Synthesis and Antiproliferative Activities of 5-Azacytidine Analogues in Human Leukemia Cells

Twenty-six 5-azacytidine analogues have been synthesized, including 4-amino-6-alkyl-1-pyranosyl/ribofuranosyl-1,3,5-triazin-2(1H)-ones 1a-j, 6-amino-4-alkyl/aryl-1-pyranosyl/ribofuranosyl-1,3,5-triazin-2(1H)-ones 2a-f and 4-amino-6-alkyl-1,3,5-triazin-2-yl-1-thio-pyranosides/ribofuranosides 3a-j. The antiproliferative activities of these synthetic analogues were investigated in human leukemia HL-60 cells. Ribofuranosyl S-nucleoside 3a, a bioisostere of 5-azacytidine, had a similar antiproliferative ability as that of the latter. Introduction of a methyl at the 6 position of 5-azacytidine and/or replacement of the ribofuranosyl moiety with pyranosyl sugars or disaccharides significantly decreased the antiproliferative activities of the 5-azacytidine derivatives. Several compounds with the replacement of pyranosyl sugars enhanced all-trans retinoic acid-induced differentiation ability in human leukemia HL-60 cells.

Introduction 5-Azacytidine (5-aza-CR) and 5-aza-2'-deoxycytidine (5-aza-CdR, Decitabine) ( Figure 1) are known DNA methyltransferase inhibitors and have been approved for the treatment of myelodysplastic syndrome (MDS) and chronic myelomonocytic leukemia (CMML) [1]. Although both agents are cytotoxic at high concentrations, the therapeutic effects of both compounds in MDS have been thought to be mediated through inhibition of the DNA methyltransferase at low concentrations [1]. Aberrant DNA methylation in the promoter region of genes can silence their expression [2]. Some tumor suppressor genes have been found to be silenced due to DNA hypermethylation and these genes can be reactivated by DNA demethylation [3]. Treatments of malignant cells with 5-aza-CR or 5-aza-CdR have been found to be associated with reversal of specific gene suppression [1,4]. Zebularine ( Figure  1), a derivative of 5-aza-CR with the increased stability, has been reported to have little cytoxicity but to have maintained the ability of inhibiting DNA methytransferase activity [5,6]. Although 5-fluorodeoxycytidine (FDAC) has both cytotoxic effects to malignant cells and inhibitory effects on DNA methyltransferase activity, 5-fluorouracil (5-FU) does not have the inhibitory effects on DNA methyltransferase activity [4]. These observations suggest that the riboside moiety of these nucleoside inhibitors is required for the inhibition of DNA methyltransferase activity [7]. Since the cytotoxic effects of these compounds are due to their incorporation into DNA or RNA [8,9], it seems that replacement of the riboside with other types of sugars will decrease the cytotoxic effects and will keep the abilities of DNA methyltransferase inhibition.
We have synthesized a series of 5-aza-CR derivatives with introduction of a methyl or an ethyl group at the 6 position of 5-aza-CR, and/or with a replacement of O by S, or with a replacement of ribofuranosyl moiety by pyranosyl sugars or disaccharides. Addition of a methyl or an ethyl group at the 6 position of 5-aza-CR could block the attack of water and will improve the chemical stability of 5aza-CR. The replacement of ribofuranosyl moiety with pyranosyl sugars or disaccharides is estimated to reduce cytotoxicity of 5-aza-CR. The antiproliferative activities as well as the differentiation induction of these derivatives alone and in combination with all-trans retinoic acid (ATRA) were investigated in human leukemia HL-60 cells.
The structures of all compounds were determined by application of IR, 1 H-NMR and MS spectral data. The structure assignments for compound 1, 2 and 3 were supported by comparing the 1 H-NMR and HMBC spectral data with each other. The H-1' peak was found to be correlated with C 2 (C=O) and C 6 (C-R 1 ) in HMBC spectrum in N 1 -nucleosides 1. There was a single peak for the two hydrogen signal of the amino group in 1 H-NMR. However, the H-1' peak was correlated with C 2 (C=O) and C 6 (C-NH 2 ) in HMBC spectrum in N 3 -nucleosides 2. There were two single peaks at about 8.31-8.46 and 6.94-7.87 ppm in 1 H-NMR spectra corresponding to the two hydrogen signals of the amino group. This non-equivalence of the amino protons was caused by the pyranosyl sugar, which enhances the rotational barrier of the amino group. The S-nucleosides 3 were identified by HMBC spectra analysis. The H-1' peak was found to be correlated with C 2 (C-S) in HMBC spectrum. The H-1' peak of N 3nucleosides was observed at about 5.85 ppm and these values slightly shifted to low field comparing to that of the N 1 -nucleosides and S-nucleosides (5.35 ppm). The configuration of the nucleosides was deduced from the trans-diaxial coupling between H-1' and H-2' of pyranose in the 1 H-NMR spectra. 1c, 1d, 1j, 2c, 3d, and 3i were α-manopyranosyl or α-rhamnopyranosyl and β-configuration nucleosides.

Bioactivity
The antiproliferative activities of these synthetic compounds were determined in human leukemia HL-60 cells. 5-aza-CR is a potent growth inhibitor with a GI50 value of 0.29 μM. By using a trypan blue exclusion assay, we found that 5-aza-CR killed half of cells at a concentration of 1.0 μM. Among all the listed compounds, only compound 3a has a lower GI50 value of 1.7 μM. Compound 1g has a GI50 value of 18.5 μM. The GI50 values of other compounds can not be obtained since these compounds do not inhibit 50% cell growth at concentrations less than 50 μM. Compound 3a is the bioisostere of 5-aza-CR, with a ribofuranosyl S-nucleoside. Compound 1g only has an introduction of a methyl group at the 6 position of 5-aza-CR without replacement of the ribofuranosyl group. Comparing the structures of 1a-f with 5-aza-CR it was found that replacement of the β-Dribofuranosyl with a β-D-glucopyranosyl, a β-D-xylopyranosyl, a α-D-mannopyranosyl, a α-Lrhamnopyranosyl, a β-D-maltopyranosyl or a β-D-lactopyranosyl significantly decreased the antiproliferative activities. Comparison of the antiproliferative activities of 5-aza-CR with compound 1g revealed that replacement of the H with a CH 3 (1g) slightly decreased the antiproliferative activities. However, replacement of the H (3a) with a CH 3 (3f), where both are ribofuranosyl Snucleosides, evidently decreased the antiproliferative activity. The antiproliferative activity of 5-aza-CR is due to its incorporation into DNA and/or RNA [8], it seems that replacement of the β-Dribofuranosyl in these nucleosides with other sugar moieties would not incorporate into DNA or RNA, that may explain their non-toxic effect to HL-60 cells. It has been shown that ATRA induced differentiation of human leukemia cells with induction of RARβ2 that has been found to be hypermethylated [17]. We have investigated the differentiation activities of these compounds alone and in combination with ATRA in HL-60 cells using NBT reduction assay as a differentiation marker. ATRA treatment at 0.1 μM induced 25.9% of HL-60 cells undergoing differentiation after 5 days. All of these compounds alone did not induce NBT reduction but some compounds such as 1c, 1d and 3i enhanced ATRA-induced differentiation. The enhancement on ATRA differentiation induction by these compounds may be due to the enhanced RARβ2 expression through inhibition of DNA methyltransferase that needs to be further investigated. Since non-nucleoside compounds such as (-)epigallocatechin-3-gallate, hydralazine and procainamide inhibit DNA methyltransferase activity [18,19], we prospect that some of these synthetic compounds will keep the ability of inhibiting DNA methyltransferase activity without incorporating into DNA. The inhibitory abilities of these compounds on DNA methyltransferase activity are under investigation.   In summary, our data indicate that 1) changing 5-aza-CR into a bioisosteric ribofuranosyl Snucleoside (3a) does not influence the antiproliferative ability; 2) replacement of the ribofuranosyl moiety with pyranosyl sugars or disaccharides significantly decreases the antiproliferative efficacies; 3) compounds with the ribofuranosyl moiety replaced by the pyranosyl sugars (1c, 1d and 3i) enhance ATRA differentiation induction ability.

General
The melting points were determined on an electrically heated X4 digital visual melting point apparatus and were uncorrected.. IR spectra were recorded on Bruker IFS 55 (KBr). 1 H-NMR spectra were recorded on a Bruker ARX-300 spectrometer at 300MHz with DMSO-d 6 as solvent and TMS as an internal standard. Mass spectra were taken in ESI mode on Agilent 1100 LC-MS. Elemental analysis was determined on a Carlo-Erba 1106 Elemental analysis instrument. All solvents and reagents were commercially available.

Cell growth inhibition
Cells were seeded at 5 × 10 4 cells/mL and incubated with various concentrations of the indicated agents for three days. The total cell number in each group was determined with the aid of a hemocytometer and the drug concentration that inhibits half of cell growth (GI50) was calculated. The cell viability was estimated by trypan-blue exclusion assay.

NBT reduction assay
The nitroblue tetrazolium (NBT) reduction assay as a determination of cell differentiation was performed as reported previously [21] and the cells were treated with 0.15 μM 5-aza-CR, 0.85 μM 3a, 10 μM 1g or 50 μM each of other compounds alone and in combination with 0.1 μM ATRA for 5 days. Total 300 cells were counted and percentages of NBT positive cells were calculated.