The major goal of epigenetic therapy is to reverse aberrant promoter hypermethylation and restore normal function of tumor suppressor genes by the use of chromatin-modifying drugs. Decitabine, or 5-aza-2′-deoxycytidine (5-aza-CdR), is a well-characterized drug that is now Food and Drug Administration approved for the treatment of myelodysplastic syndrome. Although 5-aza-CdR is an extremely potent inhibitor of DNA methylation, it is subject to degradation by hydrolytic cleavage and deamination by cytidine deaminase. We show that short oligonucleotides containing a 5-aza-CdR can also inhibit DNA methylation in cancer cells at concentrations comparable with 5-aza-CdR. Detailed studies with S110, a dinucleotide, showed that it works via a mechanism similar to that of 5-aza-CdR after incorporation of its aza-moiety into DNA. Stability of the triazine ring in aqueous solution was not improved in the S110 dinucleotide; however, deamination by cytidine deaminase was dramatically decreased. This is the first demonstration of the use of short oligonucleotides to provide effective delivery and cellular uptake of a nucleotide drug and protection from enzymatic degradation. This approach may pave the way for more stable and potent inhibitors of DNA methylation as well as provide means for improving existing therapeutics. [Cancer Res 2007;67(13):6400–8]
Aberrations in DNA methylation are frequently observed in various types of cancer (1–3). Several tumor suppressor genes or cancer-related genes acquire de novo DNA methylation in promoter or regulatory regions leading to inactivation, contributing to tumorigenesis (4–8). DNA hypermethylation can be reversed by demethylating agents and crucial cellular functions reestablished in the cells. In recent years, the use of DNA methylation inhibitors has become a promising alternative to patients with myelodysplastic syndrome and hematologic malignancies (9, 10).
The most widely known examples of DNA methylation inhibitors are 5-azacytidine (5-aza-CR) and 5-aza-2′-deoxycytidine (5-aza-CdR); both drugs were initially synthesized as anticancer agents and were later shown to inhibit DNA methylation (11, 12). The clinical use of nucleotides rather than nucleosides is essentially impossible due to the negative charge on the phosphate group that prevents effective cellular uptake. Thus, nucleoside analogues, which are taken up intracellularly and phosphorylated to their respective monophosphates, diphosphates, and triphosphates before incorporation into replicating DNA, leading to covalent trapping of DNA methyltransferases (DNMT), are used (13). 5-Aza-CR and 5-aza-CdR are powerful demethylating agents; nevertheless, they have a number of drawbacks. The aza pyrimidine ring is unstable in aqueous solution, making it difficult to administer, and is quite toxic both in vitro and in vivo (14). Furthermore, the drugs have transient effects, and DNA is gradually remethylated after removal of the drug (15). Yet, another problem arises due to cytidine deaminase, which renders the drugs inactive by converting them into 5-azauridine compounds.
In our attempts to synthesize more stable and potent inhibitors of DNA methylation, we found that short oligonucleotides containing an azapyrimidine effectively inhibit DNA methylation in living cells. Here, we focus on S110, a 5′-AzapG-3′ dinucleotide, whose aqueous stability and toxicity are quite similar to that of 5-aza-CdR but is protected from deamination by cytidine deaminase. The demethylating activity seems to require incorporation of the azapyrimidine into DNA presumably after degradation of the oligonucleotide by phosphodiesterases and is not limited to a dinucleotide but is seen in trinucleotides and tetranucleotides as well, showing that short oligonucleotides are effective prodrugs for delivery of inhibitors of DNA methylation. The utilization of short oligonucleotides as nucleoside drug delivery vehicles that provides protection against enzymatic degradation might have application for delivery of other nucleoside drugs to cells.
Materials and Methods
Synthesis of oligonucleotides containing 5-aza-CdR. Dinucleotides, trinucleotides, and tetranucleotides containing 5-aza-CdR were synthesized by standard procedures with modifications to increase coupling times, different oxidizing agents, and use of phenoxyacetyl decitabine phosphoramidite, instead of phenoxyacetyl cytidine phosphoramidite. A polystyrene-based solid support with loading of 240 μmol/g dG(pac) or D(pac) was used (16, 17).
Briefly, synthesis of S53, 5′-GpAza-3′ dinucleotide, is described here. Amersham ÄKTA Oligopilot 10 system was loaded with a protected decitabine-linked CpG solid support (phenoxyacetyl protection of amino function) and coupled with 2 to 2.5 equivalents of tert-butyl phenoxyacetyl 2′-deoxyguanosine phosphoramidite in the presence of 60% of 0.3 mol/L benzylthiotetrazole activator in acetonitrile for 2.5 min. The CpG solid support containing protected S53 was treated with 20 mL of 50 mmol/L K2CO3 in methanol for 1 h and 20 min. The coupled product was oxidized with 2 mol/L tert-butylhydroperoxide in dry acetonitrile prepared by dissolving tert-butylhydroperoxide in 80% tert-butylperoxide for 5 min. The dimethoxy trityl protective group was removed with 3% dichloroacetic acid in toluene. The CpG solid support was washed with dry methanol; the filtrate was neutralized by addition of 2 mL of 1 mol/L acetic acid in methanol. The solution was concentrated by rotary evaporation; the residue was taken up in 200 mmol/L triethylammonium acetate (pH 6.9), washed with 500 μL 50% aqueous acetonitrile, and filtered through a syringe filter. S53 was subsequently purified to 95% purity by the ÄKTA Explorer 100 high-performance liquid chromatography (HPLC) with a Gemini C18 preparative column (Phenomenex), 250 × 21.2 mm, 10 μm with guard column (Phenomenex), 50 × 21.2 mm, 10 μm, with 50 mmol/L triethylammonium acetate (pH 7) in MilliQ water (Mobile Phase A) and 80% acetonitrile in MilliQ water (Mobile Phase B), with 2% to 20%/25% Mobile Phase B in column volumes. The electrospray-mass spectrometry (+) of triethylammonium salt of S53 exhibited m/z 556.1 [M-H]− and 1,113.1 for [2M-H]−, and the calculated exact mass for the neutral compound C18H24N9O10P is 557.14.
Cytidine deaminase assay. Recombinant human cytidine deaminase was kindly provided by Alberto Vita (University of Camerino, Italy) and prepared as described in Vincenzetti et al. (18). Purified cytidine deaminase (0.09–0.1 unit) was incubated with 0.2 mmol/L 5-aza-CdR or S110 in 3 mL of 0.1 mol/L Tris-HCl (pH 7.5) at 38°C. Percent substrate remaining was measured after 45 and 90 min of incubation by a Waters 2695 HPLC system with a 996 photodiode array detector (Waters Corp.).
Determination of nucleic acid stability. The absorbance of 5-aza-CdR and S110 was measured with a Beckman DU Series 600 UV/visible Spectrophotometer (Beckman Coulter, Inc.). The wavelengths of maximum absorbance of 5-aza-CdR and S110 were 241 and 245 nm, respectively. The initial absorbance was measured immediately after 5-aza-CdR and S110 were dissolved in PBS, and both compounds were incubated at 37°C except when absorbance reading was taken.
Cell lines and drug treatment. T24 bladder carcinoma cells and HCT116 colon carcinoma cells were obtained from the American Type Culture Collection and cultured in McCoy's 5A medium supplemented with 10% heat-inactivated FCS, 100 units/mL penicillin, and 100 μg/mL streptomycin (Invitrogen) in a humidified incubator at 37°C in 5% CO2. Cells were plated (1.5 × 105 per 60-mm dish) and treated 24 h later with 5-aza-CdR (Sigma-Aldrich) or S110 continuously for 6 days. Each compound was dissolved in PBS. The medium was changed 3 days after the initial treatment and supplemented with a fresh dose of 5-aza-CdR or S110.
Determination of cytotoxicity. A colony formation assay was used to compare cytotoxicity of 5-aza-CdR and S110 as previously described (19). Briefly, T24 cells were plated at a low density (100 per 60-mm dish) and treated with varying concentrations of 5-aza-CdR and S110. Colonies were allowed to form for 10 to 14 days, fixed with methanol, and stained with 10% Giemsa. The number of colonies from an untreated control plate was used to calculate the plating efficiency in percent at each concentration. Triplicate dishes were used, and error bars are represented by 1 SD of the mean.
Nucleic acid isolation. RNA was collected from T24 and HCT116 cells with the RNeasy Mini kit (Qiagen) according to the manufacturer's protocol. DNA was collected using the DNeasy Tissue kit (Qiagen) according to the manufacturer's protocol.
Quantitative reverse transcription-PCR analysis. Total RNA (5 μg) was reverse transcribed with Moloney murine leukemia virus reverse transcriptase (Invitrogen) and random primers (Invitrogen). The reverse transcription was carried out in a total volume of 50 μL as previously described (4). The quantitation of mRNA levels was carried out by a real-time fluorescence detection method as described previously (15, 20). All samples were normalized to the reference gene GAPDH. The primer and probe sequences are as follows: for p16, sense 5′-CTGCCCAACGCACCGA-3′, probe 5′-6-FAM-TGGATCGGCCTCCGACCGTAACT-BHQ-1 3′, and antisense 5′-CGCTGCCCATCATCATGAC-3′; for GAPDH, sense 5′-TGAAGGTCGGAGTCAACGG-3′, probe 5′-6-FAM-TTTGGTCGTATTGGGCGCCTGG-BHQ-1 3′, and antisense 5′-AGAGTTAAAAGCAGCCCTGGTG-3′. The conditions for real-time reverse transcription-PCR are 94°C for 9 min followed by 45 cycles at 94°C for 15 s and 60°C for 1 min.
Western blot analysis. Cell pellets were lysed in radioimmunoprecipitation buffer containing 0.1% SDS, 0.5% NP40, and 0.5% sodium deoxycholate in PBS and incubated on ice for 30 min. The lysates were centrifuged at 4°C for 30 min at 14,000 rpm. The supernatant was collected and stored at −80°C. Approximately 60 μg of protein was electrophoresed on a Ready Gel Tris-HCl Gel, 4% to 15% linear gradient (Bio-Rad Laboratories) and transferred to a polyvinylidene difluoride membrane using Trans-Blot SD Semi-Dry Electrophoretic Transfer Cell (Bio-Rad Laboratories). The membrane was hybridized with antibodies against human DNMT1 (H-300; 1:500; Santa Cruz Biotechnology), p16 (N-20; 1:500; Santa Cruz Biotechnology), and proliferating cell nuclear antigen (PCNA; PC10; 1:500; Santa Cruz Biotechnology) in TBS-Tween (TBS-T) buffer (0.01 mol/L Tris, 0.15 mol/L NaCl, and 0.1% Tween 20) with 5% nonfat dry milk overnight at 4°C. The membranes were washed four times with TBS-T buffer at room temperature and incubated with secondary antibodies for 1 h at room temperature. Secondary antibodies used were anti-rabbit-IgG-horseradish peroxidase (HRP) for DNMT1 and p16 (1:7,500; Santa Cruz Biotechnology) and anti-mouse-IgG-HRP for PCNA (1:7,500; Santa Cruz Biotechnology). The membranes were washed six times with TBS-T at room temperature. Proteins were detected with the ECL Western Blotting Detection Reagents (GE Healthcare Biosciences Corp.) and by exposure to Kodak X-OMAT AR film.
Quantitation of DNA methylation. Genomic DNA (4 μg) was treated with sodium bisulfite as previously described (19). Methylation analysis was done using the methylation-sensitive single-nucleotide primer extension (Ms-SNuPE) assay for p16 5′-region as previously described (21). The PCR primers used were sense 5′-TTTGAGGGATAGGGT-3′ and antisense 5′-TCTAATAACCAACCAACCCCTCC-3′. An initial denaturation at 94°C for 3 min was followed by 94°C for 45 s, 62°C for 45 s, and 72°C for 45 s for 40 cycles. Primers used for Ms-SNuPE analysis were 5′-TTTTTTTGTTTGGAAAGATAT-3′, 5′-TTTTAGGGGTGTTATATT-3′, and 5′-GTAGAGTTTAGTT-3′. Conditions for primer extension were 94°C for 1 min, 50°C for 30 s, and 72°C for 20 s. Methylation analysis of X58075 LINE-1 element was carried out with the following primer set. The PCR primers were sense 5′-TTTTTTGAGTTAGGTGTGGG-3′ and antisense 5′-CATCTCACTAAAAAATACCAAACAA-3′, and the conditions were 94°C for 3 min followed by 94°C for 45 s, 52°C for 45 s, and 72°C for 45 s for 39 cycles. Primers for Ms-SNuPE analysis were 5′-GGGTGGGAGTGATT-3′, 5′-GAAAGGGAATTTTTTGATTTTTTG-3′, and 5′-TTTTTTAGGTGAGGTAATGTTT-3′. Conditions for primer extension were 94°C for 1 min, 50°C for 30 s, and 72°C for 20 s.
In vitro hemimethylation assay. An intronic region of p16 was amplified using human genomic DNA as a substrate and a primer set with a methylated CpG to yield a hemimethylated CpG site. The amplicon (1 ng) was treated with M.Sss1 or human DNMT1 (New England Biolab) in 50 μL of reaction buffers, as recommended by the manufacturer, for 30 min at 37°C, in the presence of S110 of various concentrations. The reaction was stopped by heat inactivation of enzymes. Bisulfite-converted DNA was amplified by 5′-CTCTTACCATCCTCTT-3′ and 5′-GAGTTATATTTATGTGATTATTTT-3′, and Ms-SNuPE assay was done using 5′-TTTTAAAATTTTGTTAATAGTTTGAATT-3′ as a primer.
Sucrose density gradient ultracentrifugation. Nuclei were prepared according to the procedure described in Gal-Yam et al. (22). Briefly, cells were trypsinized and washed once with PBS. The cells were then resuspended in ice-cold RSB buffer [10 mmol/L Tris-HCl (pH 7.4), 10 mmol/L NaCl, 3 mmol/L MgCl2] containing Complete Mini proteinase inhibitors (Roche) and kept on ice for 10 min before dounce homogenization with 0.5% NP40 to break up cell membranes. Nuclei were washed twice with RSB plus proteinase inhibitors without the detergent. Purified nuclei (1 × 108) were resuspended in 1 mL of RSB containing 0.25 mol/L sucrose, 3 mmol/L CaCl2, and 100 μmol/L phenylmethylsulfonyl fluoride and digested with MNase (Worthington Biochemical Corp.) for 15 min at 37°C, and then the reaction was stopped with EDTA/EGTA (up to 10 mmol/L). After microcentrifugation at 5,000 rpm for 5 min, the nuclear pellet was resuspended in 0.3 mL of the buffer [10 mmol/L Tris-HCl (pH 7.4), 10 mmol/L NaCl] containing 5 mmol/L EDTA/EGTA, gently rocked for 1 h at 4°C, and followed by microcentrifugation to obtain soluble nucleosomes, which were then fractionated through a sucrose density gradient solution [5-25% sucrose, 10 mmol/L Tris-HCl (pH 7.4), 0.25 mmol/L EDTA, 300 mmol/L NaCl] at 30,000 rpm for 16 h at 4°C. Fractions were taken from the top of the centrifuge tube into 16 aliquots and subjected to Western blot analysis.
Reexpression of p16 in T24 bladder cancer cells by short oligonucleotides. We first treated T24 cells with varying concentrations of oligonucleotides containing the 5-azapyrimidine ring for 6 days to test their abilities to induce the expression of the methylation-silenced p16 gene in T24 bladder carcinoma cells (Table 1). Induction of p16 is an indirect, yet straightforward, method to detect inhibition of DNA methylation because demethylation of the 5′ CpG island of p16 results in reexpression of the gene (4, 23). All short-chain nucleotides tested, regardless of their chain lengths, were able to induce robust p16 expression with the exception of S52R (Fig. 1B) and S112 (Fig. 1E), which induced expression to lesser extents (Table 1). S52R, a phosphorothioate derivative of S110 (Fig. 1B), is less prone to cleavage by phosphodiesterases and caused a small induction of p16 at high concentration, suggesting that the cleavage of the oligonucleotides into nucleotides and nucleosides is a determining factor in inducing gene expression. S112 is an S110 derivative with a hexaethylene glycol phosphate linker at the 5′-end (Fig. 1E), which is subject to cleavage in cells. This extra cleavage requirement probably slows down the ability of S112 to induce the p16 gene and thus explains the weaker activity of the compound. Demethylation of the 5′-end of the p16 gene was observed in cells treated with the short oligonucleotides, showing that indeed, these compounds are bona fide inhibitors of DNA methylation (data not shown). From here on, we focused our detailed characterization on S110 (Fig. 1D) as an example of short oligonucleotide demethylating agents.
|Compound name .||Structure .||Dose (μmol/L) .||p16 induction .|
|Compound name .||Structure .||Dose (μmol/L) .||p16 induction .|
NOTE: List of short oligonucleotides that inhibit DNA methylation and induce p16 expression in T24 cells after 6 d of continuous treatment. Dose indicates the lowest concentration at which the induction of p16 expression is observed.
Abbreviations: Aza, 5-aza-2′-deoxycytidine.; HEG, hexaethylene glycol phosphate linker.
Effects of S110 on DNA methylation and p16 gene expression in T24 and HCT116 cells. We extended on the initial screen done in Table 1 to determine the dose dependency of S110 using T24 bladder and HCT116 colon cancer cells. First, we assessed global methylation status by determining the methylation level of long interspersed nucleotide element-1 (LINE-1) sequences. Repetitive DNA elements, such as LINE-1 retrotransposable elements, serve as a useful marker of genome-wide methylation changes and have previously been shown to be demethylated upon treatment with 5-aza-CdR (24). In both T24 and HCT116 cells, the decrease in the level of methylation was dose dependent and comparable for 5-aza-CdR and S110 after 0.1 and 1 μmol/L treatment (Fig. 2A , top). At 10 μmol/L concentrations, only a small decrease in methylation was noted, probably due to side effects of high drug concentrations as we observed previously (12). In fact, 10 μmol/L treatment may be too cytotoxic for effective demethylation to take place as the plating efficiency of T24 cells indicates. It is well established that its cytotoxic dose is not ideal for optimal epigenetic therapy because these drugs inhibit DNA methylation best at low doses in cell lines as well as in the clinic (25, 26).
We then analyzed the DNA methylation status of exon 1 of the p16 gene in these cells after 6-day treatment by quantitative Ms-SNuPE. In untreated T24 cells, the three CpG sites under analysis are located in the first exon of the gene and are almost always fully methylated. The methylation level decreased with increasing concentrations of 5-aza-CdR or S110 in T24 cells (Fig. 2A,, middle left) and HCT116 cells (Fig. 2A , middle right) with the greatest demethylation seen at 1 μmol/L 5-aza-CdR.
Next, we measured the expression of p16 in both cancer cell lines. Untreated T24 bladder carcinoma cells do not express p16 and dose-dependent increases in p16 expression were observed after 6 days of continuous treatment with 5-aza-CdR or S110 (Fig. 2A,, bottom left). After HCT116 colorectal carcinoma cells were treated for 6 days, a dose-dependent increase in p16 expression was observed with S110, whereas the highest p16 expression was seen at 1 μmol/L dose of 5-aza-CdR (Fig. 2A,, bottom right). In addition, T24 and HCT116 cells treated with either agent for 3 days also showed a dose-dependent increase in the level of p16 protein (Fig. 2B), showing the competence of S110 to inhibit DNA methylation and induce p16 at both mRNA and protein levels as well as 5-aza-CdR. Thus, S110 is able to inhibit DNA methylation at 5′-region and induce the expression of the p16 gene in T24 and HCT116 cells at concentrations comparable to 5-aza-CdR, and the induction of p16 expression by both agents correlated with the demethylation at the 5′-end region of the gene in both cell lines.
Depletion of DNMT1 levels by S110. To provide more clues to the mechanism of action of S110, DNMT1 protein levels in T24 and HCT116 cells treated with the compound were analyzed by Western blot analysis (Fig. 2B). A number of studies have provided evidence that incorporation of the azapyrimidine ring into DNA is necessary for the drug to inhibit DNA methylation (13, 27, 28). Work by Hurd et al. (29) showed a covalent bond formation between a DNMT and zebularine-incorporated oligodeoxynucleotide, strengthening the notion that DNA incorporation is an essential step toward inhibition of methylation by mechanism-based inhibitors. It is probable then that S110 works via a similar mechanism and causes a depletion of extractable DNMT1 in cells. Indeed, partial and complete depletion of DNMT1 protein was observed in T24 and HCT116 cells treated with 5-aza-CdR for 3 days (Fig. 2B). Similarly, in S110-treated T24 and HCT116 cells, partial depletion of the enzyme was seen at the 1 μmol/L dose, and a trace of DNMT1 remained after 10 μmol/L treatment. Our results suggest that S110 causes depletion of extractable DNMT1 in cells and may work by a similar mechanism as that of 5-aza-CdR in inhibiting DNA methylation.
Inhibition of DNMT1 in vitro by S110 and fractionation of DNMT1 in nuclear extracts using sucrose density gradient ultracentrifugation. To validate the notion that S110 works via DNA incorporation after phosphodiesterase cleavage, we next determined whether the dinucleotide could inhibit DNMTs in a cell-free hemimethylation assay. We tested the ability of S110 to inhibit a bacterial methylase, M.Sss1, or human DNMT1 activity to methylate a double-stranded DNA containing a hemimethylated CpG site. Figure 3A shows that both M.Sss1 and DNMT1 were able to methylate DNA even in the presence of high concentrations of S110. This result provided important clues to the mechanism of S110 by showing that the dinucleotide cannot inhibit DNMT1 directly.
Although Fig. 3A showed that the dinucleotide cannot directly inhibit DNA methylation, we still lacked direct evidence that S110 metabolite had to be incorporated into DNA after cleavage. We therefore extracted nuclei from 5-aza-CdR or S110-treated HCT116 cells, collected soluble nucleosomes, and size fractionated the nucleosomes by sucrose density gradient centrifugation to determine whether the drugs would trap DNMT1 on chromatin (Fig. 3B and C). The absorbance showed that DNA is concentrated in fractions 5 to 16 (Fig. 3B) as are the nucleosomes (Fig. 3C). Under the salt concentrations used (300 mmol/L NaCl), DNMT1 was not associated with nucleosomes, and the majority of DNMT1 protein in untreated HCT116 cells was not bound to chromatin (Fig. 3C). After 5-aza-CdR or S110 treatment, the distribution of DNMT1 changed dramatically, and although the majority of the enzyme was present in the low molecular portion of the gradient (fractions 4 and 5), it was also found in fractions 6 to 16 along with high molecular weight DNA (Fig. 3C). We interpret this result to show that the incorporation of 5-aza-CdR or the hydrolyzed product of S110 caused the DNMT1 enzyme to bind very strongly to the DNA in chromatin. Our experiment strongly suggests that S110 works through a similar mechanism as 5-aza-CdR after its cleavage by phosphodiesterases, and that both 5-aza-CdR and S110 byproducts are incorporated into DNA where they covalently interact with DNMTs. It is most probable that the S110 dinucleotide is cleaved into nucleotides and nucleosides because incorporation of a dinucleotide or oligonucleotide is not known to occur during DNA synthesis.
Reverse dinucleotide and phosphorothioate analogues of S110. To gain further insight as to how the S110 dinucleotide inhibits DNA methylation, we used S110 analogues (i.e., S53), which is a reverse dinucleotide of S110, and S52R and S52S, which are two optical isomers of phosphorothioate analogues of S110 (Fig. 1B and C). S53 would allow us to determine whether the sequence of the dinucleotide plays a role in inhibiting DNA methylation. In addition, the phosphorothioate analogues would help us determine whether a free 5-aza-CdR is released from the dinucleotide, or if S110 directly inhibits DNA methylation. The phosphate backbones of phosphorothioate analogues S52R and S52S are not readily cleaved, hence retaining the dinucleotide structure for longer period of time. T24 cells were treated continuously for 6 days with these compounds at varying concentrations. A dose-dependent induction of p16 and demethylation of p16 exon 1 were seen in T24 cells treated with S53, whereas the expression and the DNA methylation status at the 5′ end region of p16 remained unaffected in cells treated with S52R and S52S (Fig. 4A and B). Overall, this suggests that the order of bases in the dinucleotide does not play a crucial role in inhibiting DNA methylation, but the cleavage of the dinucleotide is critical in rendering the compound active, suggesting once again that the dinucleotide does not work as a whole and has to be cleaved into individual nucleotides and nucleosides.
Stability of 5-aza-CdR and S110 at 37°C in neutral aqueous solution. A drawback of 5-aza-CdR is its short half-life in aqueous solution due to rapid hydrolysis of the azapyrimidine. To compare the stability of S110 with that of 5-aza-CdR, we measured the absorbance of 5-aza-CdR and S110 at their respective wavelengths of maximum absorbance (λmax) over time at 37°C in PBS solution at pH 7.4. The absorbance of 5-aza-CdR showed a sharp decay in the first 30 h of incubation with a half-life of about 20 h (Fig. 5A). The half-life of S110 was 21 h, with 65% of its initial absorbance remaining for as long as 200 h (Fig. 5A). This residual absorbance at 245 nm is attributed to the guanosine ring in the dinucleotide, the absorbance of which did not decrease during the course of the experiment. In addition, the stability of both drugs was comparable when monitored for 90 min at 38°C (pH 7.5) by HPLC (see Fig. 5B). Furthermore, additional HPLC analysis of 5-aza-CdR and S110 stability at room temperature over time showed that both compounds have similar half-lives (data not shown). Therefore, whether the 5-azacytosine ring was in a single nucleotide or a dinucleotide, the stability of the compound was not affected, and the half-lives of these two compounds in aqueous solution at 37°C remained the same.
Enzymatic degradation of 5-aza-CdR and S110 by human cytidine deaminase. 5-aza-CdR is quickly catabolized into 5-azadeoxyuridine by cytidine deaminase in vivo (30). We incubated either 5-aza-CdR or S110 with human recombinant cytidine deaminase at 38°C and measured the amount of substrate remaining over time by HPLC (Fig. 5B). A slow decrease in the total amount of 5-aza-CdR in the absence of cytidine deaminase (▴) over time is indicative of hydrolysis of the 5-azapyrimidine ring in water. A rapid decrease in 5-aza-CdR was observed in the first 45 min of incubation with human cytidine deaminase, and only about 2% of the total substrate remained after 90 min (Fig. 5B,, •). In contrast, S110 was subject to hydrolytic cleavage as found earlier (Fig. 5A) but was resistant to enzymatic degradation by cytidine deaminase; the majority of S110 remained in the solution after 90 min of incubation with cytidine deaminase (Fig. 5B , ♦). Incubation of S110 in human plasma showed similar results, and the dinucleotide persisted for 40 min with very little decrease when measured by HPLC compared with a 33% decrease in the level of 5-aza-CdR (data not shown). Although the dinucleotide structure cannot prevent the hydrolysis of the 5-azacytosine ring, it may greatly lengthen the half-life of the drug by protecting it from deamination, potentially increasing the efficacy of the drug in patients as a result.
Cytotoxicity of 5-aza-CdR and S110 in T24 cells. Finally, we compared the cytotoxicities of 5-aza-CdR and S110 in T24 bladder carcinoma cells by measuring effects of the compounds on the plating efficiency of T24 cells. 5-Aza-CdR decreased the plating efficiency in a dose-dependent manner, with no colonies forming at 10 μmol/L concentration (Fig. 5C). S110 was slightly less toxic than 5-aza-CdR at the doses tested up to 1 μmol/L concentration but displaying similar toxicity at 10 μmol/L concentration. Similar results were obtained for HCT116 cells (data not shown). Thus, the cytotoxicity of S110 as measured by plating efficiency is quite similar to 5-aza-CdR in T24 cells.
Short oligonucleotides were synthesized to test whether the stability of 5-aza-CdR, a cancer chemotherapeutic agent, could be improved without compromising its potency to inhibit DNA methylation. Comparison of the S110 dinucleotide containing 5-aza-CdR showed that S110 works via a similar mechanism of action to the parent compound to inhibit DNA methylation, induce expression of the p16 tumor suppressor gene, and inhibit tumor cell growth (Figs. 2–5). Characterization of S110 showed that the stability in aqueous solution and cytotoxicity is still comparable with that of 5-aza-CdR; however, a major improvement was accomplished in terms of resistance to enzymatic deamination (Fig. 5B). Deamination of 5-aza-CdR by cytidine deaminase rapidly depletes the plasma level of the drug, resulting in low bioavailability that has been frequently pointed out as one of the major drawbacks of 5-aza-CdR (30). S110 dinucleotide is resistant to cytidine deaminase deamination probably due to the substrate specificity of the enzyme, which may potentially increase the half-life of the drug, enhance bioavailability, and make the drug more efficacious. Lowering the dose requirement may also reduce the toxicity in patients as well as diminishing common side effects of 5-aza-CdR, such as myelosuppression.
Trinucleotides and tetranucleotides containing one or more 5-aza-CdR residues in the chain showed demethylating activity as well, showing that the cleavage of these oligonucleotides is extremely efficient (Table 1; data not shown). Additionally, the hydrolytic cleavage of the 5-azacytosine ring may presumably be improved in the longer-chain oligonucleotides, further improving stability. In practice, this may be very advantageous in terms of broadening the applications of epigenetic therapy to patients with solid tumors.
Oligonucleotides have been widely studied as chemotherapeutics, and in most cases, the base sequence plays a pivotal role in giving them therapeutic effects. For example, oligonucleotides with CpG motifs activate toll-like receptors and elicit immune responses (31). Antisense morpholino oligonucleotides and small interfering RNAs all use sequence specificities of endogenous molecules to inhibit either transcription or translation of target genes (32, 33). Unlike the existing approaches, the short oligonucleotides we have developed do not require any sequence specificities and simply provide protection against enzymatic degradation before the DNA incorporation of the active moiety following endonuclease and/or phosphodiesterase cleavage. Our findings suggest that once these prodrugs are delivered to the target site and taken up intracellularly, they are cleaved and metabolized to induce effects of similar levels to the parent molecule. This concept eliminates the need for chemical modifications such as lipid esterification and complex delivery vehicles that are commonly required for prodrugs.
Although it is most likely that these oligonucleotides are cleaved and incorporated into DNA to inhibit DNA methylation, it is still not clear as to whether the cleavage occurs extracellularly or intracellularly. It is not likely that the oligonucleotides are processed into 5-aza-CdR or its nucleotide by DNases in plasma before entering the cell because there is a little or weak exonuclease activity in plasma, and endonucleases are known to fragment DNA into larger pieces, thereby protecting the short oligonucleotides from degradation (34–36). Furthermore, CpG oligodeoxynucleotides must retain the CpG motifs to generate immune response, which provides evidence that short oligonucleotides are not subject to nuclease degradation extracellularly (37). It is probable then that the short oligonucleotide DNA methylation inhibitors retain their structure when they enter the cell and are only cleaved in the cells.
Oligonucleotides are thought to internalize themselves into cells through both passive and active mechanisms (31). Much effort has been devoted to expedite such uptake with liposomal encapsulation and peptide-mediated delivery of these oligonucleotides (38, 39). Several membrane-bound proteins and receptors were found to interact with oligonucleotides and facilitate their cellular uptake (40–43). There are also speculations about cell surface channel proteins that conduct nucleic acids and oligonucleotides (44). Other transmembrane proteins are thought to cleave phosphodiester and phosphosulfate bonds at the cell surface (45). There are therefore many possible mechanisms by which short oligonucleotides may enter the cell.
Once the oligonucleotides are internalized, they are probably cleaved into individual nucleotides and nucleosides. The most likely candidates for the cleavage of oligonucleotides are the 11 distinct subfamilies of phosphodiesterases that are abundant in all cell types (46). Phosphodiesterases catalyze the hydrolysis of 3′-ester bonds of cyclic AMP and cyclic GMP into their respective monophosphate moeities, leaving a phosphate group at the 5′-end (47, 48). It is probable then that these enzymes, when they process the oligonucleotides, also leave the phosphate group on the 5′-end of the nucleoside in the sequence, hence yielding both nucleotides and nucleosides. However, it is not clear from preliminary data whether the phosphodiesterases are the main player involved in processing these inhibitors, and further work is necessary to assess their role in oligonucleotide metabolism.
Regardless of the mechanism, our data show that these short oligonucleotides are quite efficient in entering the cell and inhibiting DNA methylation. Dinucleotides, trinucleotides, or tetranucleotides containing one or more 5-aza-CdR residues are as effective as 5-aza-CdR in inducing p16 expression and decreasing global methylation levels. These short-chain inhibitors may be more resistant to cytidine deaminase degradation and be particularly useful for clinical utility. We have therefore introduced a novel drug delivery system in which the oligonucleotides are not restricted to their sequences for therapeutic potential and provide protection against enzymes involved in the drug metabolism. This may find numerous applications in delivering 5′-phosphates to cells. In addition, other nucleoside drugs that are subject to enzymatic degradation may benefit by using this delivery method. The true merits of these short oligonucleotide DNA methylation inhibitors may only be realized in a clinical setting.
Grant support: NIH grant R01 CA82422.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.