dC, dG and 5mdC prepared in the mobile phase were adopted for the method.

Decitabine incorporation vs. hypomethylation of DNA in leukemia cell lines. The LC-MS/MS
method developed was first tested by human leukemia cell lines HL-60 and U937 for quantitation of decitabine incorporation in cellular DNA and cell responsiveness to decitabine treatment (i.e., decrease of DNA methylation or DNA hypomethylation). In this study, the effects of drug dose (0.000, 0.100, 0.500, 1.00 and 10.0 μM decitabine) and treatment times (24-and 48-h incubation) were investigated. The cellular DNAs were isolated from samples of each dose and time point. The enzyme-released decitabine, dC, dG and 5mdC from the cellular DNA were determined by the method developed and percent methylations were calculated.
As shown in Fig. 5 and summarized in Table 4, the amounts of decitabine incorporated in HL-60 and U937 DNAs were directly proportional to the drug dose and the treatment time. The higher dose and the longer treatment time produced larger amounts of decitabine incorporation in DNA and greater degree of DNA hypomethylation. Also, HL-60 cells have greater viability than U937 cells. This was revealed by the DNA extracted from each of the cell lines after 24-h and 48-h treatment with decitabine. On average, the amounts of DNA produced by HL-60 cells were 1.3 and 2.0 times higher than those of U937 cells after 24-h and 48-h treatment with decitabine. Therefore, HL-60 cells showed more sensitivity to the drug treatment in comparison to U937 with larger 1.60 × 10 3 (HQC) (1.61 ± 0.02) × 10 3 1 1 (1.61 ± 0.02) × 10 3 1 2 Table 2. Intra-and inter-assay precision and accuracy. a Intra-assay precision and accuracy were assessed by five replicate measurements of individual QC at each concentration. b Inter-assay precision and accuracy were assessed by five parallel measurements of five identical QCs at each concentration.
www.nature.com/scientificreports www.nature.com/scientificreports/ decitabine incorporation and greater degree of DNA hypomehtylation, which suggested that for U937 cell line, it may need higher decitabine exposure (either dose or time) than HL-60 cell line to reach the same drug effect.
Decitabine incorporation and hypomethylation effect in a different human leukemia cell line MOLM-13 were also investigated using both a single dose of decitabine (0.   Table 3. Matrix effect and recovery of decitabine from enzyme digestion matrix (n = 5). a MF Ana = (mean peak area of analyte in extracted enzyme digestion matrix)/(mean peak area of analyte in the mobile phase).  www.nature.com/scientificreports www.nature.com/scientificreports/ AML cells for quantitative measurement of DNA-incorporated decitabine and DNA hypomethylation. The experimental details were described in "In vivo animal study" section under "Materials and Methods". As shown in Fig. 7A, decitabine incorporation in bone marrow cells from four genetically identical mice after the drug treatment ranged 18.4 to 25.9 fmol per µg DNA, and such treatment significantly reduced DNA methylation in all mice (P = 0.006, t-test) (Fig. 7B). We had previously reported the efficacy of this treatment in reducing leukemia burden 17,18 . Decitabine incorporation vs. hypomethylation of DNA in clinical study. Finally, the method developed was applied to PBMCs isolated from peripheral blood obtained from five MDS patients before and after six weeks of decitabine treatment by a very low dose (0.2 mg/kg) subcutaneous regimen administered twice per week 2 . The experimental details were described in "Clinical study" section under "Materials and Methods". Among the five patients, four responded well to the decitabine therapy, and one did not. The molecular mechanism of decitabine action and drug resistance in these patients would be depicted by Fig. 8. For patients #1 to #4 who were sensitive to decitabine treatment, decitabine incorporation and DNA hypomethylation were large and significant (P = 0.07, t-test) (Fig. 8A,B); whereas for patient #5 who was resistant to the drug, there were minimum decitabine incorporation (Fig. 8C), and even higher DNA methylation in posttreatment (Fig. 8D).  Table 4. The effects of decitabine dosage and treatment time on decitabine incorporation in DNA and DNA hypomethylation. www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Target engagement is necessary for any drug to work, and the measurement of target engagement of decitabine (i.e., decitabine incorporation and DNA hypomethylation) is the first step to understand patients' sensitivity and resistance in decitabine therapy; however, a practical method for this purpose was lacking. The method we developed is practical, very sensitive and specific for measurement of decitabine, 5mdC, and dG, as no interferences from endogenous molecules were observed. Furthermore, this method has been fully validated in enzyme digestion matrix and essentially met the validation criteria established by US-FDA. Such method may help us not only to understand drug action and resistance mechanisms of decitabine, but also to design more effective dosing regimens for patients.
In the in vitro dose and time course studies, no cell death was observed with the treatment doses and durations used. Decitabine incorporation into DNA of HL-60 and U937 was both dose and time dependent, and notably, uptake via passive nucleoside transporters 1 (ENT1) 19 is drug concentration-dependent. Although decitabine triphosphate incorporation into DNA by DNA polymerase occurs only during S-phase 20 , the drug incorporation in the in vitro studies was around-the-clock because cycling by the cells was unsynchronized. In the time course study of decitabine incorporation in DNA of MOLM-13 cells (Fig. 6), the decitabine incorporation profile deviated from the steady-state profile, likely because of decomposition of decitabine in cell culture medium (decitabine has an in vitro half-life of 5-16 hours at 37 °C) 21 .
Different responses to decitabine treatments in in vitro and in vivo may be a consequence of differences in expression of key pyrimidine metabolism enzymes in different tissues and different models 17,18 . DCK performs the initial phosphorylation of decitabine and rate-limits the conversion of decitabine to decitabine triphosphate. CDA and DCTD metabolize decitabine and, decitabine mono-phosphates into non-DNMT1-depleting uridine derivatives. In the case of the patient with the myeloid malignancy that did not respond to decitabine, decitabine triphosphate incorporation into DNA was about 9% (3 fmol/μg DNA) in comparison to ~10-fold greater incorporation in the patients with disease that did respond to treatment (33 ± 11 fmol/μg DNA); furthermore, in the patient with unresponsive disease, DNA hypomethylation was not observed. Instead DNA methylation increased from 4.22% (pretreatment) to 5.60% (posttreatment) 3,8 .
Previous attempts at correlating DNA hypomethylation in peripheral blood cells with response to decitabine treatment have had mixed results [22][23][24] One difference is that these previous studies administered decitabine by standard pulse-cycled regimens, and DNA methylation changes may therefore have been more substantially influenced by the timing between peripheral blood collection and drug administration. By contrast, the patients analyzed in this study received decitabine twice weekly continuously for 6 weeks prior to the peripheral blood sample collection. A limitation of our study and on our conclusions remains the small number of clinical samples analyzed. www.nature.com/scientificreports www.nature.com/scientificreports/ In sum, quantitative measurement of decitabine incorporation rates and DNA hypomethylation effects can potentially determine early time points where patients with disease are non-responsive to their decitabine dosing regimens, and guide dosage adjustment or selection of alternative therapies. In other words, since the molecular pharmacodynamic intention with decitabine therapy is its incorporation into DNA, it may be useful to prospectively evaluate if the method for measurement of this pharmacodynamic effect that we describe here can guide individualization and optimization of therapy in patients with myeloid malignancies.  www.nature.com/scientificreports www.nature.com/scientificreports/ as control for 0, 4, 8, and 24 h at 37 °C in a humidified 5% CO 2 incubator. After treatment, the cells were removed from the medium by centrifugation at 1,500 × g, 4 °C for 5 min; and washed twice with 5.00 mL of PBS (1X, pH 7.4) each. The cell pellets were collected and stored in −20 °C until DNA extraction.

Chemicals. Decitabine was obtained from Developmental Therapeutics
In vivo animal study. The animal study for this work was approved by the Cleveland Clinic Institutional Animal Care and Use Committees (IACUC) and all methods were performed in accordance with the relevant guidelines and regulations. NSG mice were obtained from Jackson Laboratory (Bar Harbor, ME, USA). Primary acute myelogenous leukemia (AML) cells from patients were transplanted by tail-vein injection (0.4 × 10 6 cells/ mouse) into non-irradiated 6-8 week old NSG mice (n = 4 per/group). Mice were anesthetized with isoflurane before transplantation. On day 15 of inoculation, animal groups were treated subcutaneously with vehicle (PBS 1X, pH 7.4) or decitabine 0.100 mg/kg for three consecutive days per week during the course of study. The control-group animals (median survival of 45 days) were euthanized by an IACUC approved method for signs of distress, and the treatment-group animals were continued with treatment for 90 days. At 90 day of the treatment, animals were euthanized by the IACUC approved method.
DNA was extracted from bone marrow using the procedure published in the ref. 25  DNA isolation and hydrolysis. Each cell pellet collected from the in vitro, in vivo and clinical studies was lysed by mixing gently with 2.00 mL of TE buffer (containing 10.0 mM Tris and 1.00 mM EDTA at pH 8.0) and 240 μL of 10% SDS solution for 2 min. Then, the lysate was incubated with 25.0 μL of proteinase K (600 mAU/ mL) for 1 h at 37 °C. After the incubation, the sample solution was transferred to a phase lock gel tube (5 Prime, Gaithersburg, MD, USA), and DNA was extracted using standard phenol/chloroform extraction method 27 . The DNA extracted was dissolved in 5.00 mM BIS-TRIS buffer (pH 7.0) to a concentration of 1.00 mg/mL measured by UV spectrophotometry. The co-extracted RNA was removed by adding 2.00 µL of RiboShredder ™ RNase Blend (1 U/µL) to each 100 µL of DNA sample (1.00 mg/mL). After a 30-min incubation at 37 °C, the sample was mixed with 1.00 mL of pre-chilled ethanol (−20 °C) and kept at −20 °C for overnight to precipitate the DNA in the sample. After centrifugation at 15,000 × g for 15 min, the supernatant was discarded, and the DNA pellet was washed with 1.00 mL of pre-cooled 70% ethanol (−20 °C) twice. Then, the DNA pellet was air-dried and reconstituted with 100 µL of 5.00 mM BIS-TRIS buffer (pH 7.0) for subsequent enzyme digestion.
DNA hydrolysis was performed by denaturing ds-DNA to ss-DNA in boiling water for 30 min followed by a one-step tetra-enzyme digestion 15,28 . The tetra-enzyme cocktail was prepared by mixing the enzyme solutions as follows: 10.0 μL of DNase I (20,000 U/mL), 15.0 μL of NP1 (200 U/mL), 40.0 μL of PDE I (100 U/mL) and 0.50 μL of ALP (40,000 U/mL). For each 50.0 μL of DNA (0.500 μg/μL for the measurement of DNA incorporated decitabine; 1.25 ng/μL for the measurements of dC, dG, and 5mdC), 4.00 μL of the tetra-enzyme cocktail was added and mixed well; then, the mixture was placed in a 37 °C water bath for overnight incubation to secure a complete digestion of DNA to mononucleosides. After incubation, 5.00 μL of 2′-deoxycytidine 15 N 3 (the internal standard, IS, 2.00 µM in the LC mobile phase) was added to a digested sample and vortex-mixed for 30 s. Each DNA digest was deproteinized by 450 μL of acetonitrile, then centrifuged at 15,000 x g for 10 min. LC-Ms/Ms system. The LC-MS/MS instrumentation used for this work consisted of a Prominence UFLC system (Shimadzu, Columbia, MD, USA) for analyte separation and a QTRAP 5500 tandem mass spectrometer (AB Sciex, Foster City, CA, USA) for quantitation. The system was controlled by AB Sciex Analyst ® software (version 1.6.1) for its operation, data acquisition, and processing. The UFLC system included a system controller (CBM-20A), two binary pumps (LC-20AD), a temperature-controlled autosampler (SIL 20AHT) and an online degasser (DGU20A3).
Mixed calibrators of decitabine, dC, dG and 5mdC (1.00, 4.00, 20.0, 40.0, 200, 400 and 2.00 × 10 3 nM) together with single-and double-blanks, and mixed QCs (3.00, 70.0 and 1.60 × 10 3 nM) were prepared in 5.00 mM BIS-TRIS buffer (pH 7.0) and subjected to the same sample preparation procedure as DNA samples. The linear calibration plots for decitabine, dC, dG, and 5mdC were constructed using peak area ratios (y) of each analyte to the IS versus the concentrations of calibrators (x) with 1/x weighting, and the least squares linear regression equations were obtained as the calibration equations for individual analytes.
The concentrations of decitabine, dC, dG, and 5mdC in each unknown sample were back calculated by the calibration equations using the peak-area ratios of each analyte in the unknown sample to that of the IS. The accurate mass concentration of DNA was determined using the measured molar concentration of dG by the following equation: [DNA] (mg/L) [dG] 618 (g/mol)/0 41 (1) where [dG] is the measured concentration of dG in mM; 618 (g/mol) is the molar mass of G/C pair, and 0.41 is the percentage of G/C pair in human DNA 29 . The amount of DNA incorporated decitabine was expressed as pmol of decitabine per μg of DNA. The percent methylation was calculated using the following equation 29-31 : = + × % methylation [5mdC/(dC 5mdC)] 100% (2) statistical analysis. Statistical analysis was performed using GraphPad Prism 5.0 (GraphPad Software Inc., San Diego, CA).

Conclusion
We have developed a sensitive and specific LC-MS/MS method for quantitative determination of decitabine DNA incorporation and hypomethylation effect both in vitro and in vivo. This method is practical and can be easily adopted in preclinical and clinical studies of decitabine. The preliminary data from decitabine-treated AML patients show that this method not only allows us to quantitatively measure the pharmacokinetic endpoints of decitabine but also provides supporting evidence for drug action mechanisms on patients' sensitivity and resistance in decitabine therapy. It may be useful for determining suitable decitabine dosing regimens for individual patients to achieve desirable therapy outcomes.