Current Topics : Reviews
Synthesis of 4′-Thionucleosides as Antitumor and Antiviral Agents
2018 Volume 66 Issue 2 Pages 139-146
Details
2018 Volume 66 Issue 2 Pages 139-146
Many attempts have been made to synthesize structurally novel nucleoside derivatives in order to identify effective compounds for the treatment of tumors and virus-caused disease. At our laboratories, as part of our efforts to synthesize 4′-thionucleosides, we have identified and characterized biologically active nucleosides. During the course of our synthetic study, we developed the Pummerer-type thioglycosylation reaction. As a result, we synthesized a potent antineoplastic nucleoside, 1-(2-deoxy-2-fluoro-β-D-4-thio-arabino-furanosyl)cytosine (4′-thioFAC), and several novel 4′-thionucleosides that possess antiherpes virus activities.
To date, many nucleoside antimetabolites have been used as chemotherapeutics for cancers and virus-caused diseases. For example, cytarabine, a cytidine analogue that contains D-arabinose in place of its ribose sugar, is used for treatment of leukemia.1,2) Gemcitabine hydrochloride (2′-deoxy-2′-difluorocytidine) developed by Eli Lilly, is an anticancer drug approved for the treatment of pancreas cancer, and also used clinically for lung cancers and other tumors.3,4) Aciclovir, developed by Welcome, is a 2′,3′-nor derivative (acyclo nucleoside) of guanosine and one of the most successful antiherpes drugs.5) Ganciclovir, which also belongs to the category of acyclo nucleosides, is used for treating diseases related to human cytomegalovirus.6,7) 3′-Azido-3′-deoxythymidine (zidovudine, AZT), discovered by Mitsuya of the National Cancer Institute (NCI; Bethesda, MD, U.S.A.), was the first anti-human immunodeficiency virus (HIV) drug approved for the treatment of AIDS, and still plays a significant role in AIDS treatment.8) In addition, the recent success of sofosbuvir, an anti-hepatitis C drug developed by Pharmasset (Gilead), has had a huge impact in the field of medicinal chemistry9) (Chart 1). The discovery of these important drugs is the fruit of concerted research efforts into the synthesis of nucleoside derivatives. However, the use of nucleoside antimetabolites also has the serious drawback that these agents may acquire drug resistance during clinical use.10–12) To overcome this problem, the synthesis of diverse and structurally novel nucleoside derivatives is urgently needed in order to continue the search for new antitumor and antiviral agents.
About 25 years ago, when we started a new project to explore antitumor and antiviral nucleosides, we were intrigued by the reports of Walker and colleagues13) and Secrist et al.14) that 2′-deoxy-4′-thionucleosides, in which the ring oxygen of the 2-deoxyribose constituting 2′-deoxynucleosides was replaced with sulfur, had potent anti-herpes virus activity and, in some cases, cytotoxicity as well. In addition, 2′-substituted cytidine derivatives, e.g., 2′-deoxy-2′-methylenecytidine (DMDC) (2)15,16) and gemcitabine,3,4) were developed and reported to have potent antitumor activity. Most of the nucleoside antimetabolites are prodrugs, which need to be converted to their active form in cells. The corresponding nucleoside mono-, di- and triphosphates are often the active forms that inhibit target enzymes, e.g., DNA polymerase and reverse transcriptase.10–12) Therefore, the nucleoside derivatives must be recognized by intracellular or viral-coded deoxynucleoside kinase as a substrate in order to exert their antitumor or antiviral effects. The above-described results for 4′-thionucleosides and 2′-substituted nucleosides strongly suggest that 2′-substituted 4′-thionucleosides could be recognized by these enzymes and would have promising biological activity. Thus, we chose a novel 2′-substituted 4′-thiocytidine, 4′-thio DMDC (3), as our target molecule for potential antitumor and antiviral agents (Chart 2).
To our knowledge, at the time that we began this project, the only 4′-thionucleosides that had been synthesized were 4′-thioribonucleosides,17–19) 4′-thioarabinonucleosides20–22) and 2′-deoxy-4′-thionucleosdes.13,14) The development of a synthetic strategy for 4′-thionucleosides applicable to the synthesis of 2′-substituted derivatives was thus a key to the success of this project. From the viewpoint of the structure–activity relationship, the synthesis should include a glycosylation reaction of nucleobase with 4-thiosugars, since various base-modified analogues could be obtained from a single 4-thiosugar intermediate by the glycosidation strategy. Thus, our first goal was to achieve the synthesis of 4-thiosugars by the method generally applicable to the synthesis of the 2-substituted 4-thiosugar derivatives.
To construct a glycosidic linkage between the base and sugar moiety of the nucleoside skeleton, the Vorbrüggen reaction is generally used.23) It was clear, at the time when we started our project, that the reaction could be used in the synthesis of 4′-thionucleosides as well as normal “4′-oxy” nucleosides. In addition, 1-acetoxy-4-thiosugar 5, one of the most effective substrates for the Vorbrüggen reaction, could be easily obtained from a corresponding sulfoxide 4 by classical Pummerer rearrangement. On the other hand, we noticed that the Vorbrüggen reaction of 1-acetate 5 generated a sulfenium ion 6 which could also be formed by the sila-Pummerer reaction, developed by Kita et al.,24,25) of the sulfoxide 4. This new type of glycosylation reaction seemed quite attractive since it could bypass one step and achieve direct access to the sulfenium ion 5 from the sulfoxide 4 (Chart 3). Thus, our second goal for the planned synthesis of 4′-thionucleoside was to develop this new “Pummerer-type” thioglycosylation reaction.
Starting from a xylose derivative 8,26) thiabicyclo sugar 9 was prepared by methanolysis of the 1,2-isopropyliden group, mesylation of the 2- and 5-hydroxy groups and subsequent inter/intramolecular SN2 reaction with sodium sulfide. Hydrolysis of 9 followed by hydride reduction gave a 4-thioarabinose derivative 10 which was protected at the primary hydroxyl group to give 11. Through the cyclic intermediate 9, chirality of the 2, 3, and 4 positions of xylose was transferred to the 4, 3, and 2 positions of the 4-thioarabinose derivative 10, respectively. Oxidation of 11 followed by the Wittig reaction gave a methylene derivative which was debenzylated and oxidized to the corresponding sulfoxide to give 12. Next, we tried the Pummerer-type thioglycosylation of N4-acetylcytosine with the sulfoxide 12. Fortunately, we found that the reaction of 12 with excess persilylated N4-acetylcytosine in the presence of trimethylsilyl trifluoromethanesulfonate (TMSOTf) gave an anomeric mixture of 4′-thioDMDC derivative 15 in good yield. The results suggested that the reaction occurred by forming the expected sulfenium ion 14 and that the concept of the Pummerer-type glycosylation was actually effective for formation of the glycoside bond of 4′-thionucleosides. Deprotection of 15 and the separation of anomers by octadecyl silica (ODS) column chromatography furnished 4′-thioDMDC (3) along with its α-anomer.27,28) We also synthesized 4′-thiogemcitabine (16) from 11 by using a similar synthetic scheme27,28) (Chart 4).
As described above, the intermediate for 4′-thioDMDC could also be used for the synthesis of other 4′-thionucleosides. Compound 10 was benzylated at the hydroxyl groups and converted to 1-acetate derivative 17 via Pummerer rearrangement. The Vorbrüggen reaction of 17 with silylated nucleobases in the presence of Lewis acid and the subsequent deprotection gave 4′-thioarabinonucleosides 18.29) Introduction of a fluoro group at the 2-position was achieved by treatment of 11 with N,N-diethylaminosulfur trifluoride (DAST) to give the 2-fluorinated compound 19 with retention of the stereochemistry at the reaction site. Similar results reported by Marquez and colleagues suggested that the reaction proceeded by a neighboring group participation of ring sulfur to form an episulfonium ion prior to the nucleophilic attack of fluoride ion.30) The synthesis of 2′-flluoro-arabino-4′-thionucleosides 21 from 19 was achieved by the Vorbrüggen reaction and the subsequent deprotection as in the case of 4′-thioarabinonucleosides28,31) (Chart 5).
The obtained 2′-substituted 4′-thionucleosides were evaluated for their antitumor and antiviral activities. Among the cytidine derivatives, 4′-thioDMDC (3) and 2′-deoxy-2′-fluoro-arabino-4′-thiocytidine (4′-thioFAC, 21: B=cytosine) were shown to have potent cytotoxicities against human T-cell leukemia, CCRF-HSB-2 and solid tumor KB cells comparable to those of cytarabine and the parental DMDC.27,28) On the other hand, 4′-thiogemcitabine (16) showed only moderate activity against the same cell lines.27,28) It is worth noting that 21 was 15–30 times more active against KB cells than cytarabine or DMDC (Table 1). Further evaluation of the antitumor activity of 21 revealed it had broad antineoplastic activity against various tumor cell lines including pancreas and colon cancers, and was also active in an in vivo assay using nude mice implanted with human tumor cells.32–34)
| Compound | 2′-Substituent | Antineoplastic activities IC50 (µg/mL) | |
|---|---|---|---|
| CCRF-HSB-2 | KB cells | ||
| 4′-ThioDMDC (3) | =CH2 | 0.0091 | 0.12 |
| 4′-Thiogemcitabine (16) | F2 | 1.5 | 17 |
| 4′-ThioFAC (21: B=Cyt) | F (arabino) | 0.051 | 0.015 |
| Cytarabine | — | 0.052 | 0.26 |
| DMDC (2) | — | 0.022 | 0.44 |
The mechanism of action of 4′-thioFAC was investigated, revealing that 4′-thioFAC inhibited cellular DNA synthesis, but did not inhibit either RNA or protein synthesis. Potent inhibition against DNA polymerase α by the action of 4′-thioFAC triphosphate (4′-thioFACTP) was observed, whereas 4′-thioFACTP showed moderate to little inhibition against DNA polymerase β and γ (Fig. 1). The kinetic analysis showed that the inhibition mode of 4′-thioFACTP against DNA polymerase α was mixed-type inhibition. The results implied that 4′-thioFACTP inhibited DNA polymerase α in two ways: (a) retarding DNA chain elongation by competing with dCTP at the nucleotide-binding site, (b) acting as a chain terminator after incorporation into DNA.35)
●: 4′-thioFACTP; ▲: dFdCTP; ×: ddCTP.
The triphosphate of gemcitabine (dFdCTP), on the other hand, did not show potent inhibition against these three DNA polymerases. This result was consistent with the knowledge that the inhibition of ribonucleotide reductase by the diphosphate of gemcitabine was more responsible for its antitumor action.12) Since these findings showed that the main target enzymes of 4′-thioFAC and gemcitabine appeared to be different, the combinatory use of 4′-thioFAC and gemcitabine was expected to have a synergic effect, which was confirmed in an in vitro model.35)
It is well known that 2′-arabinonucleosides36) and 2′-deoxy-2′-fluoroarabinonucleosides37) exhibit antiviral activities, especially against herpesviruses. We therefore evaluated the anti-herpesvirus activities of the 4′-thio counterparts of these nucleosides by culturing them with several viruses, including human herpes simplex virus types 1 and 2 (HSV-1 and -2), varicella zoster virus (VZV), and human cytomegalovirus (HCMV), followed by analysis with a plaque reduction assay.29,31,38) The results are summarized in Tables 2 and 3.
| B | Antiviral activities ED50 (µg/mL) | Anti-cell proliferative activity IC50 (µg/mL) | |||
|---|---|---|---|---|---|
| HSV-1a,e) | HSV-2b,e) | VZVc,e) | HCMVd,e) | CCRF-HSB-2f) | |
| 5-Me-Ura | 0.77 | 4.6 | 6.6 | 44 | >100 |
| 5-Et-Ura | 0.43 | 6.5 | >50 | >50 | >100 |
| 5-I-Ura | 3.50 | 13.6 | 27.1 | >50 | >100 |
| Ade | 18.4 | 13.5 | 1.85 | 1.36 | 14.8 |
| 2,6-DAP | 0.52 | 0.40 | 0.11 | 0.022 | 0.20 |
| Gua | 0.49 | 0.59 | 0.11 | 0.010 | 0.29 |
| Aciclovir | 0.14 | 0.23 | — | — | >100 |
| Ganciclovir | 0.016 | 0.039 | 3.1 | 0.21 | 17 |
a) VR-3 strain. b) HSV-2 MS strain. c) VZV Oka strain. d) HCMV AD 169 strain. e) Plaque reduction assay. f) MTT assay.
| B | Antiviral activities ED50 (µg/mL) | Antineoplastic activities IC50 (µg/mL) | |||
|---|---|---|---|---|---|
| HSV-1a,e) | HSV-2b,e) | VZVc,e) | HCMVd,e) | CCRF-HSB-2f) | |
| 5-Me-Ura | 0.43 | 1.0 | 24.5 | 0.088 | 0.49 |
| 5-Et-Ura | 0.015 | 0.073 | 29.4 | >50 | >100 |
| 5-I-Ura | 0.018 | 0.18 | 6.8 | >50 | 78 |
| Ade | 1.61 | 3.0 | 3.0 | 1.41 | 54 |
| 2,6-DAP | 0.0057 | 0.050 | 0.101 | 0.066 | 2.1 |
| Gua | 0.0091 | 0.063 | 0.095 | 0.078 | 3.6 |
| Aciclovir | 0.14 | 0.23 | — | — | >100 |
| Ganciclovir | 0.016 | 0.039 | 3.1 | 0.21 | 17 |
a) VR-3 strain. b) HSV-2 MS strain. c) VZV Oka strain. d) HCMV AD 169 strain. e) Plaque reduction assay. f) MTT assay.
The 5-substituted-4′-thioaraU derivatives (B=5-methyl, 5-ethyl- and 5-iodouracils) showed anti-HSV-1 activity (ED50: 0.43–3.50 µg/mL), but moderate to weak inhibitory activity against HSV-2 and VZV. Among the purine derivatives examined, 4′-thioaraG and 2,6-diaminopurine (2,6-DAP) 4′-thioarabinonucleoside (4′-thioaraDAP) showed potent antiviral activity against all of the herpesviruses assayed. It was noteworthy that their antiviral activities were particularly potent against HCMV and were 10 times more active than that of a commercial anti-HCMV drug, ganciclovir (0.010, 0.022 µg/mL, respectively, Table 2).
The 2′-fluoroarabinonucleosides showed results similar to those for the 2′-arabino derivatives mentioned above. Based on the comparison of the antiviral activities of 2′-fluoro derivatives with those of arabino derivatives (Tables 2, 3), the antiviral activity was clearly improved by replacing the 2′-substituent from a hydroxyl to a fluoro group on the 4-thioarabinose moiety. In the pyrimidine series, the 5-ethyluracil, 5-iodouracil, and 5-iodocuracil derivatives showed potent anti-HSV-1 as well as anti-HSV-2 activities in vitro (ED50: 0.015–0.43 µg/mL toward HSV-1, 0.073–1.0 µg/mL toward HSV-2). Among them, the 5-ethyluracil derivative was quite interesting, since it had potent inhibitory activities against both HSV-1 and HSV-2 with lesser cytotoxicity than the 4′-thioarabino counterpart. In the purine series, the guanine and 2,6-diaminopurine derivatives showed broad and prominent antiviral activities. In addition, the cytotoxicity of the purine 2′-fluoro derivatives was slightly improved in comparison to that of the arabino derivatives, but they were still cytotoxic, which impeded further development of these analogues as antiviral drugs.
We achieved the synthesis of 2′-modified 4′-thionucleosides starting from a D-xylose derivative as a chiral synthon. Considering D-xylitol instead of D-xylose as a chiral synthon, we focused on its pseudosymmetrical structure. As described above, during the synthetic scheme adopted, the chirality of the 2, 3, and 4 positions of D-xylitol was transferred to the 4, 3, and 2 positions of the D-4-thioarabinose derivative. The pseudosymmetrical structure of D-xylitol provided a clue that we should shift the carbon units of D-xylitol utilized for 4′-thionucleosides from the 2–5 to the 1–4 positions, which allowed us to synthesize the L-enantiomers of 4′-thioarabinonucleosides (Chart 6). The L-isomers of nucleoside derivatives have garnered much attention as candidates for alternative antiviral agents.39) Indeed, the L-isomer of lamivudine was more potent and less cytotoxic than its D-isomer.40–43) We concluded that our synthetic approach was effective. Adopting a similar synthetic scheme for the D-isomers depicted above, we were able to obtain L-4′-thioarabinonucleosides from D-xylose by shifting the chiral carbons employed. Interestingly, the α-L-cytosine derivative showed moderate anti-HSV-1 activity.44)
The synthesis of 4′-thionucleosides was achieved by developing a new synthetic route for the 4-thiosugar portion and combining it with the Pummerer-type glycosylation reaction. To prove the usefulness of the latter Pummerer-type glycosylation, we applied it to the synthesis of 4′-thioribonucleosides.45) We then further extended our study to the synthesis of purine 4′-thioribonucleosides and found that some of the purine 4′-thioribonucleoside derivatives exhibited anti-angiogenesis activity based on our primary screening system.46) To obtain several purine analogues, including 4′-thioguanosine, we modified the original synthesis of 4′-thioguanosine.47) A mixture of α- and β-anomers of 2-amino-6-chloropurine 4-thioriboside 24 was prepared by the Vorbrüggen reaction of tetraacetyl-4-thioribofuranose. We found that treatment of an inseparable mixture of α,β-24 with adenosine deaminase gave a pure β-anomer of 4′-thioguanosine (25) along with the unreacted α-24 (Chart 7).
We established a screening system in which the growth of human umbilical vein endothelial cells (HUVECs) was induced by the addition of the conditioned medium of lung carcinoma cell line PC-9. The primary screening using HUVECs resulted in the identification of α,β-24 and 25 as potent inhibitory compounds. The assay showed that these compounds were more potent than fumagillin (Chart 7), a fungal metabolite which was reported to have anti-angiogenic activity in vitro and in vivo (Table 4).
| Base | IC50 (µg/mL) | |
|---|---|---|
| 4′-Thionucleosides | Guanine (25) | 0.021 |
| 2-Amino-6-chloropurine (α,β-24) | 0.027 | |
| Fumagillin | 2.8 |
To evaluate the in vivo anti-angiogenic effect of 4′-thioguanosine, we performed a chick embryo chorioallantoic membrane (CAM) assay. We implanted a murine S-180 tumor onto a CAM that was grown by subcutaneous implantation in mice, and then examined the inhibitory effect of 4′-thioguanosine on the growth of the tumor. The tumor growth was inhibited by the administrations of 0.001–0.01 µg/d of 4′-thioguanosine (25) at 1, 3, 5 and 7 d after implantation. Fumagillin also inhibited the tumor growth at the same doses as 4′-thioguanosine (0.001–0.01 µg/d, Fig. 2). Since 25 was almost inactive against the cancer cell line panel tested (its IC50 value was >50 µM), 25 was suggested to inhibit the tumor growth by inhibiting the pathway that transmits the angiogenesis signal from certain unidentified growth factors produced by the tumor.46)
* p<0.05 vs. Control, Dunnett’s multiple comparison test.
As described above, our first target was 4′-thioDMDC (3), which was synthesized by developing the Pummerer-type thioglycosylation. Following our synthesis of 3, this new glycosylation method was widely used and became one of the standard methods for constructing a glycosidic bond between a nucleobase and 4-thiosugar unit.48–50) We continued to study the utility of this reaction by applying it to the synthesis of nucleoside derivatives having a 5-thiopyranose system. As a target of potential anti-HIV agents, we designed dihydrothiopyranonucleoside 27, a ring-expanded analogue of L-4′-thioD4C, which was reported to possess anti-HIV activity51) (Chart 8). As we mentioned in the context of the synthesis of L-4′-thionucleosides, both the D- and L-isomers of nucleosides were considered to be potentially active against HIV. Therefore, we planned to synthesize the target analogue as a racemate.
To synthesize the substrate of the Pummerer-type thioglycosylation for preparing 27, diene 29, prepared from 2-butene-1,4-diol, was subjected to a ring-closing metathesis (RCM) reaction catalyzed by the second generation Grubbs catalyst to give a dihydrothiopyran derivative 30 in excellent yield. After the process of one-carbon deletion, protection at the resulting primary hydroxyl group of 31 and the subsequent oxidation gave a sulfoxide 32. The Pummerer-type thioglycosylation reaction of sulfoxide 32 by treatment with bis(trimethylsilyl)uracil, TMSOTf, and N,N-diisopropylethylamine (DIPEA) gave the dihydrothiopyranyluracil derivative 33 in 45% yield along with the formation of its α-anomer. The results clearly proved that the Pummerer-type thioglycosylation reaction was effective for the synthesis of 5′-thiopyranonucleoside derivatives as well as 4′-thionucleosides. Finally, conversion of the uracil moiety of 33 to cytosine and desylilation afforded the desired dihydrothiopyranyl cytosine 27.52,53) Although 27 did not show any anti-HIV activity, its bis(hydroxymethyl) analogue 28, which was obtained by a method similar to that used for synthesis of 27, was revealed to have anti-HIV activity54) (Chart 9).
Our early efforts to identify biologically active nucleosides were made to synthesize 4′-thioDMDC and 4′-thiogemcitabine. To achieve their synthesis, we developed the Pummerer-type thioglycosylation. This reaction was successfully applied to synthesize dihydrothiopyranonucleosides as well as 4′-thionucleosides. At present, the reaction has become the standard method for building a glycosidic linkage of 4′-thionucleosides. It is also emphasized that we developed a new synthetic route accessing 4-thiosugar, by which we achieved the synthesis of various 4′-thionucleoside derivatives including L-enantiomers of 4′-thioarabinonucleosides. As a result of these synthetic efforts, we were able to identify many biologically interesting nucleosides that were active against tumors or viruses. In particular, 4′-thioFAC, a 2′-fluoro analogue of 4′-thiocytidine, showed prominent antitumor activity even in an in vivo assay. The results unambiguously demonstrated the importance of 4′-thionucleosides for drug development and their promise as drug candidates for antineoplastic and antiviral agents.
We are thankful to co-workers, whose names are cited in the references, for their intellectual and experimental contributions. This work was supported in part by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (JSPS) and by a Grant of the Strategic Research Foundation Grant-in-Aid Project for Private Universities from the Ministry of Education, Culture, Sport, Science and Technology, Japan (MEXT).
The authors declare no conflict of interest.
