Metabolism, Biochemical Actions, and Chemical Synthesis of Anticancer Nucleosides, Nucleotides, and Base Analogs

Abstract

Nucleoside, nucleotide, and base analogs have been in the clinic for decades to treat both viral pathogens and neoplasms. More than 20% of patients on anticancer chemotherapy have been treated with one or more of these analogs. This review focuses on the chemical synthesis and biology of anticancer nucleoside, nucleotide, and base analogs that are FDA-approved and in clinical development since 2000. We highlight the cellular biology and clinical biology of analogs, drug resistance mechanisms, and compound specificity towards different cancer types. Furthermore, we explore analog syntheses as well as improved and scale-up syntheses. We conclude with a discussion on what might lie ahead for medicinal chemists, biologists, and physicians as they try to improve analog efficacy through prodrug strategies and drug combinations.

Graphical Abstract

1. INTRODUCTION AND MOTIVATION

Cancer is the second leading cause of death in the United States, accounting for 1 in 4 deaths annually. Importantly, cancer incidence continues to increase worldwide.¹ Cancer encompasses a broad range of diseases in which host cells escape their normal cell cycle regulation. This phenomenon can be linked to several factors, including gender, ethnicity, age of onset, and lifestyle,² but cancer can also be caused by cellular transformation linked to viral infection, chemical exposure, or radiation exposure, or the cause can be unknown (spontaneous) in nature.^3–5 Advances in oncology treatments have increased the number of cancer survivors from 3 million in 1971 to more than 13 million in 2012.⁶ This has led to a 39% improvement in 5-year survival rates for different cancers and stages of disease as indicated by the 2010 data, and rates should continue to improve with recent advances in new therapies.^1,7

The primary standard-of-care treatment for cancer includes surgery, radiation, chemotherapy, hormone therapy, immuno-therapy, and targeted therapy.¹ Cutting-edge therapies may include one or more of these procedures listed above depending upon the type and stage of the cancer being treated. Sometimes surgery and radiation may be a second tier treatment, with chemotherapy being used first to reduce the tumor burden. Chemotherapies are divided into several main drug classes: alkylating agents,⁸ antimetabolites, anti-tumor antibiotics,⁹ topoisomerase inhibitors,¹⁰ mitotic inhibitors,^11,12 and corticosteroids.¹³ Recently, antibodies directed against programmed cell death 1 (PD-1) protein, PD-1 ligands 1 and 2, and cytotoxic T-lymphocyte-associated protein 4 (CLTA-4) have been found to activate the immune system to attack tumors, and they are therefore used with varying success to treat some types of cancer.^14–16

Chemotherapeutic nucleoside, nucleotide, and base analogs, herein referred to as “nucleoside analogs”, are antimetabolites. They are chemically modified analogs of natural nucleosides, nucleotides, and bases, which are endogenous metabolites involved in many essential cellular processes, such as DNA and RNA synthesis, and purinergic signaling. Nucleoside analogs have been a cornerstone of anticancer and antiviral chemotherapy for decades. Currently, there are 15 FDA-approved nucleoside analogs used to treat various cancers, and they account for a large percentage of the current cancer chemotherapeutic arsenal.¹⁷ Whereas chemotherapeutic nucleoside analogs alone seldom lead to a cure for cancer, they do provide a valuable treatment option for cancer patients. In addition, many other nucleoside analogs are currently being investigated in clinical trials as monotherapy or combination therapy for multiple cancers. These studies attempt to increase the potency or bioavailability of these agents while diminishing associated dose-limiting toxicity.

In this review, we focus on both the biology and chemical syntheses of the current FDA-approved anticancer nucleoside analogs (Figure 1), investigational anticancer nucleoside analogs in development since 2000 or analogs used outside of the United States (Figure 2), and nucleoside analogs in development since 2000 that have stalled in clinical trials (Figure 3). Additionally, we present an overview of the cellular metabolism, the biochemical mode of action, and the types of cancers treated with these agents, as well as focusing on current methodologies for their chemical synthesis.

Figure 1.

FDA-approved anticancer nucleoside analogs.

Figure 2.

Various nucleoside analogs in clinical trials or used outside of the U.S.

Figure 3.

Various nucleoside analogs that have stalled in clinical trials.

2. GENERAL BIOLOGICAL ACTIVITY OF ANTICANCER NUCLEOSIDE ANALOGS

The two main routes for administering nucleoside analogs are by intravenously infusion and oral formulation. Once inside the body, the analogs can face a major bottleneck in cellular uptake. Both protein facilitated diffusion using either concentrative nucleoside transporters (CNT or SLC28; three members) and/or equilibrative nucleoside transporters (ENT or SLC29; four members) are involved in the cellular uptake for the majority of agents.^18,19 Passive diffusion is limited in regard to cellular uptake for parent nucleoside analogs, with troxacitabine 48 being the exception, requiring very high extracellular concentrations in order to occur.²⁰ Certain nucleoside analog prodrugs, such as NUC-1031 32, NUC-3373 37, or elacytarabine 51, however, can enter the cell independent of transporters.

Nucleoside analogs are prodrugs and require cellular metabolism to be converted to their active metabolites. This is accomplished by the cellular salvage pathway—a group of enzymes that can phosphorylate nucleoside and nucleotide substrates. After transport across the plasma membrane, nucleoside analogs undergo an initial phosphorylation—or if a base analog, ribosylation followed by phosphorylation—to generate nucleoside-5′-monophosphate forms,²¹ which is often the rate-limiting step within the cell in the process to generate active metabolites (Figure 4).²² The cell primarily uses four kinases for nucleoside-5′-monophosphate phosphorylation: 2′-deoxycytidine kinase, thymidine kinase 1, thymidine kinase 2, and deoxyguananine kinase.²³ Each kinase has multiple substrates and can have feedback inhibition to regulate their activities.^24,25 The initial phosphorylation is often performed by 2′-deoxycytidine kinase. Cancer cells usually express 2′-deoxycytidine kinase at a 3- to 5-fold higher protein level than most normal cells, affording some degree of selectivity.²⁶ Phosphoramidate prodrugs such as NUC-1031 32 or NUC-3373 37, however, enter the cell as masked monophosphates and pass through a different enzymatic pathway in order to be converted to their nucleoside-5′-monophosphate forms.

Figure 4.

General biological mechanism of action of anticancer nucleoside analogs.

Nucleoside-5′-monophosphate analogs are converted to nucleoside-5′-diphosphate and nucleoside-5′-triphosphate forms by various cellular kinases. Nucleoside-5′-triphosphate analogs are substrates for DNA polymerases and can be incorporated into DNA during replication or DNA excision repair synthesis, which gives rise to stalled replication forks and chain termination. These events activate various DNA damage sensors, which stimulate DNA repair, halt cell progression, and often lead to apoptosis.^27,28 Since most cancer cells replicate their genome more frequently than a majority of normal adult cells, which are quiescent and not actively synthesizing their DNA, this phenomenon allows for a degree of cancer cell selectivity.¹⁷ Moreover, certain nucleoside-5′-triphosphate analogs can be incorporated into RNA, leading to transcriptional termination, and messenger RNA (mRNA) and ribosomal RNA (rRNA) instability.^29,30

Nucleoside and nucleotide (mono-, di-, or triphosphates) derivatives can also inhibit key cellular enzymes, providing a secondary mode of action that inhibits cell growth. Such enzymes include ribonucleotide reductase,³¹ which removes the 2′-OH group from the ribose sugar in order to generate de novo 2′-deoxyribonucleoside diphosphate,³² purine nucleoside phosphorylase,^33,34 which is involved in purine metabolism catalyzing the reversible phosphorolysis of purine nucleosides,³⁵ and thymidylate synthase,³⁶ discussed below (see Section 5.1.1.1). A few newer nucleoside analogs, however, exert their anticancer activities as adenosine receptor antagonists, or by inhibiting cellular enzymes that are not involved in nucleic acid synthesis. These cellular enzymes include NEDD8-activating enzyme, which is involved in the ubiquitin-proteasome degradation system and has an important role in cellular processes associated with cancer cell growth,³⁷ and DOT1L (histone-lysine N-methyltransferase, H3 lysine-79 specific), which is involved in post-translational gene modification.³⁸ Furthermore, the enzymatic activity of DOT1L represents an oncogenic driver of MLL-r leukemia.³⁹

Enzymes that dephosphorylate nucleotides are also present in the cell. 5′-Nucleotidases (5′-NTs) can dephosphorylate noncyclic nucleoside-5′-monophosphates to nucleosides and inorganic phosphates, leading to the regulation of cellular nucleotide and nucleoside levels.^40,41 There are seven intracellular 5′-NTs: cN-IA, cN-IB, cN-II, cN-III, cN-II-like, cdN, and mdN in the cell,^42,43 and they have been reported to modulated antiviral and anticancer nucleotide levels.⁴¹ Their cellular functions involve cell–cell communication and signal transduction,⁴⁴ as well as nucleic acid repair.⁴⁵ Other enzymes involved in nucleotide degradation include eN (CD73), an ecto-5′-nucleotidase that converts AMP to adenosine plus phosphate,⁴³ and ectonucleoside triphosphate diphosphohydrolase-1 (CD39), which can hydrolyze nucleoside-5′-triphosphates into nucleoside-5′-monophosphate and nucleoside-5′-diphosphate products. Tumor expression of these two enzymes, in an animal model, appears to enhance tumor growth by negatively regulating the immune response.^46,47 Furthermore, alkaline phosphatases are also present and can be targets for anticancer treatment, but they might have a lesser impact on antimetabolite metabolism.⁴⁸ Recently, a sterile alpha motif and histidine/aspartic acid domain containing protein 1 (SAMHD1) was discovered to be a dGTP/GTP-dependent deoxynucleotide triphosphohydrolase,^49,50 generating 2′-deoxynucleosides and inorganic triphosphates from the cellular canonical 2′-deoxyribonucleoside-5′-triphosphates. Clofarabine-5′-triphosphate has been shown to be a substrate for SAMHD1.⁵¹ Furthermore, the SAMHD1 gene was shown to be mutated in CLL cancer,⁵² and the protein level downregulated in lung cancer.⁵³

Cancer chemotherapy is limited in effectiveness by off target toxicity and drug resistance, with the latter leading to a decreased ability to deliver an adequate concentration of drug to cancer cells. Drug resistance can arise from pre-existing intrinsic properties of cancer cells or by acquired traits, which cancer cells develop in response to limited drug exposure.⁵⁴ Cancer cells can acquire resistance to nucleoside analogs by several different mechanisms. First, many studies using cell lines exposed to nucleoside analogs have shown transporter mutations, which generally do not occur in cancer patients. Although a decrease in nucleoside transport in patients was detected, it was dependent on cell differentiation state: myeloblasts versus lymphoblasts.^55,56 For example, high expression of hENT1 was a potential co-determinant of poor clinical response to 5-FU 2 in cells ex vivo from colorectal cancer patients,⁵⁷ or using RNA1 to downregulate hENT1 in the pancreatic cell line.⁵⁸ In contrast, high hENT1 expression was reported as a biomarker for survival in patients with pancreatic cancer treated with gemcitabine.⁵⁹ Second, modulation of phosphorylating enzymes, such as 2′-deoxycytidine kinase, has been reported when using cancer cell lines.⁶⁰ In patients, however, resistance to nucleoside analogs by phosphorylating enzymes has been reported to be linked to decreases in mRNA levels and a decrease in enzyme activity.^61,62 Third, resistance to nucleoside analogs can also occur by augmentation of nucleotide excision repair machinery.⁶³

An increase in cellular nucleopside analog efflux provides another mechanism of resistance.⁶⁴ Cellular nucleoside analog efflux is performed by several ATP-Binding Cassette (ABC) transporters, such as ABCB1 (MDR1), ABCC4 (MRP4), ABCC5 (MRP5), and ABCC11 (MRP8).^65,66 Furthermore, metabolic inactivation of nucleoside analogs can occur by the actions of cellular enzymes such as adenosine deaminase, cytidine deaminase, and purine nucleoside phosphorylase.^67–69

3. GENERAL SYNTHETIC APPROACHES TO ANTICANCER NUCLEOSIDE ANALOGS

Syntheses highlighted in this review use three general approaches to access anticancer nucleoside analogs: divergent, convergent, and trans-glycosylation (Figure 5).⁷⁰ The divergent approach begins with an intact nucleoside that is subsequently modified at either the sugar or base. A major advantage of this method is that the stereochemistry, especially of the key glycosidic bond (β-anomer), is already fixed. A drawback of this approach is the presence of two or three hydroxyl groups, which are relatively similar in chemical reactivity, and often require multiple protection/deprotection steps.

Figure 5.

General synthetic approaches to anticancer nucleoside analogs.

The convergent approach couples a nitrogenous base with a modified sugar in a glycosylation reaction. This method allows for more synthetic diversity, and it is more widely used, although the stereochemistry of the glycosylation can be an issue. The reaction is most often performed under Silyl–Hilbert–Johnson (Vorbrüggen) conditions or metal salt coupling conditions.

Benzoyl-protected furanose sugars are often used in glycosylations performed under Vorbrüggen conditions, because the stereochemistry of the glycosylation reaction is almost completely selective. Anchimeric assistance of the 2′-O-benzoyl group, protected furanose sugar II, leads to intermediate III, which effectively has its α-face blocked. This allows for selective formation of the protected β-anomer IV. Difficulties arise when dealing with 2′-deoxy furanose sugars. The absence of a 2′-O-benzoyl group precludes formation of an intermediate with a sterically hindered α-face, often leading to an anomeric mixture of products VII, and many times requires laborious chromatographic separation.

The metal salt glycosylation generally couples a furanose sugar possessing a leaving group (often halogens) with α-stereochemistry and a salt of a nitrogenous base. Conditions are sought to make the reaction predominantly (or exclusively if possible) S_N2 in nature, in an attempt to elude anomerization of the leaving group or avoid the formation of an oxonium ion intermediate similar to VI, favoring a S_N1-like reaction.

Enzyme-mediated base exchange is an addition method for nucleoside synthesis. Purine nucleoside phosphorylase or uridine phosphorylase can be used to replace one nucleoside base with another in a trans-glycosylation reaction. Overall, enzymes can selectively generate the desired β-anomer nucleoside.

4. PURINE AND ACYCLIC ANALOGS

4.1. Thiopurines

4.1.1. Thiopurines Biology.

In the early 1950s, Elion’s group, while at the Wellcome Research Laboratories, discovered that hypoxanthine and guanine substituted with sulfur at the 6 position led to purine synthesis inhibition, which in turn decreased growth of Lactobacillus casei.⁷¹ Subsequently, 6-mercaptopurine (6-MP, 1) and 6-thioguanine (6-TG, 3) were demonstrated to have potent anticancer activities against a wide variety of rodent tumors. The more promising 6-MP 1 was rapidly advanced into clinical trials; in 1953 it received FDA approval for treating pediatric acute lymphocytic leukemia (ALL), a rapid-developing cancer of the blood and bone marrow in which immature lymphocytes are overproduced.

Currently, 6-MP 1 and 6-TG 3 have various clinical applications. They can effectively treat hematologic cancers (childhood and adult leukemias).⁷² 6-TG 3 is used to treat acute myelogenous leukemia, a cancer involving abnormal white blood cells of myeloid linage that rapidly divide and interfere with normal white blood cell production, whereas 6-MP 1 is used in combination with other chemotherapeutic agents and remains part of the standard of care for acute lymphocytic leukemia.⁷³ These compounds are also being used to treat autoimmune disorders, such as rheumatoid arthritis, psoriasis, and ulcerative colitis,^74–76 and they have also been employed to prevent solid organ transplant rejection.^30,77,78

Thiobases require further metabolism by cellular enzymes to reach their active forms (Figure 6).⁷⁹ 6-MP 1 is converted by hypoxanthine/guanine phosphoribosyl transferase to 6-thioinosine-5′-monophosphate III and then converted to 6-thio-guanosine-5′-monophosphate VI, which is also the product of 6-TG 3 and hypoxanthine/guanine phosphoribosyl transferase. Further phosphorylation leads to 6-thio-guanosine-5′-triphosphate VIII, which can be incorporated into RNA.²⁹ However, Nelson et al. have reported that 6-MP 1 activity was not directed toward specifically inhibiting RNA synthesis, suggesting that 6-MP-5′-triphosphate VIII incorporation into RNA was not a major mechanism for its anticancer activity in vitro.⁸⁰

Figure 6.

Metabolism of 6-MP 1 and 6-TG 3.

A major active metabolite of both 6-MP 1 and 6-TG 3 is 6-thio-2′-deoxyguanosine-5′-triphosphate XI, which is produced from ribonucleotide reductase-mediated deoxygenation of 6-TG-5′-diphosphate VII, followed by phosphorylation.¹⁷ 6-Thio-2′-deoxyguanosine-5′-triphosphate XI can be incorporated into DNA. DNA polymerases continue polymerization after 6-thio-2′-deoxyguanosine-5′-monophosphate incorporation, resulting in multiple thiobase derivatives embedded in the newly synthesized DNA strand. The reactive thiol group of the incorporated active metabolite is susceptible to nonenzymatic methylation, leading to incorporated 6-methylthioguanine-5′-monophosphate IX, which preferentially base-pairs with a thymine during DNA replication.⁸¹ The 6-methylthioguanine:-thymine base-pairings resemble DNA replication errors, which are recognized by cellular mismatch repair enzymes. It is hypothesized that 6-methylthioguanine:thymine base-pairing leads to a futile cycle of repair, which ultimately generates cytotoxic DNA damage.⁸¹

Another major active metabolite of 6-MP 1 is methylthioino-sine-5′-monophosphate IV, which is formed by the methylation of thioinosine-5′-monophosphate III by thiopurine S-methyltransferase.⁸² Metabolite IV inhibits phosphoribosylpyrophosphate (PRPP) amidotransferase, the first enzyme in the de novo purine biosynthesis pathway.⁸³ PRPP amidotransferase inhibition leads to a decrease in purine nucleotide pools, which can disrupt DNA synthesis and repair, ultimately leading to cell death in vitro.⁸⁴ 6-Thioguanine-5′-monophosphate VI can also be methylated by S-methyltransferase to form methylthioguano-sine-5′-monophosphate IX. However, this metabolite is a weak inhibitor of PRPP amidotransferase.

Two enzymes involved in the detoxification of 6-MP 1 and 6-TG 3 are thiopurine methyltransferase and xanthine oxidase. The former directly methylates 6-MP 1 and 6-TG 3, to metabolites II and V, respectively. Xanthine oxidase catabolizes 6-MP 1 to the 6-thiouric acid derivative I. However, because xanthine oxidase expression is low in hematopoietic cells, thiopurine S-methylation is considered to be the predominant pathway of detoxification of 6-MP 1 and 6-TG 3.⁷⁹ In addition, six different variant alleles have been identified from clinical samples that lead to lower thiopurine methyltransferase activity.⁷⁹

Thiopurine resistance can arise due to the action of ATP-Binding Cassette (ABC) transporters. 6-MP-5′-monophosphate III and 6-TG-5′-monophosphate IV are substrates for ABC transporters: MRP4 and MRP5, for active transport out of cancer cells.^85,86 Additional mechanisms of resistance identified using cell lines are the up regulation of P-glycoprotein⁸⁷ and decreased hypoxanthine-guanine phosphoribosyltransferase activity.⁸⁸ A genomic approach revealed that overexpression of the ATM/p53/p21 pathway, overexpression of TNFRSF10D, and overexpression of CCNG2 may also contribute to thiopurine resistance in cell lines.⁸⁹ Besides the natural mutations in thiopurine S-methyltransferase gene,⁷⁹ mutations in NT5C2 transporter gene led to an increase in nucleotidase activity and resistance to 6-mercaptopurine 1 and 6-thioguanine 3 in ALL patients.^90,91 Over 170 clinical trials evaluating thiopurines as a cancer treatment are listed at ClinicalTrials.gov.

4.1.2. Thiopurines Synthesis.

Elion and Hitchings reported the preparation of 6-MP 1 in 1952 (Scheme 1).⁹² For their studies, they developed a multigram-scale synthesis of hypoxanthine 57 starting from 4-amino-6-hydroxy-5-nitrosopyrimidine-2-thiol 54.⁹³ The heterocycle 54 was reduced, dethiolated, and cyclized, forming 57, which was subsequently treated with P₂S₅, furnishing the desired 6-MP 1 Conversion of hypoxanthine 57 to 6-MP 1 using P₂S₅ has also been used in a recent process synthesis, which was done in toluene on a 200-g-scale with a yield of 93%.⁹⁴ Additionally, this transformation has been effected on a gram-scale using the crystalline and storable P₄S₁₀-pyridine complex reagent 58.⁹⁵

Scheme 1. Synthesis of 6-Mercaptopurine 1^a.

^aReagents and conditions: (a) Na₂S₂O₄, H₂O, 60–70 °C, 80–85%; (b) Raney Ni, Na₂CO₃, H₂O, reflux, 2 h then H₂SO₄, 88%; (c) HCO₂H, reflux, 2 h then aq NaOH, AcOH, 93%; (d) P₂S₅, tetralin, 190–200 °C, 12 h, 40%; or 58, DMSO₂, 165–170 °C, 85%.

Elion and Hitchings synthesized 6-thioguanine (6-TG, 3) in 1954 (Scheme 2).⁹⁶ Guanine 59 was converted to 6-TG 3 by treatment with P₂S₅ in refluxing pyridine. Diacylated guanine 60 has also been used to produce 3. Sulfuration with P₂S₅, followed by deprotection, furnished the product on 100-g-scale (Scheme 2).⁹⁷ In addition, the sulfuration of 60 has been effected by treatment with Lawesson’s reagent.⁹⁸

Scheme 2. Synthesis of 6-Thioguanine 3^a.

^aReagents and conditions: (a) P₂S₅, pyr, reflux, 2.5 h, 32%; (b) P₂S₅, pyridine hydrochloride, pyr, 110 °C, 4 h, then HCl, H₂O, pH 4, recrystallization, 75%.

4.2. GS-9219

4.2.1. GS-9219 Biology.

Several acyclic nucleotide derivatives, such as 9-(2-phosphonylmethoxyethyl)guanine (PMEG), 9-(2-phosphonylmethoxyethyl)-2,6-diaminopurine (PMEDAP), and 9-(2-phosphonylmethoxyethyl)adenine (PMEA), display antiproliferative activities in cancer cells (Figure 7).⁹⁹ The derivatives bear a phosphonate group, which is a bioisostere of a phosphate, but metabolically and chemically more stable.¹⁰⁰ The acyclic nucleotide-5′-diphosphates are structural analogs of naturally occurring 2′-deoxyribonucleoside-5′-triphosphates and can be used as substrates for cellular DNA polymerases, acting as absolute chain terminators.¹⁰¹

Figure 7.

Structures of PMEG, PMEDAP, and PMEA, and proposed cellular metabolism of GS-9219 53.

PMEG has been shown to be the most cytotoxic among the acyclic derivatives studied.⁹⁹ Within the cell, PMEG is converted to its diphosphate form, which is the active metabolite. The PMEG-5′-diphosphate potently inhibits DNA polymerases α, δ, and ε (DNA chromosomal replication enzymes), leading to inhibition of DNA synthesis and repair.¹⁰² The 3′ → 5′ exonuclease activity of DNA polymerase can remove PMEG incorporated during DNA synthesis.⁹⁹ However, if not excised from the growing DNA strand, the PMEG-5′-monophosphate, which lacks a 3′-OH moiety, acts as an absolute DNA synthesis chain terminator.

Unfortunately, PMEG displays poor cellular permeability properties, in addition to kidney and gastrointestinal tract associated toxicities.¹⁰² Therefore, researchers at Gilead developed GS-9219 (53, VDC-1101), a double prodrug of PMEG, in an attempt to preferentially target lymphoid tissues.¹⁰³ The N⁶-cyclopropyl moiety was installed to increase specific intracellular activation as well as decrease plasma exposure to the toxic PMEG. The phosphorodiamidate prodrug was installed to increase lymphoid cell and tissue loading efficacy. Such prodrugs are also often employed to increase compound solubility, cellular penetration, and selective tissue accumulation.¹⁰⁴ GS-9219 53 showed potent antiproliferative activity against activated lymphocytes and hematopoietic cancer cells in vitro and in canines.¹⁰³ It was proposed that GS-9219 53 is intracellularly converted to PMEG-5′-diphosphate in three steps: enzymatic hydrolysis of the amino acid of the prodrug (involving cathepsin A), deamination of the cyclopropyl amino moiety (involving N⁶-methyl-AMP-aminohydrolase), and phosphorylation of PMEG (Figure 7).¹⁰⁵

Mutations in human adenosine deaminase-like proteins (H286R and S180N) have recently been found to promote resistance to GS-9219 53 and to 9-(2-phosphonylmethoxyethyl)-N⁶-cyclopropyl-2,6-diaminopurine, an intermediate in the conversion pathway of GS-9219 53 to PMEG.¹⁰⁶ Furthermore, point mutations in human guanylate kinase (S35N and V168F) were shown to promote GS-9219 resistance in cell lines.¹⁰⁷

Unfortunately, GS-9219 53 has stalled in human clinical trials for treatment of non-Hodgkin’s lymphoma, chronic lymphocytic lymphoma, and multiple myeloma due to an unacceptable safety profile (ClinicalTrials.gov identifier: NCT00499239—terminated—no results posted). However, the agent was FDA-approved for canine lymphoma in June 2013,^108,109 and an additional Phase II study was done evaluating the effectiveness in canine cutaneous T-cell lymphoma.¹¹⁰ GS-343074 and GS-424044 are two additional prodrugs synthesized after GS-9219 53 and are being evaluated as anticancer prodrugs, particularly for canine cancers. No structures have been revealed for these two compounds.

4.2.2. GS-9219 Synthesis.

In 2005, Gilead disclosed their discovery synthesis of GS-9219 53 (Scheme 3).¹¹¹ 2-Amino-6-chloropurine 61 was selectively N⁹-alkylated, and subsequent nucleophilic aromatic substitution with cyclopropylamine furnished alkyl diester 63. TMSBr-mediated deprotection led to the phosphonic acid derivative 64, which was sufficiently pure to require no chromatography after the reaction. Bis-amidate formation was effected by treatment of acid 64 with 2,2′-dithiodipyridine, triphenylphosphine, and D-alanine ethyl ester HCl.

Scheme 3. Synthesis of GS-9219 53^a.

^aReagents and conditions: (a) HO(CH₂)₂OCH₂PO(O-iPr)₂, DIAD, PPh₃, DMF, −15 °C to rt, 4 h, 63%; (b) cyclopropylamine, CH₃CN, 100 °C, 4 h, 90%; (c) TMSBr, CH₃CN, rt, overnight, 90%; (d) d-alanine ethyl ester HCl, 2,2′-dithiodipyridine, PPh₃, Et₃N, pyr, 60 °C, overnight, 50%.

Recently, Jansa et al. reported an improved route to bisamidate prodrugs.¹¹² These researchers surmised that the intermediate 65 would react directly with 2,2′-dithiodipyridine, triphenylphosphine, and the amino acid to produce the bisamidate (Scheme 4). This would obviate the need for the laborious purification step often needed to isolate and purity the intermediate diacid. The approach was successful and was used to synthesize GS-9219 53.¹¹³ TMS bromide-mediated deprotection of the phosphonate diester 63¹¹³ formed the bis(TMS)ester intermediate 65, maintained by strictly avoiding any contact with air. The bis(TMS)ester intermediate was then directly converted to GS-9219 53.

Scheme 4. Synthesis of GS-9219 53^a.

^aReagents and conditions: (a) Cl(CH₂)₂OCH₂P(O)(O-iPr)₂, Cs₂CO₃, DMF, 80 °C, 8 h, 56%; (b) cyclopropylamine, EtOH, reflux, 3 h, 65%; (c) TMSBr, CH₃CN, rt, overnight; (d) d-alanine ethyl ester, aldrithiol-2, PPh₃, Et₃N, pyr, 50 °C, 3 h, 92–98% (over two steps).

5. PYRIMIDINE ANALOGS

5.1. Fluorinated Pyrimidines

5.1.1. Fluorouracil (5-FU), Prodrugs, and Combinations Biologies.

5.1.1.1. Fluorouracil (5-FU) Biology.

In 1954, Rutman and colleagues observed that rat liver tumor cells utilized more uracil 18 as compared to normal liver cells.¹¹⁴ Thymidylate synthase can catalyze the conversion of deoxyuridine-5′-monophosphate to thymidine-5′-monophosphate using 5,10-methylenetetrahydrofolate as a methyl source. These observations led Heidelberger and colleagues to synthesize fluorouracil (5-FU, 2). Their hypothesis was that the 5-FU 2 would be converted to 5-fluorouracil-5′-monophosphate, and this product would selectively inhibit thymidylate synthase by forming a complex that could not breakdown due to prevention of elimination by the 5-fluorine (Figure 8).^36,115 Furthermore, they surmised that this would selectively target tumor cells, since 5-FU-5′-monophosphate would be found at higher concentrations in these cells. Moreover, the inhibition of thymidylate synthase and subsequent decrease of thymidine would be deleterious for cancer cells growth.

Figure 8.

Complex of 5,10-methylenetetrahydrofolate and thymidylate synthase with 5-FU-5′-monophosphate.

These researchers subsequently validated their hypothesis by showing that 5-FU 2 inhibited the production of thymine in vitro,³⁶ and soon thereafter reported that the agent inhibited tumor growth in animals.¹¹⁶ Other reports have shown that the main mechanism of action for 5-FU 2 was the inhibition of synthesis,¹¹⁷ although RNA metabolism was also influenced.¹¹⁸

5-FU 2 is used as a palliative treatment for colorectal, head and neck, and breast cancers.¹¹⁹ Treatment can also include a biomodulator, such as leucovorin, which increases folate cofactors and the efficacy of 5-FU 2.^120,121 Although introduced over 50 years ago, 5-FU 2 is still extensively studied, with over 2000 clinical trials listed at ClinicalTrials.gov.

Mechanisms of action for 5-FU 2 are complex, and an overview of its metabolism is shown in Figure 9. The agent is administered by intravenous infusion and is rapidly eliminated from the plasma, having a half-life of 8–20 min.¹²² 5-FU 2 enters the cell by facilitated transport,¹²³ and subsequent ribosylation and phosphorylation occur by orotate phosphoribosyltransferase or in two steps via uridine phosphorylase and uridine kinase.^119,124 Nucleotide kinases convert 5-FU-5′-monophosphate IV to the active metabolite 5-FU-5′-triphosphate VIII, which may be incorporated into many species of RNA. This leads to several levels of RNA disruption, including rRNA maturation inhibition^101,125 and disruption of post-translational modifications of tRNAs,¹²⁶ which can lead to cell death.

Figure 9.

Metabolism of 5-FU 2 and derivatives.

5-FU-5′-diphosphate VII is also a substrate of ribonucleotide reductase, resulting in 2′-deoxyribose-5-FU-5′-diphosphate generation. It is further phosphorylated to 2′-deoxyribose-5-FU-5′-triphosphate VI (an additional active metabolite), which then can be incorporated into DNA by cellular polymerases. 2′-Deoxyribose-5-FU-5′-monophosphate embedded into DNA does not act as a chain terminator or delayed chain terminator as compared to other antimetabolites. Instead, uracil glycosylase can excise the embedded 2′-deoxyribose-5-FU-5′-monophosphate from the DNA, or the activation of homologous recombination can occur.¹²⁷ This apyrimidinic site is recognized by apurinic/apyrimidinic endonuclease 1 and can promote a DNA single-strand break.¹⁷ DNA repair enzymes recognize this break in the DNA, and a futile cycle of excision, repair, and nucleotide misincorporation ensues, which eventually causes cell death.

2′-Deoxyribose-5-FU-5′-triphosphate VI is converted by dUTPase to 2′-deoxyribose-5-FU-5′-monophosphate II, which then inhibits the enzyme thymidylate synthase. Inhibition of thymidylate synthase leads to a decrease in dTMP cellular concentration and an increase in dUMP concentration. As a result, less dTTP is available for DNA polymerases during DNA synthesis. This can promote the misincorporation of dUTP into newly synthesized DNA, which may result in the futile cycle mentioned above. Incorporation of dUTP into DNA has been shown to be mutagenic and promote genomic instability.^128–130

There are several major obstacles for the effective use of 5-FU 2 as a chemotherapeutic. First, up to 80% of administered 5-FU 2 is catabolyzed by dihydropyrimidine dehydrogenase, which is abundantly expressed in the liver and acts as an innate resistance mechanism. Dihydropyrimidine dehydrogenase converts 5-FU 2 to dihydrofluorouracil III and is the rate-limiting enzyme in the catabolic pathway.¹²² Secondly, 2′deoxy-5-FU-5′-monophosphate and 5-FU-5′-monophosphate metabolites (II and IV) can be exported from cancer cells via the ABC transporters MRP5 and MRP8, limiting the therapeutic impact of this compound in vitro.^131,132

Several mechanisms for 5-FU 2 resistance have been reported. The miRNA-320a has been shown to downregulate PDCD4 (programmed cell death protein 4) gene expression, thereby promoting 5-FU 2 resistance in vitro.¹³³ Also, miRNA-21 induces resistance by regulating PTEN and PDCD4 levels in vitro, which correlate to cell survival and cell death pathways, respectively.¹³⁴ Overexpression of nicotinamide N-methyltransferase is common in many cancer cell lines and has been linked to 5-FU 2 resistance by regulating nitric oxide production in vitro. Reducing reactive oxygen species leads to inactivation of the ASK1-p38 mitogen-activated protein kinase (MAPK) pathway, leading to a reduction in 5-FU 2-induced apoptosis in vitro.¹³⁵ Furthermore, putrescine released by macrophage has been shown to promote 5-FU 2 resistance in vitro, which might be important for regulating 5-FU 2 concentration in the microenvironment of solid tumors in vivo.¹³⁶ Overall, 5-FU 2 resistance appears to be a multifactorial event which includes transport mechanisms, metabolism, molecular mechanisms, protection from apoptosis, and resistance via cell cycle kinetics.

In patients, 5-FU 2 resistance has been linked to thymidylate synthase methylenetetrahydrofolate reductase, dihydropyrimidine dehydrogenase, and thymidine phosphorylase, which influence 5-FU 2 metabolism.^137,138 There have been over 1500 clinical trials performed that evaluated 5-FU 2 therapy listed at ClinicalTrials.gov.

5.1.1.2. Doxifluridine Biology.

5-FU 2 is administered intravenously due to the high level of dihydropyrimidine dehydrogenase in the gut wall, which rapidly degrades the agent. Researchers at Roche used prodrug strategies to develop an orally available prodrug that overcomes degradation in the gut leading to enhanced absorption and an improved pharmacokinetic profile.¹³⁹ This work led to the development of doxifluridine (5′-deoxy-5-fluorouridine) 23, an orally available prodrug which is converted to 5-FU 2 in tumors by either thymidine phosphorylase or pyrimidine-nucleoside phosphorylase.^140,141

High expression levels of pyrimidine-nucleoside phosphorylase have been reported for esophageal squamous cell carcinoma and colorectal cancer patients.^142,143 Furthermore, thymidine phosphorylase is expressed at relatively high levels in tumor tissues such as esophageal, breast, cervical, pancreatic, and hepatic,^144,145 which increase drug specificity. However, high thymidine phosphorylase expression is also found in the human intestinal tract. Therefore, doxifluridine treatment can result in dose-limiting toxicity (diarrhea) in some individuals.¹⁴⁶ In addition, the most frequent adverse effects for doxifluridine 23 were neurotoxicity and mucositis, whereas leukopenia and nausea were reported for 5-FU 2 in patients.¹⁴⁷ Several clinical trials have been done using doxifluridine 23 as a monotherapy or in combination therapy with mitomycin-C and cisplatin (administered intravenously).¹⁴⁸

5.1.1.3. Capecitabine Biology.

As mentioned above, doxifluridine 23 treatment had gastrointestinal toxicity in patients. This led Roche to develop capecitabine 10 (5′-deoxy-5-fluoro-N-[(pentyloxy)carbonyl]cytidine), a third-generation prodrug of 5-FU 2 that was FDA-approved in 1998.^149,150 Capecitabine 10 is almost 100% orally bioavailable and is not a substrate for thymidine phosphorylase.¹⁵¹ After passage through the intestinal mucosa, carboxylesterases in the liver cleave the N⁴-pentyl carbamate. Next, cytidine deaminase, in either the liver or tumor, catalyses the formation of doxifluridine 23, which is subsequently converted by thymidine phosphorylase to 5-FU 2.¹⁵² Capecitabine 10 was highly effective in HCT116 human cancer xenograft models and showed better selective tumor delivery as compared to either doxifluridine 23 or 5-FU 2.¹⁵³ In addition, it was found that radiation therapy stimulated thymidine phosphorylase activity, and combination therapy with capecitabine 10 enhanced tumor cell selectivity.¹⁵⁴

Drug resistance to capecitabine 10 appears to involve the interplay between dihydropyrimidine dehydrogenase and thymidine phosphorylase, which are involved in the conversion of capecitabine 10 to 5-FU 2 and the degradation of 5-FU 2 within tumors.¹⁵⁵ In addition to these resistance mechanisms, drug efflux is associated with gene polymorphisms of the ABCB1 transporter in patients.⁶⁶

Currently, capecitabine 10 is FDA-approved for treatment of metastatic colorectal and breast cancers. Moreover, several clinical trials are underway using capecitabine 10 in combination therapy. A Phase II study for capecitabine 10 and bendamustine (DNA alkylating agent) in women with pretreated locally advanced or metastatic Her2-negative breast cancer (MBC-6) is being conducted (ClinicalTrials.gov identifier: NCT01891227—active status—not recruiting patients). A Phase III study is being conducted that examines capecitabine 10 maintenance therapy following treatment with capecitabine 10 in combination with docetaxel (an antimitotic taxane) in treatment of mBC (CAMELLIA study) (ClinicalTrials.gov identifier: NCT01917279—recruiting patients). Capecitabine 10 is also being evaluated as a follow up therapy for efficacy of capecitabine 10 metronomic chemotherapy to triple-negative breast cancer (SYSUCC-001 study) (ClinicalTrials.gov identifier: NCT01112826—recruiting patients). It is also under examination for maintenance therapy: a Phase II study of maintenance capecitabine 10 to treat resectable colorectal cancer (CAMCO study) (ClinicalTrials.gov identifier: NCT01880658—recruiting patients).

5.1.1.4. NUC-3373 Biology.

In 2011 a novel 5-FU 2 prodrug NUC-3373 37 (aka. NUC-3073 and FUDR) was reported having potent anticancer activity in vitro.^156,157 NUC-3373 37, a phosphoramidate prodrug of floxuridine 5, was shown to be resistant to thymidine phosphorylase activity, and showed (in comparison to 5) a significantly decreased loss of potency in CEM cells deficient in hENT1 transporters.¹⁵⁷ The authors also proposed a mechanism for the prodrug decomposition sequence including enzyme-mediated carboxyl ester hydrolysis, spontaneous cyclization with concomitant loss of the aryloxy leaving group, water-mediated opening of the mixed anhydride, and phosphoramidase-mediated P–N bond cleavage, resulting in 2′-deoxyribose-5-FU-5′-monophosphate II (Figure 10).

Figure 10.

Proposed mechanism for phosphoramidate decomposition of NUC-3373 37.

NUC-3373 37 is a mixture of phosphoramidate diastereomers, and it has been shown to have many impressive prodrug attributes, overcoming the major 5-FU 2 resistance mechanisms. The agent retains activity in thymidine kinase deficient cells, is resistant to thymidine phosphorylase activity, and maintains activity in hENT1 deficient cells by entering independently of nucleoside transporters. It is resistant to dihydropyrimidine dehydrogenase catabolism, shows lower toxicity than 5-FU 2 in a variety of cell lines, and achieves high intracellular levels of 2′-deoxyribose-5-FU-5′-monophosphate II. Furthermore, the agent rapidly distributes to tissues following bolus infusion, shows low plasma levels of degradation metabolites, and reduces tumor weigh in HT29 colorectal xenografts significantly more than 5-FU 2.¹⁵⁷ This impressive profile has led to NUC-3373 37 entering a Phase I clinical trial in late 2015 for treatment of advanced solid tumors (ClinicalTrials.gov identifier: NCT02723240—recruiting patients).

5.1.1.5. Floxuridine and 5′-Fluoro-2′-deoxycytidine with Tetrahydrouridine Biologies.

The National Cancer Institute was an early developer of floxuridine 5 (5-fluoro-2′-deoxyuridine, FdUrd), which has been reported to be 10–100 times more effective at inhibiting tumor cell proliferation in vitro than 5-FU 2.¹⁵⁸ Floxuridine 5 gained FDA approval in 1970, and was marketed by Roche.¹⁵⁹ It is approved, but not widely used, for colon and colorectal cancers that have metastasized to the liver, kidney, and stomach.¹⁶⁰

Floxuridine 5 is a hydrophilic compound, and is likely transported into cells by pyrimidine nucleoside transporters.¹⁶¹ Thymidine kinase converts floxuridine 5 to floxuridine-5′-monophosphate II, which, as mentioned above, inhibits thymidylate synthase (Figure 9).

In the intestine, a significant amount of floxuridine 5 is converted to 5-FU 2 by thymidine phosphorylase. Therefore, hepatic arterial infusion has been employed to bypass the intestine and deliver more intact floxuridine 5 to the liver.¹⁶² Furthermore, the agent exhibits two properties associated with drugs that are best fit for hepatic arterial infusion: a high hepatic extraction and a short plasma half-life.¹⁶⁰ These properties can lead to increased hepatic levels of floxuridine 5 and decreased systemic levels of the compound, which can decrease overall toxicity. This procedure has shown improvements for floxuridine 5 treatment for colorectal cancers.^163–165

Several mechanisms of floxuridine resistance have been reported. Studies have shown a decrease in orotate phosphoribosyltransferase activity and thymidine kinase activity in vitro, as well as an increase of thymidylate synthase mRNA expression in vitro.¹⁶⁶ In patients, an increase in thymidylate synthase levels has been correlated with floxuridine resistance.¹⁶⁷

Several clinical trials that include floxuridine 5 are currently underway. These include hepatic arterial infusion with floxuridine 5 and the anti-inflammatory and immunosuppressive steroid derivative dexamethasone (ClinicalTrials.gov identifier: NCT01862315—recruiting patients). Another trial examines hepatic arterial infusion with floxuridine 5 and dexamethasone in combination with gemcitabine hydrochloride 9 or the antiangiogenic monoclonal antibody bevacizumab (ClinicalTrials. gov identifiers: NCT01938729—recruiting patients and NCT00200200—active status—not yet recruiting patients). Recent results from a Phase I trial using floxuridine 5 with modified oxaliplatin, 5-fluorouracil 2, and leucovorin (m-FOLFOX6) were reported.¹⁶⁴

A major problem for anticancer cytidine analogs (gemcitabine, cytarabine 4, azacytidine 11, decitabine 14, capecitabine 10, and 5-fluoro-2′-deoxycytidine 38) is that they are substrates for cytidine deaminase and can be rapidly converted to uridine analogs, which are often inactive metabolites.¹⁶⁸ For example, the liver expresses high levels of cytidine deaminase that provide a microenvironment to protect cancer cells from decitabine 14 treatment.¹⁶⁹ In attempts to overcome the problem of deamination, cytidine deaminase inhibitors, tetrahydrouridine (39, NSC-112907) and (4R)-2′-deoxy-2′,2′-difluoro-3,4,5,6-tetrahydrouridine (Figure 11), are being investigated with nucleoside analogs as potential combination therapies.

Figure 11.

Structures of tetrahydrouridine 39 and (4R)-2′-deoxy-2′,2′-difluoro-3,4,5,6-tetrahydrouridine.

Tetrahydrouridine 39 has been investigated since the 1970s and been evaluated in cell lines and mice with cytarabine 4, gemcitabine, azacytidine (11, and decitabine 14.^170–174 How ever, clinical data has shown that patients with the cytidine deaminase gene G208A polymorphism had several adverse reactions to gemcitabine treatment.^175,176 In addition, the cytidine deaminase gene K27Q polymorphism promotes greater catalytic activity toward deoxycytine and cytarabine 4.¹⁷⁷ Collectively, these data suggest prescreening patients for polymorphisms would be helpful in guiding treatment options.

Currently 5-fluoro-2′-deoxycytidine (FdCyd, 38) is being studied in combination with tetrahydrouridine 39 in mice.^168,174 Coadministration of 38 with tetrahydrouridine 39 led to higher levels of FdCyd-5′-monophosphate, produced by 2′-deoxycyti-dine kinase in mice.¹⁷⁸ Once in the cell, FdCyd 38 can be converted to FdUrd by the action of cytidine deaminase, and subsequently, phosphorylation leads to FdUrd-5′-monophosphate, which inhibits thymidylate synthase. Furthermore, FdCyd-5′-monophosphate can be phosphorylated to FdCyd-5′-triphosphate and incorporated into DNA. Once incorporated into DNA, FdCyd 38 has been shown to inhibit DNA methyltransferase with comparable activity to azacytidine 11 (see Section 5.2.1).¹⁷⁹ Therefore, FdCyd 38 exposure can lead to the following direct and indirect cytotoxic activities in cells: (1) FdCyd-5′-monophosphate inhibition of DNA methylation, (2) 2′-deoxyribose-5-FU-5′-monophosphate II inhibition of thymidylate synthase, and (3) 2′-deoxy-5-FU-5′-triphosphate VI and 5-FU-5′-triphosphate VIII incorporation into DNA and RNA, respectively.¹⁸⁰

Both preclinical and clinical trials are currently underway using FdCyd 38 with tetrahydrouridine 39. In 2016, a preclinical evaluation of FdCyd 38 and tetrahydrouridine 39 in pediatric brain tumors was reported,¹⁸¹ and a methods development report was published evaluating a combination treatment of tetrahydrouridine 39 and FdCyd 38 from human serum using HPLC-MS analysis.¹⁸² In 2015, Newman et al. reported a Phase I trial that examined the toxicity, plasma exposures, peak response, and dosing for the two compounds.¹⁸³ Additional clinical studies are underway for patients with advanced non-small cell lung cancer, breast cancer, bladder cancer, head and neck cancer, and solid tumors (ClinicalTrials.gov identifiers: NCT00978250—recruiting patients and NCT01534598—recruiting patients). Recently, the oral and intravenous pharmacokinetics of FdCyd 38 and tetrahydrouridine 39 were reported for cynomolgus monkeys and humans trial.¹⁸⁴

(4R)-2′-Deoxy-2′,2′-difluoro-3,4,5,6-tetrahydrouridine is a newer cytidine deaminase inhibitor being evaluated in preclinical studies.¹⁸⁵ The agent has an IC₅₀ of 0.4 μM for cytidine deaminase and exhibited enhanced acid stability at its N-glycosyl bond as compared to tetrahydrouridine 39. Rhesus monkeys cotreated with (4R)-2′-deoxy-2′,2′-difluoro-3,4,5,6-tetrahydrouridine and decitabine 14 had higher detectable levels of 14 in the serum.

5.1.1.6. Tegafur-Uracil, TS-1, Carmofur, and Flucytosine Biologies.

Tegafur 19, TS-1, carmofur 24, and flucytosine 31 are other examples of prodrugs of 5-FU 2 (Figure 9). The tegafururacil combination, TS-1 (a triple drug combo including tegafur 19, chloro dihydropyridine 20, and potassium oxonate 21), and carmofur (24, mifurol), are used globally in cancer treatments.¹⁸⁶ In 1983, tegafur–uracil (18 + 19) was approved for use in Japan. The orally bioavailable combination is given in a 4:1 mol ratio of tegafur 19 and uracil 18. Tegafur 19 is metabolized to 5-FU 2, which can be converted to the inactive dihydrofluorouracil mainly by dihydropyrimidine dehydrogenase in the liver.¹⁸⁷ The coadministration of uracil 18 inhibits this enzyme, thus helping maintain high levels of 5-FU 2 in the liver and in the circulation. Tegafur-uracil (18 + 19) is approved in Japan and Taiwan for various advanced gastrointestinal cancers. Patients are still being recruited for a Phase II clinical trial using metronomic therapy with tegafur–uracil (18 + 19) to treat head and neck cancer (ClinicalTrials.gov identifier: NCT00855881—recruiting patients). While a Phase II study in combination with sorafenib (kinase inhibitor) for advanced hepatocellular carcinoma has been terminated in Jan 2015, with no results posted (ClinicalTrials.gov identifier: NCT01539018—terminated in 2015—no results posted).

TS-1 includes tegafur 19, chlorodihydropyridine (gimeracil) 20, and potassium oxonate 21 used in a respective molar ratio of 1:0.4:1.¹⁸⁸ TS-1 (19 + 20 + 21) is an orally administered triple combination drug regiment, which was approved in Japan in 1999. Chlorodihydropyridine 20 is a more potent inhibitor of dihydropyrimidine dehydrogenase than uracil 18.^188,189 Potassium oxonate 21 may decrease gastrointestinal toxicity by selectively inhibiting 5-FU 2 activity within the intestinal lumen. TS-1 (19 + 20 + 21) is approved for use and is currently undergoing multiple clinical trials. Some of these studies include coadministration with cisplatin (DNA alkylating agent) for treatment of advanced non-small-cell lung cell cancer (ClinicalTrials.gov identifier: NCT01874678—completed status—with no results), a Phase II study for combination treatment of advanced hepatocellular carcinoma with the DNA alkylating agent oxaliplatin (ClinicalTrials.gov identifier: NCT01429961–unknown status–no updates for 2 years), and a Phase II study for monotherapy treatment of advanced metastatic breast cancer (ClinicalTrials.gov identifier: NCT01492543—unknown status—no updates for 2 years). In addition, Phase I clinical studies using combination therapy of TS-1 (19 + 20 + 21) and TAS-114, a first-in-class dual inhibitor of dUTPase (deoxyuridine-5′-triphosphate nucleotide hydro-lase) and dihydropyrimidine dehydrogenase,¹⁹⁰ is being conducted (ClinicalTrials.gov identifier: NCT01610479—active status—not recruiting).

Product marketing for carmofur 24 started in 1981. Carmofur 24 is a lipophilic-masked analog of 5-FU 2 that can be administered orally. The carbamoyl moiety of the drug is removed in vivo to release 5-FU 2. Carmofur 24 has been used to treat colorectal cancer in China, Japan, and Finland for many years.¹⁸⁶ However, the agent has been shown to induce delayed leukoencephalopathy, characterized by progressive damage to white matter in the brain with stroke-like symptoms.^191–193 There are currently no open clinical trials for this agent. A trial treating patients with stage II hepatocellular carcinoma was suspended prematurely, because 56% of the treated patients had unacceptable side effects and offered no survival advantage for certain cancers in stage 1 and 2 of the disease.¹⁹⁴ This may be a reason why carmofur 24 was never pursued for FDA-approval in the US.

Flucytosine 31 is another prodrug of 5-FU 2 in clinical trials. It is converted to 5-FU 2 by cytosine deaminase, which is not encoded by the human genome. Therefore, flucytosine 31 is used in combination with gene therapy of cancer cells in an attempt to have drug-specific targeting.¹⁹⁵ Engineered mesenchymal stromal cells have been examined to convert flucytosine 31 into its active compound,¹⁹⁶ and modified neural stem cells are currently being tested for the same purpose (ClinicalTrials.gov identifiers: NCT02015819—recruiting status, and NCT01172964—completed status—no results posted). Flucytosine 31 is currently in a Phase I/II clinical trial in combination with maltose and APS001F, a recombinant anaerobic bacteria Bifidobacterium engineered to express the cytosine deaminase gene, for treatment of solid tumors (ClinicalTrials.gov identifier: NCT01562626—recruiting status).

The progression of fluoropyrimidine analogs shows the importance of a biological observation–chemical modification interplay. The intravenously-administered 5-FU 2 and floxuridine 5 was followed by the orally-bioavailable carmofur 24 and flucytosine 31. The next step in the progression involved drug combinations TS-1 (19 + 20 + 21), tegafur–uracil (18 +19), and the recently FDA-approved TAS-102 (15 and 16), in order to decrease 5-FU 2 degradation. Further chemical modifications led to double (doxifluridine 23) and triple (capecitabine 10) prodrugs, which showed improvement in toxicity and pharmacokinetics profiles. Currently, NUC-3373 37 appears to overcome both innate and drug-derived resistance mechanisms. Newer analogs could be used as combination therapy with DNA damaging agents such as mitomycin-C and cisplatin, and immunotherapies—antiCTLA-4 and anti-PD-1, to improve patient survival rates.

5.1.2. Fluorouracil (5-FU), Fluorouracil Prodrugs, and Combinations Syntheses.

5.1.2.1. Fluorouracil (5-FU) Synthesis.

In 1957, a synthesis of 5-FU 2 was reported in which the compound was prepared from pseudourea salts and α-fluoro-β-keto ester enolates.^197–199 Twenty years later, Robins et al. reported a two-step synthesis of 5-FU 2 from uracil 18 (Scheme 5).²⁰⁰ Trifluoromethyl hypofluorite was used to introduce a fluorine atom into the uracil ring, and the resulting fluorinated adduct 66 underwent base-catalyzed elimination of MeOH, generating 5-fluoro-pyrimidine 2. An alternative synthesis used the cheap and readily available ortic acid (vitamin B13, 67) as the starting material.²⁰¹ Fluorination of 67 with fluorine gas and trifluoroacetic acid furnished acid 68, which in turn was decarboxylated to produce 5-FU 2 on a gram-scale.

Scheme 5. Synthesis of Fluorouracil 2^a.

^aReagents and conditions: (a) CF₃OF/CCl₃F, MeOH, −78 °C, 5 min; (b) Et₃N, MeOH, H₂O, rt, 11 min, 90% (over two steps); (c) F₂ (g), TFA, 0 °C to 10 °C, 3 h, 80%; (d) triethyleneglycol dimethyl ether, 130 °C to 200 °C, 20 min, 90%.

Baasner et al. reported a 100-g-scale preparation of 5-FU 2.²⁰² This procedure used selective fluorine/chlorine exchange and chlorine hydrogenolysis reactions (Scheme 6). Tetrafluoropyr-imidine 72, which can be prepared in four steps from urea 69 and diethylmalonate 70 via 2,4,5-trichloropyrimidine 71, was treated with HCl gas forming the dichlorinated intermediate 73. Dechlorination by selective palladium-catalyzed hydrogenolysis was followed by hydrolysis with aqueous sodium hydroxide to complete the synthesis.

Scheme 6. Synthesis of Fluorouracil 2^a.

^aReagents and conditions: (a) Na, MeOH; (b) POCl₃, PhNMe₂; (c) proton sponge hydrofluoride; (d) NaF, F₂, CFCl₃; (e) HCl (g), 160 °C, 4 h, 38%; (f) Pd/C, H₂, Et₃N, EtOAc, 3.5 h, 82%; (g) aq NaOH, 80 °C, 4 h, 93%.

5.1.2.2. Doxifluridine and Capecitabine Syntheses.

The first synthesis of capecitabine 10 was reported in the mid-1990s (Scheme 7).^150,203 Compound 79 is a key intermediate and is synthesized in six steps from d-ribose through a cyclization, mesylation, iodidation, reductive dehalogenation, hydrolysis, and cyclization/acetylation sequence. 5-Fluorocytosine, prepared by treating cytosine with trifluoromethyl hypofluorite in MeOH,²⁰⁴ was glycosylated with protected sugar 79 in good yield. N⁴-Amino acylation followed by selective deprotection of the 2′,3′-acetyl groups by treatment with aqueous sodium methoxide afforded crude capecitabine 10. Finally, pure capecitabine 10 was precipitated as a white solid from ethyl acetate/hexanes on a 100 gram-scale.

Scheme 7. Synthesis of Capecitabine 10^a.

^aReagents and conditions: (a) HCl, MeOH/Me₂CO, 78%; (b) MsCl, pyr; (c) NaI, DMF; (d) Pd/C, H₂, 27%; (e) aq HCl, 100 °C, 97%; (f) Ac₂O, pyr, 64%; (g) 5-fluorocytosine, HMDS, SnCl₄, DCM, 76%; (h) n-pentyl chloroformate, pyr; (i) aq NaOH, MeOH.

Two large-scale process syntheses of capecitabine 10 were reported by Hoffman-La Roche²⁰⁵ and Gore et al.²⁰⁶ Both syntheses use 5′-deoxy-5-fluorocytidine 85 as a key intermediate (Scheme 8). 5-Fluorocytidine 81, synthesized by treating protected cytidine with CF₃ OF/CCl₃ F,²⁰⁰ was protected, tosylated, and subsequently converted to 5′-iodo 83.²⁰⁷ Reductive dehalogenation followed by deprotection produced intermediate 85.²⁰⁸ This same sequence can also be employed to prepare doxifluridine 23, although using 5-fluorouridine as the starting material.

Scheme 8. Synthesis of Doxifluridine 23 and Capecitabine 10^a.

^aReagents and conditions: (a) 2,2-dimethoxypropane, TsOH, Me₂CO, rt, 2 h, 95%; (b) methyltriphenoxyphosphonium iodide, DMF, rt, 2.5 h, then MeOH, rt, 1 h, 74%; (c) Et₃N, H₂, Pd/C, MeOH, rt, 1.5 h, 93%; (d) TFA, 40 min, 78%; (e) CH₃(CH₂)₄OCOCl, pyr, DCM, −5 °C to rt, overnight, 92%; (f) NaOH, MeOH, −10 °C, 15 min, then concd HCl, 87%; (g) 87, THF, reflux, 66%.

The Hoffmann-La Roche synthesis converted 85 to the trisacylated intermediate 86, using 3 equiv of n-pentyl chloroformate in cold dichloromethane and pyridine. Selective removal of 2′,3′-ester groups was accomplished by treatment with aqueous sodium hydroxide in cold methanol, and precipitation from cold ethyl acetate afforded pure capecitabine 10.

The Gore et al. approach converted 85 directly to the product 10 by selective N⁴ acylation. In refluxing THF, intermediate 85 was reacted with pentyloxycarbonyl-1-hydroxybenzotriazole 87, which can be easily made from pentyl chloroformate and HOBt. This was followed by aqueous workup and precipitation to afford 10. Other acylating agents, such as pentyloxycarbonyl-imidazole, pentyloxycarbonyl-4-nitrophenoxy, and pentyloxycarbonyl-pentafluorophenoxy were also used, and produced similar yields.

The Jamison group used continuous flow chemistry to prepare capecitabine 10.²⁰⁹ The reaction exhibited impressive yields, “green” conditions, and very short reaction times (Scheme 9). The glycosylation reaction between triacetate intermediate 88 and silylated pyrimidine 90, while under BrØnsted acid-catalyzed conditions, was completed in 20 min. This procedure does not require aqueous workup. Mixing conditions were carefully monitored to form the N⁴-carbamate intermediate. Deprotection with a mixture of NaOH/MeOH/H₂O occurred at room temperature in only 2 min. The overall process produced milligrams of capecitabine 10 in less than 1 h and a 72% yield from starting materials 88 and 90.

Scheme 9.

Flow Synthesis of Capecitabine 10

In the same report,²⁰⁹ doxifluridine 23 was also synthesized using protected sugar 88, which was condensed with silylated 5-FU 2 in a glycosylation reaction. Deprotection of the resulting intermediate with sodium methoxide furnished the product. This one-flow, two-step process produced doxifluridine 23 in 89% yield in 10 min.

5.1.2.3. NUC-3373 Synthesis.

In 2011, McGuigan and co-workers reported the synthesis of NUC-3373 37.¹⁵⁶ The agent was prepared by reaction of phosphochloridate 93 (produced using l-alanine benzyl ester salt and 1-naphthyl-dichlorophosphate 92) with floxuridine 5 in THF in the presence of t-BuMgCl (Scheme 10). The synthesis produces milligrams of the final product and requires column chromatography to produce the purified mixture of diastereomers. Furthermore, no large-scale syntheses of NUC-3373 37 have been reported.

Scheme 10. Synthesis of NUC-3373 37^a.

^aReagents and conditions: (a) l-alanine benzyl ester salt, Et₃N, DCM, −78 °C, 1–3 h, 74%; (b) t-BuMgCl, THF, rt, 18 h, 8%.

5.1.2.4. Floxuridine, 5-Fluoro-2′-deoxycytidine, and Tetrahydrouridine Syntheses.

In the late 1950s, floxuridine 5 was originally prepared enzymatically and later converted to 5-fluoro-2′-deoxycytidine 38 via chemical synthesis.^210–212 Robin’s method (see Sections 5.1.2.1) was used to prepare both compounds in only two steps from 2′-deoxyuridine 94a and 2′-deoxycytidine 94b, with the key ring fluorination and sugar deprotection occurring in one pot (Scheme 11). This approach has the advantage of starting from a deoxynucleoside, which already has the needed stereochemistry fixed.

Scheme 11. Synthesis of Floxuridine 5, 5-Fluoro-2′-deoxycytidine 38, and Tetrahydrouridine 39^a.

^aReagents and conditions: (a) Ac₂O, DMAP, rt, 24 h, 88% (for 95a); Ac₂O, pyr, rt, 4.5 h, 99% (for 95b); (b) CF₃OF, CCl₃F, CHCl₃, −30 °C, evaporation, then Et₃N, MeOH, H₂O, rt, (55% for 5),(69% for 38); (c) Rh/Al, H₂ (g), H₂O, overnight; (d) H₂O, pH 6, 92% (over two steps); (e) Rh/Al, H₂ (g), NaOH, H₂O, 24 h, 45% of 99, 35% of 39; (f) NaBH₄, H₂O, freezer, overnight.

Surprisingly few syntheses of tetrahydrouridine 39 have been reported. In 1967, Hanze reported two approaches to the compound (Scheme 11).²¹³ Cytidine 96 was “overreduced” to tetrahydrocytidine 97 in one step using rhodium on aluminum in water, and subsequent hydrolysis afforded tetrahydrouridine 39. Uridine 98 was reduced to 5,6-dihydrouridine 99 also using rhodium and aluminum, and subsequent treatment with sodium borohydride furnished the target compound. The syntheses suffered from complex reaction mixtures, and no yields were reported.

Aoyama reported a gram-scale synthesis of floxuridine 5,²¹⁴ in which the key step was the condensation of 5-fluoro-2,4-bis(trimethylsilyloxy)pyrimidine 102 with 3,5-bis(O-p-chorobenzoyl)-2-deoxy-α-D-ribofuranosyl chloride 101. Chlorosugar 101 was made as the pure α anomer in three steps from 2′-deoxyribose (Scheme 12).^215,216 The condensation was performed in the presence of p-nitrophenol, which led, interestingly, to the stereoselective production of the desired β-anomer 103, whereas the combination of p-nitrophenol and pyridine as catalysts led to formation of the α-anomer. Recrystallization from acetic acid gave the pure β-anomer, which was deprotected with methanolic ammonia, concluding the synthesis.

Scheme 12. Synthesis of Floxuridine 5^a.

^aReagents and conditions: (a) AcCl, MeOH, 25 °C, 45 min, then pyr; (b) pyr, DMAP, 4-ClBzCl, 0 °C, 1 h then 25 °C, 12 h; (c) HOAc, HCl (g), 0 °C, 7–10 min, 82% (α-anomer crystallizes from the solvent); (d) p-nitrophenol, CHCl₃, 30 °C, 12 h, 92%; (e) NH₃/MeOH, 30 °C, 16 h, 81%.

5.1.2.5. Tegafur, Carmofur, and Flucytosine Syntheses.

In 1976, an original synthesis of tegafur 19 was reported that coupled 2-acetoxytetrahydrofuran with silylated 5-FU 2 in the presence of sodium iodide in 95% yield.^70,217 Five years later, Miyashita et al. reported an alternative preparation (Scheme 13).²¹⁸ The synthesis started with a one-pot condensation of urea 69, triethyl orthformate 104, and methyl malonate 105 under solvent-free conditions. The resulting methyl ureidomethylenemalonate 106 was then cyclized in methanolic sodium methoxide to give 5-methoxycarbonyluracil 107. Condensation of 107 with 2,3-dihydrofuran followed by ring fluorination and subsequent hydrolysis concluded the synthesis of 19.

Scheme 13. Synthesis of Tegafur 19 and Carmofur 24^a.

^aReagents and conditions: (a) 130 °C, 4 h, 69%; (b) NaOMe, MeOH, reflux, 10 min, 84%; (c) 2,3-dihydrofuran, pyr, 180 °C; (d) F₂, AcOH, rt; (e) 1 N NaOH, rt, 1 h, 62% (over two steps); (f) aq HCHO, 55 °C, 6 h, 82%; (g) KBrO₃, 80 °C, then H₂N(CH₂)₅CH₃, DCC, MeCN, 0 °C to rt, 6 h, 40%.

Carmofur 24 has been synthesized by treating 5-FU 2 with phosgene and hexylamine.^219,220 An alternative approach also utilized 5-FU 2 as starting material but used aqueous form-aldehyde to effect N-formylation. This was followed by oxidation with potassium bromate and finally condensation with hexylamine to conclude the synthesis of 24 (Scheme 13).²²¹

Flucytosine 31 was synthesized by reacting cytidine 110 with CF₃OF/CCl₃F in methanol for 5 min followed by addition of triethylamine in water (Scheme 14).²⁰⁰

Scheme 14. Synthesis of Flucytosine 31^a.

^aReagents and conditions: (a) CF₃OF, CCl₃F, MeOH, −78 °C to rt; (b) Et₃N, MeOH, H₂O, rt, 8 h, 85% (over two steps).

5.1.2.6. Gimeracil and Oteracil Syntheses.

In 1953, Kolder and Hertog reported a synthesis of the TS-1 additive gimeracil 20, which was completed in seven steps using 4-nitropyridine N-oxide as starting material.²²² Later, Yano et al. reported an alternative gram-scale synthesis (Scheme 15).²²³ The one-pot, three component condensation of malononitrile 111, 1,1,1-trimethoxyethane, and 1,1-dimethyoxytrimethylamine generated the dicyano intermediate 112, which was into 2(1H)-pyridinone 113.²²⁴ Selective chlorination of 113 was followed by acid-mediated demethylation, hydrolysis, and decarboxylation, to afford gimeracil 20. Interestingly, Xu et al. found that treatment of intermediate 113 with sulfuryl chloride resulted in dichloro 115 formation, which could still be converted to gimeracil 20 by treatment with sulfuric acid.²²⁵

Scheme 15. Synthesis of Gimeracil 20^a.

^aReagents and conditions: (a) CH₃C(OCH₃)₃, MeOH, then (CH₃)₂NHCH(OCH₃)₂, reflux, 92%; (b) aq AcOH, 130 °C, 2 h, 95%; (c) SO₂Cl₂, HOAc, 50 °C, 0.5 h, 91%; (d) 40% H₂SO₄, 130 °C, 4 h, 91%; (e) SO₂Cl₂, HOAc, 50 °C, 45 min, 86%; (f) 75% H₂ SO₄, 140 °C, 3 h, then NaOH, then pH 4–4.5, 89%.

Poje et al. reported a two-step, gram-scale preparation of the TS-1 additive oteracil 21 (Scheme 16).²²⁶ Iodine-mediated-oxidation of uric acid 116 produced dehydroallantoin 117 as the major product, and subsequent treatment with potassium hydroxide resulted in the rearranged product oteracil 21.²²⁷

Scheme 16. Synthesis of Oteracil 21^a.

^aReagents and conditions: (a) LiOH, I₂, H₂O, 5 °C, 5 min, then AcOH, 75%; (b) aq KOH, 20 min, rt, 82%.

5.1.3. Trifluorothymidine and Tipiracil Hydrochloride (TAS-102) Biology.

In 1964, Heidelberger and co-workers synthesized trifluorothymidine (TFT, 15) and demonstrated that it had promising anticancer activity.^228–231 Initial animal studies showed that TFT 15 was degraded to trifluorothymidine and catabolite 5-carboxyuracil, in the liver, spleen, and intestines.²³⁰ The agent showed reductions in some tumor sizes during a clinical trial but, like other antimetabolites, exhibited a plasma half-life of 15 min in cancer patients.²³² Research on TFT 15 was discontinued due to inadequate information on the pharmacokinetics and toxicity profile.^233,234

TFT 15 is a substrate for thymidine kinase, generating TFT-5′-monophosphate (an active intracellular metabolite). Similar to 5-FU-5′-monophosphate II, TFT-5′-monophosphate inhibits thymidylate synthase (Figure 8).²³⁵ In contrast to 5-FU-5′-monophosphate II, TFT-5′-monophosphate does not form a ternary complex with thymidylate synthase but inhibits it by binding to the active site of the enzyme.²³⁶ Furthermore, TFT-5′-monophosphate is a reversible inhibitor of thymidylate synthase, and removal of the agent results in rapid recovery of enzyme activity, whereas inhibition caused by formation of the 5-FU-5′-monophosphate II ternary complex has prolonged effects.²³⁷ TFT-5′-monophosphate can be further phosphorylated to its triphosphate form (another active metabolite) and then incorporated into DNA to promote single-strand breaks.²³⁸ DNA-incorporated TFT-5′-monophosphate is resistant to DNA glycosylase,²³⁹ and incorporated TFT-5′-monophosphate can promote DNA instability and double-stranded breaks.^240,241

The major drawback of TFT 15 monotherapy is its rapid degradation within the body, primarily by the action of thymidine phosphorylase, providing a natural TFT resistance mechanism.²⁴² However, coadministration of TFT 15 and a thymidine phosphorylase inhibitor improved the pharmacokinetic profile in animal models and antitumor activity in cell lines and animal models.²⁴³ Currently, Taiho Pharmaceuticals is developing TAS-102 (15 + 16) as a combination therapy that uses TFT 15 with tipiracil hydrochloride 16, a thymidine phosphorylase inhibitor with an IC of 35 nM,²⁴³ in a respective molar ratio of 2:1. This combination was shown to greatly decreased the biodegradation of TFT 15 in vitro.²⁴⁰ Interestingly, tipiracil hydrochloride 16 has also been shown to have antiangiogenic activity.²²⁸

TAS-102 (15 + 16) was FDA-approved in Sept 2015 for treatment in patients with colorectal cancer,²⁴⁴ and it is also currently being evaluated in clinical trials for the treatment of advanced solid tumors and metastatic refractory colorectal cancer²³³ (ClinicalTrials.gov identifier: NCT01607957—active status, not recruiting). A study was initiated evaluating TAS-102 (15 + 16) plus nivolumab in patients with microsatellite stable refractory metastatic colorectal cancer ClinicalTrials.gov identifier: NCT02860546—recruiting patients). In addition, TFT 15 has been pursued as an antiviral agent (registered as Viroptic) for use against the herpes simplex virus.^245,246

5.1.4. Trifluorothymidine and Tipiracil Hydrochloride (TAS-102) Syntheses.

Heidelberger et al. reported the first synthesis of TFT 15 (Scheme 17), featuring an enzyme-mediated transglycosylation of 5-trifluoromethyluracil 126 with thymidine.²²⁹ The trifluoro base 126 was prepared in an eight-step sequence. The synthesis began with treatment of trifluoroacetone 118 with hydrogen cyanide to produce cyanohydrin 119, and subsequent acetylation and ester pyrolysis afforded alkene 120. Treatment of 121 with anhydrous HBr and urea provided ureidoamide 123, which was then refluxed in acid to generate 5,6-dihydro-5-trifluoromethyluracil 124. This intermediate was subjected to a bromination–dehydrobromination sequence, furnishing 5-trifluoromethyluracil 126.

Scheme 17. Synthesis of Trifluorothymidine 15^a.

^aReagents and conditions: (a) NaCN, H₂SO₄, H₂O, 10 °C, 3 h, 99%; (b) Ac₂O, concd H₂SO₄, reflux, 1 h, 78%; (c) 500 °C, 4 h, 64%; (d) HBr (g), MeOH, 0 °C, 36 h, 82%; (e) (NH₂)₂CO, H₂O, 100 °C, 30 min, 28%; (f) aq HCl, reflux, 1 h, 58%; (g) Br₂/AcOH, reflux, 3 h, 85%; (h) DMF, 140 °C, 75 min, 80%; (i) thymidine, bactotryptone, NaCl, Escherichia coli B, 0.067 M phosphate buffer (pH 6.7), 37 °C, 3.5 h, 14%.

Komatsu et al. have reported a more modern and enzyme-free preparation of TFT 15.²⁴⁷ The preparation featured a green glycosylation reaction performed on the 100-g-scale using the TMS-protected analog of thymine 127 and the corresponding chlorosugar 101, made in two steps from 2′-deoxyribose. Conditions were sought to increase the rate of glycosylation (e.g., 101α to 128) and decrease the rate of chlorosugar anomerization (e.g., 101α to 101β), thereby increasing the yield of the β-glycosylation product 128 (Scheme 18).²⁴⁸

Scheme 18. Synthesis of Trifluorothymidine 15^a.

^aReagents and conditions: (a) AcCl, MeOH, 25 °C then pyr; (b) pyr, DMAP, 4-ClBzCl, 0 °C, 1 h, then 25 °C, 12 h; (c) HOAc, HCl (g), 0 °C, 7–10 min, 82% (over three steps); (d) 127, anisole, 50 °C, 3.5 h, 85% (β/α = 85.7:14.3); (e) NaOMe, MeOH, 4 °C, 3.5 h, then AcOBu, 97% (β/α = 99.92:0.08).

Nonpolar solvents were used to decrease chlorosugar anomerization, and temperature and reactant concentrations were explored empirically. Optimized large-scale conditions included condensation of silylated trifluorothymidine with an equimolar amount of chlorosugar 101 in minimal amounts of anisole (96% w/w of 101) at 50 °C, producing an 85:15 β/α anomeric mixture of 128/129 in 71% yield. Deprotection with NaOMe at 4 °C, precipitation and filtration, and finally washing with AcOBu (to remove residual methyl 4-chlorobenzoate) produced pure trifluorothymidine 15.

A concise 100-g-scale synthesis of tipiracil hydrochloride 16 has been reported (Scheme 19).²⁴⁹ Amination of 134 with methanolic ammonia led to the cyclized intermediate 2-iminopyrrolidine 135.²⁵⁰ In parallel, 6-chloromethyl uridine, which can be prepared in four steps from urea 69 and ethyl 2-acetoacetate 130 via intermediate 131,²⁵¹ was chlorinated at the 5-position to furnish the dichloromethyl uracil derivative 133.²⁵² Finally, condensation of 133 with 135 produced tipiracil hydrochloride 16.²⁴⁹

Scheme 19. Synthesis of Tipiracil Hydrochloride 16^a.

^aReagents and conditions: (a) SO₂Cl₂, AcOH, 50 °C, 2.5 h, 83%; (b) NH₃/MeOH, 120 °C, 10 h, 83%; (c) NaOEt, DMF, rt, 16 h, then aq HCl, 38%.

5.2. Azanucleosides

5.2.1. Decitabine, Azacytidine, CP-4200, and SGI-110 Biologies.

The azanucleosides, decitabine (Dacogen, 14), and azacytidine (Videza, 11), were first developed as cytostatic agents nearly half a century ago.^253,254 These cytostatic agents were later found to inhibit DNA methylation in human cell lines and were developed as epigenetic drugs.

Decitabine 14 is a 2′-deoxy-5-azanucleoside analog of cytidine that enters the cell by ENT-1.²⁵⁵ It is subsequently converted to decitabine-5′-monophosphate by 2′-deoxycytidine kinase. Further phosphorylation leads to the active metabolite decitabine-5′-triphosphate, which is a substrate for DNA polymerase α and is incorporated into the DNA.²⁵⁶ The incorporated decitabine-5′-monophosphate cannot be methylated, which can influence epigenetic gene regulation.²⁵⁷ DNA methylation usually involves the transfer of a methyl group from S-adenosyl-l-methionine, catalyzed by DNA methyl transferases (DNMTs), to the 5-position of a cytosine base within a cytosine-phosphate-guanosine dinucleotide.^258,259 Furthermore, DNA hypermethylation often silences tumor suppressor genes in hematopoietic malignancies,²⁶⁰ and inhibition of DNA methylation at regions of decitabine-5′-monophosphate incorporation in the DNA most likely restores the expression of some of these genes.^261–264

The ability of decitabine 14 to inhibit DNA methylation is often attributed to the formation of a complex between the azacytosine-guanine dinucleotide and DNMT1.²⁶⁵ This maypossibly occur via the covalent trapping paradigm (Figure 12).²⁶⁶ A covalent bond is formed between DNMT1 and the C6-position of the azacytosine base. Methylation of the nitrogen at position 5 of the base then occurs; however, β-elimination cannot occur due to the lack of a hydrogen atom at this position. As a consequence, DNMT1 remains covalently bound to the aza-base and is unable to continue its methylation activity. The covalent complex also triggers DNA damage signaling to result in DNMT1 degradation.²⁶⁷ A recent study, however, indicated that decitabine 14 could induce degradation of methyltransferases without the formation of the covalent complex, suggesting that the agent may work by additional mechanisms as well.²⁶⁸

Figure 12.

Covalent trapping paradigm: mechanism-based DNMT1 degradation by decitabine 14.

Decitabine 14 suffers from some chemical stability issues. The triazine ring of the drug is prone to hydrolytic opening and deformylation, whereas the sugar is susceptible to anomerization. Several groups have reported a half-life of decitabine 14 ranging from 3.5 to 21 h at physiological pH and temperature,²⁶⁹ whereas the in vivo half-life of decitabine 14 has been reported to be 15–20 min.²⁷⁰ Once the prodrug is metabolized to release decitabine 14 in the liver or cells, the high levels of cytidine deaminase protein will generate an inactive nucleoside byproduct,¹⁶⁹ which in turn limits the intracellular concentration and toxicity of decitabine 14. Finally if patients have reduced levels of 2′-deoxycytidine kinase activity, this may also contribute to natural decitabine resistance, as observed with decitabine.²⁷¹

At low concentrations, decitabine 14 has the ability to inhibit DNA methyltransferase 1, but it is cytotoxic at high concentrations in vitro.^272,273 Optimal treatment with decitabine 14 has been tested using continuous treatment at a low concentration in patients.²⁷⁴ However, the agent has poor oral bioavailability (currently administered by intravenous infusion) and is also a substrate for cytidine deaminase, which provides a natural resistance mechanism. Additionally, ex vivo studies suggest that reduction in phosphorylation by 2′-deoxycytidine kinase may also contribute to decitabine resistance.²⁷¹ Attempts to overcome these problems include the preparation of a decitabine mesylate salt,²⁷⁵ designed to increase the oral bioavailability, as well as coadministration of oral decitabine 14 with oral tetrahydrouridine 39 in patients.²⁷⁶

Azacytidine 11, a riboside analog of decitabine 14, is given by injection and enters cells by the uridine/cytidine transport system.²⁷⁷ After phosphorylation by uridine-cytidine kinase,^278,279 the active metabolite azacytidine-5′-triphosphate is incorporated into RNA to disrupt RNA metabolism and protein synthesis.²⁷⁹ Azacytidine-5′-diphosphate can also be reduced by ribonucleotide reductase to form 2′-deoxyazacytidine-5′-diphosphate and then follows the mechanism of action of decitabine 14.

After years of clinical research and dosage refinement, conditions were finally produced to effectively treat myelodysplastic syndrome, with azacytidine 11 and decitabine 14 being FDA-approved in 2004 and 2006, respectively.^280,281 Myelodysplastic syndrome, also known as preleukemia, is a group of related diseases originating in the bone marrow in which hematopoietic stem cells produce ineffective myeloid cells.²⁸² A genetic polymorphism in the cytidine deaminase gene (79A>C) and promoter depletion (–31delC) can lead to a rapid-deaminator phenotype and an increase in mRNA expression, leading to increased toxicity in patients treated with azacytidine 11.^283,284 Sorting patients based on cytidine deaminase genotypes remains difficult due to genotype to phenotype relationships, whereas developing a functional cytidine deaminase assay would be beneficial.²⁸⁵

Azacytidine 11 and decitabine 14 treatments are being evaluated in numerous clinical studies. An oral formulation of azacytidine (CC-486, Celgene) is currently in Phase I clinical trials (alone and in combination) for treatment of refractory solid tumors and Japanese myelodysplastic syndrome (ClinicalTrials. gov identifiers: NCT01478685—completed—no results posted, and NCT01908387—terminated status—no results posted). A Phase I/II Study of azacitidine 11 and with capecitabine 10 and oxaliplatin (DNA alkylating agent) is underway (ClinicalTrials. gov identifier: NCT01193517–active status–not recruiting). A Phase II study is examining the kinase inhibitor sorafenib with azacitidine 11 for primary response and secondary toxicity profile (ClinicalTrials.gov identifier: NCT02196857–actively recruiting patients). For decitabine 14 and tetrahydrouridine 39 clinical trials, one is currently recruiting patients— adjuvant oral decitabine 14 and tetrahydrouridine 39 with or without celecoxib in people undergoing pulmonary metastasectomy (ClinicalTrials.gov identifiers: NCT02839694–recruiting patients). Four additional studies have yet to start recruiting patients with refractory/relapsed lymphoid malignancies, pancreatic cancer, and non-small cell lung cancer using different combination treatments (ClinicalTrials.gov identifiers: NCT02846935–not yet recruiting patients, NCT02847000–not yet recruiting patients, NCT02795923–not yet recruiting patients, and NCT02664181–not yet recruiting patients).

Clavis Pharma is developing CP-4200, an elaidic acid derivative of azacytidine, which has strong epigenetic modulatory potency in human cancer cell lines (Figure 13).²⁸⁶ The aim of this drug is to circumvent therapy resistance in patients with MDS/AML due to decrease in drug uptake by hENT1, because the agent enters the cell by a hENT1-independent mechanism.²⁸⁷ CP-4200 has not yet moved past pre-clinical development. CP-4200 might have an acceptable bioavailability profile for oral application.

Figure 13.

Chemical structure of CP-4200.

Potential hurdles for CP-4200 include the following. First, upon prodrug metabolism to release azacytidine 11 in the liver or tumor cells, high levels of cytidine deaminase activity will generate an inactive nucleoside byproduct.²⁸⁸ Second, patients with reduced levels of 2′-deoxycytidine kinase activity may have some level of natural azacytidine resistance, as observed with decitabine 14.²⁷¹ Third, two other elaidic acid derivative prodrugs: elacytarabine (51, Phase I trial) and CP-4126 (49, two Phase III trials) have not been successful in clinical trials.

Astex Pharmaceuticals has developed the dinucleotide SGI-110 22, a second-generation prodrug of decitabine 14, which couples the agent with deoxyguanosine via a phosphodiester bond.²⁸⁹ It was synthesized in order to increase the in vivo exposure and efficacy of decitabine 14 by protecting it from cytidine deaminase activity.^290,291 SGI-110 22 is being evaluated as a hypomethylation agent with an epigenetic mechanism of action in tumors.^292,293 In a Phase I/II dose-escalating study, SGI-110 22 showed a favorable pharmacokinetic profile.²⁹⁴ The dinucleotide was efficiently converted to decitabine 14 in vivo. Moreover, it exhibited a longer apparent half-life, lower C_max, and prolonged plasma exposures of decitabine 14 than equivalent amounts of IV-administered decitabine 14. Furthermore, SGI-110 22 exhibited a greater stability, and lower doses of the agent achieved equal or better methylation inhibition than decitabine 14 monotherapy. SGI-110 22 is currently in several Phase I/II studies (as monotherapy and combination therapy) for treatment of ovarian cancer, metastatic colorectal cancer, myelodysplastic syndrome, acute myelogenous leukemia, and advanced hepatocellular carcinoma (ClinicalTrials.gov identifiers: NCT01696032–active status–not recruiting patients, NCT01896856–recruiting patients, NCT01261312–active status–not recruiting patients, NCT01752933–completed in 2015–no results posted, and NCT01966289–recruiting patients).

5.2.2. Azacytidine, Decitabine, and SGI-110 Syntheses.

In 1964, Piskala et al. reported the first synthesis of azacytidine 11 (Scheme 20).²⁹⁵ Chlorination of the 1′-position of tetraacetate riboside 136 followed by treatment with silver isocyanate produced 1-glycosyl isocyanate 137. Conversion of 137 to isobiuret 138 using 2-methylisourea was followed by cyclization with triethyl orthoformate and deprotection with methanolic ammonia to give azacytidine 11.

Scheme 20. Synthesis of Azacytidine 11^a.

^aReagents and conditions: (a) HCl (g), Ac₂O, Et₂O; (b) AgNCO, 60%; (c) 2-methylisourea, 68%; (d) HC(OC₂H₅)₃; (e) NH₃/MeOH, 40%.

In 2006, Ionescu and Blumbergs reported a large-scale synthesis of azacytidine (11, Scheme 21).²⁹⁶ Cyanoguanidine 139 was hydrolyzed and subsequently cyclized to produce the highly water-sensitive triazine 141. The glycosylation reaction between silylated triazine 142 and 1,2,3,5-tetra-O-acetyl-β-d-ribofuranose was performed in cold dichloromethane, with CF₃SO₃H added at 0 °C, followed by reaction warming to room temperature. Cooled solutions were used during the quenching/extraction step, which also minimized decomposition of the triazine base. Deprotection with a sodium methoxide/methanol solution, followed by recrystallization from DMSO/MeOH afforded azacytidine 11 on kilogram-scale. The last three steps, from glycosylation to recrystallization, were also performed in one pot starting with 5 g of the triazine, producing azacytidine 11 in 41% yield.

Scheme 21. Synthesis of Azacytidine 11^a.

^aReagents and conditions: (a) HCO₂H, 80 °C, 15 min, 85%; (b) (CH₃)₂NCH(OCH₃)₂, NaOMe, MeOH, 40–50 °C, 4 h, 90%; (c) HMDS, (NH₄)₂SO₄, CH₃CN, reflux, 4–8 h; (d) 1,2,3,5-tetra-O-acetyl-β-d-ribofuranose, CF₃SO₃H, EtOAc, 45 °C, 30 min, then HCl, DCM, rt, 45 min; (e) n-BuNH₂, MeOH, 65 °C, 1 h, 49% (over three steps).

An additional process synthesis of azacytidine 11 using triflic acid in the glycosylation was reported.²⁹⁷ Detailed empirical studies of the reaction found that a 1:1.2:0.95 ratio (silylated triazine 142:triflic acid:1,2,3,5-tetra-O-acetyl-β-d-ribofuranose) produced the highest yield. Exposure times to water or acid during the workup were minimized in order to protect the triazine base of intermediate 143. Further empirical studies found optimized conditions for acetyl group deprotection of 143, which involved n-butylamine in refluxing methanol. Recrystallization of the crude azacytidine 11 from DMSO/toluene gave the pure product in 99.9% purity on a kilogram-scale.

In 1964, the first synthesis of decitabine 14 was reported, which used a multistep sequence similar to the preparation of azacytidine 11 shown in Scheme 20. However, no yield was reported, as the α/β-anomers were not separated.²⁹⁸ Several years later, Winkley and Robins reported a synthesis of pure β decitabine (14, Scheme 22).²⁹⁹ Triacetyl protected sugar 144 was converted to 3′,5′-di-O-acetyl-2′-deoxy-d-ribofuranosyl chloride and then treated with the silylated 5-azacytosine 142, forming 145 as a mixture of diastereomers. Deprotection with ethanolic ammonia followed by α/β anomer separation, involving fractional crystallization followed by preparative layer chromatography on silica gel, gave decitabine 14 in 7% yield. Unfortunately, the undesired α-anomer was the predominant product, forming in 52% yield.

Scheme 22. Early Synthesis of Decitabine 14^a.

^aReagents and conditions: (a) AcCl, HCl, Et₂O, 0 °C; (b) bis(trimethylsilyl)-5-azacytosine 142, CH₃CN, 10%; (c) NH₃/EtOH; (d) anomer separation, 7%.

Two difficulties for large-scale synthesis of decitabine 14 are the lack of stability of the triazine base and the absence of anomeric selectivity during the glycosylation step. Indeed, when conjugated to a carbohydrate, 5-azacytosine is sensitive to hydrolytic decomposition under acidic, basic, and neutral conditions.³⁰⁰ Furthermore, coupling of the silylated triazine with a 2′-deoxyribosugar generally leads to a 1:1 mixture of α/β anomers.

Henschke et al. recently performed detailed studies under various conditions of the glycosylation step in attempts to improve anomeric selectivity (Scheme 23).³⁰¹ Solvent choice, reaction temperature, and ratio of the silylated triazine to the protected ribosugar were all found to affect anomeric selectivity, producing glycosylation products ranging from an anomeric ratio of 1:1 to 1:3 (α:β). Optimal large-scale conditions included a 1:1 molar ratio of the protected sugar to silylated triazine 142 with the addition of 1.05 equiv of TMSOTf in cold DCM. Kilograms of protected decitabine 147 were produced in a 1:2.7 α/β ratio in up to 55% yield. Furthermore, it was discovered that immediate quenching of the reaction with an organic amine (usually a primary amine) was needed in order to preserve the α/β-ratio. Since they had observed isomerization of the β-anomer under different conditions, the authors hypothesized that prolonged reaction standing led to the resultant decrease of the β-anomer. Moreover, the decrease in the resultant was a result of its anomerization to the α-product and not due to decomposition of the triazine base of protected decitabine. The synthesis of decitabine 14 was concluded by deprotection of a finely milled 147; high quality of this intermediate was required for reaction success, in methanol with or without sodium methoxide. Subsequent filtration and recrystallization from DMSO and methanol afforded kilograms of API-grade decitabine 14.

Scheme 23. Kilogram-Scale Synthesis of Decitabine 14^a.

^aReagents and conditions: (a) bis(trimethylsilyl)-5-azacytosine 142, TMSOTf, DCM, cold, 55%; (b) NaOMe/MeOH, 65%.

Kolla et al. have recently reported a process synthesis of decitabine (14, Scheme 24).³⁰² Bis-acetyl protected sugar 149, formed by treating 2′-deoxy-α/β-d-ribose 100 with acetyl chloride in methanol followed by acetyl chloride in pyridine/DCM, was condensed with silylated 5-azacytosine in the presence of TMSOTf, giving a 1:1 α/β mixture of acylated 150. Deprotection with methanolic ammonia treatment followed by crystallization afforded decitabine 14 in an HPLC purity of 90–99%. The resulting solid was dissolved in DMSO, filtered, washed with MeOH/EtOAc, and suction dried to produce crystalline decitabine 14 at 99.8% purity on a gram-scale.

Scheme 24. Synthesis of Decitabine 14^a.

^aReagents and conditions: (a) AcCl/MeOH, 99%; (b) AcCl/pyr/DCM, 78%; (c) 5-azacytosine, HMDS, NH₄SO₄, TMSOTf, 60% (β/α = 1:1); (d) NH₃/MeOH, 34%.

Recently, Redkar reported an approach to SGI-110 22 which increased the scale of compound production from 500 mg to 1 kg (Scheme 25).³⁰³ Decitabine 14 was selectively protected at the N⁴- and 5′-positions by treatment with DMF dimethyl acetal and then vinyl acetate in the presence of Lipozyme RM IM. Conversion to the 3′-phosphoramidite 151 was followed by coupling with 3′,N²-protected guanosine 152, and subsequent oxidation by tert-butyl hydrogen peroxide furnished dinucleotide 154. Deprotection in methanolic ammonia followed by hydrolysis produced SGI-110 22, which precipitated as a sodium salt.

Scheme 25. Synthesis of SGI-110 22^a.

^aReagents and conditions: (a) DMF, dimethylacetal; (b) vinyl acetate, Lipozyme RM IM, CH₃CN, dioxane; (c) [(iPr)₂N]POCH₂CH₂CN, DCM; (d) DCM; (e) tBuOOH; (f) NH₃/MeOH; (g) NaOAc, H₂O, EtOH.

5.3. Ribosugar-Modified Cytidine Analogs

5.3.1. Gemcitabine and Gemcitabine Prodrugs Biologies.

5.3.1.1. Gemcitabine Biology.

Hertel and colleagues at Eli Lilly originally synthesized gemcitabine (2′,2′-difluoro-2′-deoxycytidine) hydrochloride 9 as an antiviral nucleoside analog agent.³⁰⁴ However, it was also discovered to be very effective at inhibiting cancer proliferation in cell culture.³⁰⁴ Gemcitabine was originally screened against hematological cancers, but was also surprisingly shown to have activity in solid tumors. The agent is the only FDA-approved cytidine analog to be effective in solid tumors.^305,306

Gemcitabine is imported into the cell by hENT-1,³⁰⁷ and to a lesser extent by hCNT-1 and hCNT-3.^308–310 The agent is initially phosphorylated by 2′-deoxycytidine kinase, which is the rate-limiting step in its activation. Subsequent phosphorylations lead to the active metabolites: gemcitabine-5′-triphosphate and gemcitabine-5′-diphosphate. Gemcitabine-5′-triphosphate inhibits DNA synthesis by being incorporated into the newly synthesized DNA strand. Once incorporated into DNA, one additional nucleotide is inserted past gemcitabine-5′-monophosphate before chain elongation is inhibited.^311,312 This “masked chain termination” cannot be detected by exonucleases and eventually causes cell cycle arrest primarily in the S phase, resulting in apoptosis.³¹³ Alternatively, gemcitabine-5′-triphosphate can also be incorporated into DNA without chain elongation termination,^17,314 and is recognized by a DNA-dependent protein kinase/p53 protein complex, which coincides with cellular apoptosis.²⁷ Gemcitabine was also shown to have anticancer activity by being incorporated into RNA.³¹⁵ Furthermore, synergistic interaction of gemcitabine and sphingomyelin can induce ceramide-mediated apoptosis through Fas-induced cytotoxicity.³¹⁶

Gemcitabine has self-potentiation mechanisms.^317,318 Gemcitabine-5′-diphosphate is a major active metabolite that inhibits ribonucleotide reductase, leading to one self-potentiation mechanism. Blocking ribonucleotide reductase activity decreases the natural deoxyribonucleotide pools, promoting more gemcitabine-5′-triphosphate to be incorporated into genomic DNA by DNA polymerase.^31,319,320 Furthermore, inhibition of ribonucleotide reductase leads to an indirect self-potentiation mechanism involving the regulation of 2′-deoxycytidine kinase. The enzyme is inhibited by a high dCTP concentration by a negative feedback loop regulation, and gemcitabine-induced ribonucleotide reductase inhibition decreases the dCTP and dATP concentrations.³²¹ This in turn leads to an increase in the activity of 2′-deoxycytidine kinase and more gemcitabine-5′-monophosphe accumulation in the cell. Gemcitabine-5′-triphosphate induces another direct potentiation mechanism involving the inhibition of deoxycytidylate, an enzyme that deaminates deoxycytidine-5′-monophosphate as well as gemcitabine-5′-monophosphate, preventing deamination of gemcitabine-5′-monophosphate into 2′,2′-difluoro-2′-deoxyuridine-5′-monophosphate, an inactive metabolite.³²²

There are several mechanisms of resistance for gemcitabine reported from tissue culture studies using cell lines, and some (but not all) of the proteins and pathways being discovered include the following. The deaminated byproduct of gemcitabine, 2′,2′-difluoro-5′-deoxyuridine, is a competitive inhibitor for its uptake by the hENT transporter,³²³ and mutations in nucleotide transporters and 2′-deoxycytidine kinase prevent the uptake and initial phosphorylation of gemcitabine in the cell.^60,324 Intracellularly, gemcitabine-5′-monophosphate can be deaminated to form 2′,2′-difluoro-2′-deoxycytidine-5′-monophosphate, and high gene expression of NT5C3 has been correlated with lower levels of phosphorylated gemcitabine metabolites.³²⁵ Furthermore, upregulation of ABCC5 and hENT1 mRNA or protein was detected in gemcitabine resistance cell lines but has not been directly linked to promoting drug resistance in patients.³²⁶ Additional studies have shown that histone methyltransferase G9a,³²⁷ p38MAPK,³²⁸ and Nutlin-3³²⁹ might be correlated with gemcitabine resistance. Furthermore, gemcitabine resistance can arise from changes in high expression of drug efflux pumps and in nucleotide metabolism enzymes, inactivation of the apoptosis pathway, changes in miRNA expression, and modulation of Hedgehog, Wnt, and Notch pathways.³³⁰

In patients, gemcitabine deamination appears to be a major mechanism limiting drug efficacy. Gemcitabine can be deaminated at the N⁴ position of the cytidine base in the serum to the inactive 2′,2′-difluoro-5′-deoxyuridine metabolite by cytidine deaminase,³³¹ or gemcitabine-5′-monophoshate can be deaminated by deoxycytdine deaminase in the cell. This metabolism is responsible for the very short half-life of the drug, thus limiting availability at the tumor.³³² Changes in ribonucleotide reductase subunit protein levels or polymorphisms, which regulate dNTP cellular concentrations, have been demonstrated.^333–335 Furthermore, collective cell line data suggest that modulation of mRNA expression, cell survival pathways, and tumor suppressor genes all might contribute to gemcitabine resistance. These mechanisms for gemcitabine resistance will need to be validated in the future using ex vivo models to determine their importance in patient populations. Since gemcitabine resistance is polygenic, combination treatments with alkylating agents, topoisomerase inhibitors, mitotic inhibitors, and biologicals (such as anti-PD-1 and anti-CTLA-4) appear to be a logical choice for new investigatory treatments.

Gemcitabine hydrochloride 9 is currently being used to treat non-small cell lung, pancreatic, breast, bladder, and ovarian cancers.^336–338 The agent is also being used to treat pancreatic cancer; however, treatment with gemcitabine hydrochloride 9 monotherapy showed only a modest benefit.^339,340 Currently, various clinical trials studying using gemcitabine hydrochloride 9 in combination therapy are underway. Some of these trials include: gemcitabine-cisplatin (DNA alkylating agent) combination therapy for advanced non-small-cell lung cancer,³⁴¹ Hodgkin’s lymphoma, and non-Hodgkin’s lymphoma,³⁴² gemcitabine-erlotinib (tyrosine kinase inhibitor) regimen to treat locally advanced and metastatic pancreatic cancers,³⁴³ and gemcitabine-oxaliplatin (DNA alkylating agent) treatment for biliary adenocarcinoma,³⁴⁴ pancreatic cancer,³⁴⁵ hepatocellular cancer,³⁴⁶ testicular cancer,³⁴⁷ and epithelial ovarian cancer.³⁴⁸ Additional combination treatments of gemcitabine with necitumumab (a monoclonal antibody which binds the epidermal growth factor receptor) and radiation treatments are under investigation (ClinicalTrials.gov identifiers: NCT01606748–active status–not recruiting patients and NCT02254681–recruiting patients).

5.3.1.2. LY2334737 Biology.

As an N⁴-valproic acid prodrug, LY2334737 52 was designed to improve the oral bioavailability of gemcitabine hydrochloride 9, primarily by blocking the site of deamination and decreasing first-pass metabolism.^349,350 The prodrug passes through the intestinal epithelium and enters the systemic circulation largely intact,³⁵⁰ thereby reducing gut exposure to gemcitabine. LY2334737 52 is hydrolyzed by slow systemic cleavage via the actions of carboxyesterase 2, liberating gemcitabine and valproic acid, and leads to prolonged gemcitabine exposure in vitro.³⁵¹ Preclinical studies showed that low-dosing of LY2334737 52 was efficient for human colon and lung tumor xenograft models.³⁵² Moreover, elevated carboxyesterase 2 activity and ENT1 expression may enhance LY2334737 tumor response.

In a clinical study to determine the maximum tolerated dose and dose-limiting toxicities of LY2334737 52, stable disease was achieved in 22 of the 65 patients with advanced solid tumors.³⁵³ In 2012, a Phase I clinical trial was conducted in persons with solid and metastatic tumors (ClinicalTrials.gov identifier: NCT01648764–completed in 2014–no study results were posted). However, in 2013 Eli Lilly discontinued the development of LY2334737 52,³⁵⁴ due to results from a Japanese study that showed hepatic toxicities in patients.³⁵⁵

5.3.1.3. CP-4126 Biology.

CP-4126 (49, CO-1.01) is an elaidic acid ester derivative of gemcitabine developed by Clavis Pharma. The lipophilic side chain was installed to increase the ability of gemcitabine to diffuse through the plasma membrane. This modification makes CP-4126 49 a transporter-independent analog as compared to gemcitabine, which enters the cell primarily by hENT1.³⁵⁶ Once in the cell, cleavage to gemcitabine is believed to occur via the action of carboxyesterases, and subsequent phosphorylation to gemcibine-5′-monophosphate is performed by deoxycytidne kinase.³⁵⁷ Furthermore, resistance studies showed that CP-4126 49 and gemcitabine can arise by down regulation of 2′-deoxycytidine kinase in vitro.³²⁴

CP-4126 49 has been assessed in several Phase I/II clinical trials for safety and efficacy as monotherapy or combination therapy,³⁵⁷ involving non-small-cell lung cancer, metastatic pancreatic adenocarcinoma, and other advance solid tumors. The results demonstrated no marked difference in overall survivals of individuals treated with CP-4126 49 versus gemcitabine hydrochloride 9 having pancreatic ductal adenocarcinoma and a low hENT1 protein expression level.³⁵⁸ Two Phase I clinical trials have been completed (ClinicalTrials.gov identifiers: NCT01392976–completed in 2013–results were published). For one clinical trial, CP-4126 49 was poorly absorbed and was subject to metabolism before systemic exposure. Overall, this led to further testing of CP-4126 49 in patients by intravenous administration.³⁵⁹ Another Phase I clinical trial had CP-4126 49 given intravenously and reported positive results (ClinicalTrials.gov identifiers: NCT00778128–completed in 2010–results were published).³⁶⁰ Another Phase I trial testing CP-4126 49 and cisplatin was stopped (ClinicalTrials.gov identifiers: NCT01641575–terminated in 2013–no results posted). As a result of these studies, Clavis suspended development of CP-4126 49.

We have not covered all the potential prodrug combinations or technologies under development for gemcitabine. Other unique modifications of the agent include mesoporous silica nanoparticles,³⁶¹ amino acid monoester prodrugs,³⁶² Hoechst conjugates,³⁶³ and further base modifications.³⁶⁴

5.3.1.4. NUC-1031 Biology.

The ProTide concept has been widely used in the development of antiviral and anticancer nucleoside-5′-monophosphate prodrugs.^104,365 Protides were designed as protected nucleotide-5′-monophosphate isomers to increase cellular penetration as well as bypass the often rate-limiting monophosphorylation of the nucleoside analog.¹⁰⁴ Furthermore, they are also believed to improve the delivery of the parent compound to the liver.³⁶⁶ NUC-1031 (Acelarin, 32) is a 5′-phosphoramidate prodrug of gemcitabine.³⁶⁷ This agent is able to overcome gemcitabine resistance mechanisms: decreased uptake by cellular transporters, decreased activity of 2′-deoxycytidine kinase (by mutation), and cytosine deamination.³⁶⁸ However, it would still be sensitive to gemcitabine-5′-monophosphate deamination by deoxycytidine deaminase. Like gemcitabine, NUC-1031 32, is administered by intravenous injection.

A Phase I study in people with advanced solid tumors has been launched in 2012.^368,369 The agent could be given at four times the maximum dose of gemcitabine, and over half the patients achieved stable disease, with a few patients having a reduction in tumor size. In addition, administration of NUC-1031 32 produced a 13-fold increase in intracellular gemcitabine concentration as compared to gemcitabine intravenous injection. An additional study funded by NuCana BioMed Limited examining combination treatment of NUC-1031 32 with carboplatin (DNA alkylating agent) for ovarian cancer is planned (ClinicalTrials.gov identifiers: NCT01621854–completed June 2015–no results posted, and NCT02303912–recruiting patients).

5.3.2. Gemcitabine and Gemcitabine Prodrugs Syntheses.

5.3.2.1. Gemcitabine Syntheses.

Hertel and co-workers reported the discovery synthesis of gemcitabine hydrochloride 9 (Scheme 26).³⁷⁰ It commenced with a Reformatsky reaction involving ethyl bromodifluoroacetate and aldehyde 155, which is easily prepared by isopropylidene protection and subsequent oxidative cleavage of d-mannitol, forming 156 as a mixture of diastereomers (3:1 anti/syn) which were separated by HPLC. Acid-mediated deprotection and cyclization were followed by 3′,5′-O-silylation, affording the protected lactone 157. Reduction, mesylation, and glycosylation under Vorbrüggen conditions produced protected sugar 158 as a 4:1 mixture of α/β anomers. Deprotection allowed for reverse phase HPLC separation of the anomers. The unfortunate α/β anomeric ratio produced under these conditions gave gemcitabine hydrochloride 9 in only 10% yield from intermediate 158.

Scheme 26. Synthesis of Gemcitabine Hydrochloride 9^a.

^aReagents and conditions: (a) BrCF₂CO₂Et, Zn, THF, Et₂O, 87% (3:1 anti/syn); (b) Dowex 50, MeOH, H₂O; (c) TBDMSOTf, lutidine, DCM, 86% (over two steps); (d) DIBAL-H; (e) MsCl, Et₃N, DCM, 71% (over two steps); (f) 2,4-bis(trimethylsilyl)cytosine, TMSOTf, DCE, reflux; (g) AG 50W-X8 resin, MeOH; (h) reverse phase HPLC, 10% (over two steps).

In 2010, Chang et al. reported a large-scale synthesis of gemcitabine 163 using 2,2-fluoro-α-ribofuranosyl bromide 162 as a key intermediate to increase the β-anomeric selectivity in the glycosylation reaction (Scheme 27).³⁷¹ The stereoisomers of 156 were protected with p-phenylbenzyl chloride and subjected to ester hydrolysis under basic conditions. Fortunately, the reaction volume reduction led to selective precipitation of the desired potassium erythro-pentonate 159, and filtration furnished the pure diastereomer, representing a marked increase in efficiency when compared to the laborious diastereomeric separations of other syntheses (e.g., HPLC purification in the Hertel method). Deprotection, cyclization, and 5′-O-benzoylation of intermediate 159 were followed by lactone reduction and phosphorylation of the crude anomeric mixture, directly furnishing intermediate 161 (1:10.8 α/β ratio). Subsequent bromination led to a 10.8:1 α/β mixture, and recrystallization from isopropanol afforded the pure α-anomer 162, with the phenylbenzoyl protecting group being necessary for recrystallization.

Scheme 27. Synthesis of Gemcitabine 163^a.

^aReagents and conditions: (a) PhBzCl, Et₃N, DCM; (b) K₂CO₃, H₂O, THF, MeOH, 67%; (c) HCl, CH₃CN, reflux; (d) BzCl/Pyr, EtOAc, rt, 72%; (e) LiAl(O-tBu)₃H, THF; (f) ClP(O)(OPh)₂, Et₃N, PhMe, rt, 77%; (g) HBr/HOAc, rt, 82%; (h) 2,4-bis(trimethylsilyl)cytosine, octane/heptane, Ph₂O, 140–150 °C; (i) NH₃/MeOH, H₂O, rt, 65%.

In order to maximize the S_N2-displacement of α-bromide 162, the glycosylation conditions were optimized. Therefore, non-polar solvents were used, and in situ generated TMSBr was removed via continuous distillation, in an attempt to decrease anomerization of 162 or formation of the oxocarbenium ion. Indeed, this led to an improved β/α anomeric mixture of 5.5:1. Deprotection with methanolic ammonia, followed by isolation of the pure β-diastereomer by simple extraction, evaporation, and crystallization from water, produced the hemi- or dihydrate of gemcitabine, depending on workup conditions.

Jiang et al. also recently reported a process preparation for gemcitabine hydrochloride 9 on a 100-g-scale (Scheme 28).³⁷² A diastereomeric mixture of difluoro lactone 164 was acylated with cinnamoyl chloride, and when the solution was cooled, stereospecific crystallization afforded the desired lactone 165. Reduction and tosylation gave a 1:1 mixture of anomeric tosylates 166, which was found to crystallize in ethyl acetate/petroleum ether. Subsequent glycosylation of the anomeric mixture 166 with N-acetyl cytosine in the presence of TMSOTf, followed by direct treatment of the reaction mixture with sodium carbonate, allowed for the precipitation of the undesired α-nucleoside analogue. Deprotection of β-isomer 167 with methanolic ammonia at room temperature gave crude gemcitabine 163, which was acidified with HCl and crystallized in acetone/H₂O, affording the target salt in 99.9% purity.

Scheme 28. Synthesis of Gemcitabine Hydrochloride 9 and LY2334737 52^a.

^aReagents and conditions: (a) Dowex 50W-X12 resin, MeOH, H₂O, rt, 4 days, 94%; (b) (E)-cinnamoyl chloride, pyr, EtOAc, 30 °C, 3 h, 43%; (c) LiAl(O-tBu)₃H, THF, −10 °C, 2 h, then TsCl, Et₃N, PhMe, −10 °C, 5 h, 62%; (d) N-acetyl cytosine, (TMS)₂NH, TMSOTf, DCE, reflux, 12 h, then 5% NaHCO₃, 47%; (e) NH₃/MeOH, rt, overnight, then HCl, Me₂CO, rt, 12 h, 80%; (f) 2-propylpentanoic acid, EDC, HOBt, NMM, DMF, DMSO, 55 °C, 17 h, 95%.

5.3.2.2. LY2334737, CP-4126, and NUC-1301 Syntheses.

The gemcitabine prodrug LY2334737 52 was formed by directly coupling valproic acid with gemcitabine hydrochloride 9, using peptide coupling conditions (EDC/HOBt/NMM).³⁵⁰ Prodrug LY2334737 52 is not crystalline; therefore, it was purified as a p-toluenesulfonic acid (pTSA) salt co-crystal (Scheme 28).

CP-4126 49 was formed by direct esterification of the 5′-OH of gemcitabine 163 using elaidic acid chloride as reagent (Scheme 29).³⁷³ NUC-1301 32 was prepared by reaction of phosphochloridate 169, which was produced by treating l-alanine benzyl ester hydrochloric salt 168 with phenyl-dichlorophosphate 92, with gemcitabine 163 in THF/pyridine in the presence of N-methyl imidazole.³⁷⁴

Scheme 29. Synthesis of CP-4126 49 and NUC-1031 32^a.

^aReagents and conditions: (a) Et₃N, DCM, −80 °C, 1 h, 98%, (b) NMI, pyr, THF, −80 °C to rt, 2 h, 16%; (c) HCl (g), DMF, elaidic acid chloride, rt, 12 h, 30%.

An improved synthesis for NUC-1031 32 has recently been reported, which affords the agent on a 20-gram-scale.³⁷⁵ Boc protection of the 3′-OH and N⁴-amino group of gemcitabine 163 was followed by reaction with phenyl-dichlorophosphate 92 and L-alanine benzyl ester hydrochloride 168, which installed the protide. Subsequent Boc deprotection completed the synthesis.

5.3.3. Sapacitabine Biology.

In the early 1990s, Matsuda et al. published the synthesis and anticancer activity of 2′-C-cyano-2′-deoxy-1-β-D-arabino-pentofuranosylcytosine (CNDAC) along with the rationale for the structure of the agent.³⁷⁶ Since a cyanoethyl group beta to a phosphate diester of a nucleoside was reported to undergo β-elimination under alkaline conditions,³⁷⁷ it was hypothesized that a 2′-β-C-cyano nucleoside when incorporated into DNA could lead to β-elimination of the 3′-phosphodiester moiety and produce DNA strand breaks (Figure 14). This mechanism of anticancer activity would be similar to that of radiation therapy, which was shown to produce similar DNA strand breaks that were believed to result in tumor cell death.^378,379

Figure 14.

CNDAC structure and hypothesized mechanism of DNA strand break with incorporated CNDAC.

Sapacitabine 25 (Cyclacel Pharmaceuticals) is an N⁴-palmitoyl derivatized analog of CNDAC and is orally bioavailable. It is primarily metabolized in the plasma, gut, and/or liver to CNDAC, and eventually transported into cells.³⁸⁰ CNDAC is phosphorylated by 2′-deoxycytidine kinase and further phosphorylated by other cellular kinases to generate the triphosphate form.³⁸¹ DNA polymerases incorporate the CNDAC-5′-triphosphate into the growing DNA strands, and further elongation of the chain is sluggish due to the steric effects of CNDAC analog.²⁶ Upon incorporation of CNDAC-5′-monophosphate into DNA, β-elimination of the 3′-phosphodiester moiety can result in a severed DNA strand that is terminated with 2′-C-cyano-2′,3′-didehydro-2′,3′-dideoxycytidine-5′-monophosphate (Figure 14).³⁸² This represents a unique anticancer mechanism of a nucleoside derivative, and results in cell cycle arrest at the G₂ phase, which is different from the majority of other anticancer nucleoside analogs that cause arrest in the S phase.³⁸³

Sapacitabine 25 is a weak substrate for cytidine deaminase, whereas CNDAC is a substrate and is converted into the inactive uracil derivative: 2′-C-cyano-2′-deoxy-1-β-d-arabino-(pentofuranosyl)uridine. In addition, unlike other nucleoside analogs in their corresponding di- or triphosphates forms, CNDAC-5′-diphosphate does not effectively inhibit ribonucleotide reductase, and CNDAC-5′-triphosphate does not act as a feedback inhibitor of 2′-deoxycytidine kinase.²⁶

Cells deficient in nucleotide excision repair (NER) of endonuclease XPF are 4- to 5-fold more sensitive to CNDAC exposure,³⁸⁴ suggesting that the NER pathway is used to excise the agent from the DNA strand. If DNA single-strand breaks caused by CNDAC exposure are not repaired, they become more lethal DNA double-stranded breaks during the following round of DNA replication.³⁸⁵ These CNDAC-induced DNA double-stranded breaks can occur when the replication fork encounters a DNA single-stranded break or after collapse of a stalled replication fork,³⁸⁶ which are repaired primarily by homologous recombination. Cells with defects or mutations in the homologous recombination pathway (ATM, RAD51, Brac2, Xrcc3) are up to 100-fold more sensitive to CNDAC exposure.³⁸⁷ In addition to DNA incorporation, CNDAC exposure was also shown to inhibit RNA synthesis to a limited extent in vitro.²⁶

Sapacitabine 25 is in several clinical trials (Phase I–III) either as a monotherapy or in combination with other drugs for treatment of acute myelogenous leukemia, chronic lymphocytic leukemia, non-small cell lung cancer, and some solid tumors.^349,387,388 However, it failed to achieve end point efficacy in a Phase III study using oral sapacitabine 25 in elderly patients with newly diagnosed acute myelogenous leukemia (ClinicalTrials.gov identifier: NCT01303796–unknown, >2 years since reported). In 2013, Cyclacel Pharmaceuticals announced that sapacitabine 25 had activity against 75% of primary ovarian cancer samples isolated from individuals. DFP-10917 (26, intravenously administered hydrochloride salt form of CNDAC) is in Phase I/II studies to treat relapsed or refractory acute leukemia (ClinicalTrials.gov identifier: NCT01702155–recruiting patients).

5.3.4. Sapacitabine Synthesis.

The discovery synthesis of CNDAC is shown in Scheme 30.^376,389 N⁴-Acylation of cytidine 96 followed by 3′,5′-TIPS protection and 2′-OH oxidation provided ketosugar 170. Treatment of 170 with sodium cyanide followed by reaction with phenyl chlorothionocarbonate produced an epimeric mixture of 2′-cyanohydrins 171. Subsequent deoxygenation under standard Barton conditions produced 172 diastereoselectively, presumably due to steric hindrance on the β-face. Silyl deprotection followed by N⁴-acetyl removal with HCl in MeOH afforded DFP-10917 (26, the hydrochloride salt of CNDAC). Interestingly, attempted deacetylation of 172 with NH₃/MeOH produced mainly cytosine, adding further evidence for the increased acidity of the 2′-α of 26.

Scheme 30. Discovery Synthesis of DFP-10917 26 and Process Synthesis of Sapacitabine 25^a.

^aReagents and conditions: (a) 1,3-dichloro-1,1,3,3-tetraisopropyldisiloxane, pyr; (b) Ac₂O, EtOH; (c) CrO₃/pyr/Ac₂O, DCM; (d) NaCN, NaHCO₃, Et₂O, H₂O; (e) PhOC(S)Cl, DMAP, CH₃CN; (f) Bu₃SnH, AIBN, PhMe (57% over 3 steps); (g) Bu₄NF, CH₃CO₂H, THF then HCl/MeOH, 49% (over two steps); (h) Et₃N, MeOH, DCM, 30 °C, 10 min, 97%; (i) palmitic anhydride, dioxane, H₂O, 85 °C, 89%.

Soon after the discovery synthesis of CNDAC hydrochloride (DFP-10917, 26), a process synthesis for sapacitabine 25 was reported.³⁹⁰ The N⁴-palmitoyl group was installed in the first step, and the rest of the synthesis followed a route similar to that of Scheme 30, with a few variations in reagent choice (i.e., pyridinium dichromate for the 2′-oxidation) producing sapacitabine 25 in nine steps in 20% overall yield.

More recently, Westwood et al. reported a kilogram preparation of sapacitabine 25.^391,392 The synthesis followed a similar route as the two above-mentioned preparations with some variation in reagent use and synthetic order. Deoxygenation was performed using the economical lauroyl peroxide with collidine on a substrate similar to intermediate 171, except with a less-bulky ethyl xanthate ester. Additionally, 3′,5′-OH silyl protection came before N⁴-acylation, generating a solid intermediate that was easy to purify and use in subsequent steps. The final step of the preparation involved N⁴-acylation with palmitic anhydride, producing sapacitabine 25 on a kilogram-scale in nine steps and 37% overall yield.

5.3.5. TAS-106 Biology.

Many antineoplastic agents exert their cytotoxic effects during the S phase (DNA synthesis) of the cell cycle. This poses a problem for more slow growing solid tumors, which have decreased DNA replication rates as compared to leukemias, making the incorporation of 2′-deoxynucleoside-5′-triphosphate analogs very inefficient. However, RNA synthesis occurs throughout the cell cycle except in the M phase. Therefore, a drug that targets both DNA and RNA was hypothesized to offer a broader interval in the cell cycle to exert anticancer effects.²⁶

Drawing on the observation that CNDAC-5′-triphosphate (see Section 5.3.3) was shown to have some RNA synthesis inhibition, TAS-106 46 (ECyd; [1-(3-C-ethynyl-β-d-ribopentofuranosyl)cytosine]) was developed by Taiho Pharmaceuticals in hopes that the TAS-106–5′-triphosphate analog would inhibit RNA polymerase.^382,393 Indeed, the main mechanism of action of TAS-106 46 was shown to be inhibition of RNA polymerases I, II, and III, which essentially blocks RNA synthesis to cause apoptosis.³⁹⁴ In addition, TAS-106 46 was shown to inhibit DNA repair proteins (BRAC2 and Rad51) in the homologous repair pathway, suggesting it might also act via some of the mechanisms of sapacitabine 25.³⁹⁵

TAS-106 46 is phosphorylated to the monophosphate form by uridine/cytidine kinase, which is preferentially expressed in tumor cells versus normal cells.³⁹⁶ It is subsequently converted to the TAS-106–5′-diphosphate and TAS-106–5′-triphosphate forms by uridine/cytidine monophosphate kinase and nucleoside diphosphate kinase, respectively.³⁹⁷ TAS-106–5′-triphosphate remains in cells for a prolonged period of time even after short-term exposure to TAS-106 46, producing significant exposure of cells to the active form of the drug in vitro.³⁹⁸ TAS-106 46 is a poor substrate for cytidine deaminase, although the deaminated uridine analog of the agent also exhibited potent cytotoxicity in vitro.^399,400 TAS-106–5′-diphosphate was anticipated to also block activity of the enzyme due to the precedent that other 2′-deoxyribonucleoside anticancer agents are ribonucleotide reductase inhibitors. However, contrary to the original hypothesis, TAS-106–5′-diphosphate was not found to inhibit ribonucleotide reductase. Additional investigational studies have examined duplex drug linking of 5FdU(5′−5′)TAS-106 against gastric adenocarcinoma cell lines⁴⁰¹ and 5-FdU(3′−5′)TAS-106 exposure to hepatoblastoma cells lines.⁴⁰² Both studies showed positive cytotoxic results in vitro, but further studies are needed to examine the full potential of using these duplex analogs.

TAS-106 46 therapy has produced modest results in clinical trials. In a Phase II clinical trial for salvage metastatic or recurrent head and neck squamous cell cancers and nasopharyngeal cancer, TAS-106 46 treatment was reasonably well tolerated for platinum-failure patients, but had myelosuppression as its main toxicity.⁴⁰³ However, the agent showed no anticancer efficacy (ClinicalTrials.gov identifier: NCT00737360–terminated in 2012–limited number of patients). TAS-106 46 treatment in combination with carboplatin, a DNA alkylating agent, recently underwent a Phase I dose-escalating study in subjects with solid tumors.⁴⁰⁴ The combination was well tolerated, and a few individuals experienced stabilized disease, in which the cancer did not progress (>4 months). There are no current clinical trials underway for TAS-106 46.

5.3.6. TAS-106 Synthesis.

The discovery synthesis of TAS-106 46 is shown in Scheme 31.³⁹⁹ The chosen starting material was 1,2-O-isopropylidene-α-d-xylofuranose 173, which was 5′-OH protected and then 3′-oxidized, forming intermediate 174.⁴⁰⁵ Stereoselective addition of lithiated (trimethyl)acetylene to 174 gave the tertiary alcohol 175 as the desired β-adduct. Subsequent desilylatation, selective 5′-OH benzoylation, isopropylidene deprotection, and 2′-OH benzoylation generated 178. Glycosylation of 178 with cytosine under Vorbrüggen conditions followed by deprotection furnished the desired TAS-106 46 in good overall yield.

Scheme 31. Synthesis of TAS-106 46^a.

^aReagents and conditions: (a) TBDMSCl, pyr, rt, 1.5 h, 100%; (b) CrO₃, Ac₂O, pyr, rt, 1.5 h, 77%; (c) trimethylsilylacetylene, n-BuLi, THF, −78 °C, 99%; (d) n-Bu₄NF, THF, rt; (e) BzCl, pyr, rt, 93% (over two steps); (f) 20% HCl, MeOH, rt; (g) BzCl, DMAP, pyr, 100 °C, 87% (over two steps); (h) concd H₂SO₄, AcOH, Ac₂O, rt, 93%; (i) 2,4-bis(trimethylsilyl)cytosine, SnCl₄, CH₃CN, 0 °C to rt, 18 h, 81%; (j) NH₃/MeOH, rt, 2 days, 90%.

In preparation for clinical trials, a large-scale synthesis of TAS-106 46 was developed (Scheme 32).⁴⁰⁶ Most of the intermediates were crystalline and required no chromatography. Selective 5′-OH protection of 1,2-O-isopropylidine-D-xylofuranose 173 with p-chlorobenzoyl chloride was followed by TEMPO-mediated oxidation of the 3′-OH group. Stereoselective introduction of a TMS-ethynyl group followed by protecting group manipulation set the stage for the key glycosylation reaction. Multiple leaving groups at the anomeric position were screened, and the isobutyryloxy leaving group proved to be most effective, affording the protected intermediate 184 in good yield and stereoselectivity. Final deprotection and crystallization from aqueous methanol furnished 128 g of crystallized TAS-106 46.

Scheme 32. Synthesis of TAS-106 46^a.

^aReagents and conditions: (a) 4-ClBzCl, Et₃N, DCM, 0 °C, 2 h, 74%; (b) TEMPO, DCM NaClO₄, NaHCO₃, H₂O, 0 °C, 30 min, then 2-propanol, rt, 10 min, 88%; (c) trimethylsilylacetylene, EtMgBr, THF, 4 °C, 30 min, 25%; (d) HCO₂H, H₂O, reflux, 45 min, 97%; (e) i-BuCl, Et₃N, DMAP, rt, 5 h, 87%; (f) 2,4-bis(trimethylsilyl)cytosine, SnCl₄, MeCN, 30 °C, 3 h, 81%; (g) DBU, MeOH, 30 °C, 3 h, 68%.

5.3.7. Tezacitabine Biology.

Tezacitabine 45 [MDL 101,731; (E)-2′-deoxy-2′-(fluoromethylene)cytidine] was designed by Hoechst Marion Roussel Inc. as a mechanism-based irreversible ribonucleotide reductase inhibitor.⁴⁰⁷ Such inhibitors are modified at the 2′-position, and often form covalent complexes with the enzyme. Tezacitabine 45 employs a 2′-fluorovinyl moiety, which was proposed to aid in the ribonucleotide reductase-catalyzed abstraction of the 3′-hydrogen atom from tezacitabine-5′-diphosphate, by resonance stabilization of the resultant 3′-radical (Figure 15).⁴⁰⁸ The design was successful, as tezacitabine-5′-diphosphate was indeed found to be an irreversible inhibitor of ribonucleotide reductase.⁴⁰⁹ Furthermore, tezacitabine-5′-triphosphate was found to be a DNA polymerase chain terminator.⁴¹⁰

Figure 15.

Proposed mechanism of catalytic reduction of nucleoside-5′-diphosphates by ribonucleotide reductase (upper), and proposed mechanism of ribonucleotide reductase inhibition by tezacitabine-5′-diphosphate (lower).

Although a cytidine nucleoside analog, tezacitabine 45, is relatively resistant to cytidine deaminase activity,⁴¹¹ upon entering the cell, the agent is phosphorylated by 2′-deoxycytidine kinase and then by other endogenous kinases to the active metabolites: tezacitabine-5′-diphosphate and tezacitabine-5′-triphosphate. Tezacitabine-5′-triphosphate is a substrate for DNA polymerase α, being incorporated into the growing DNA strand in the place of cytidine, and it prevents further chain elongation by DNA polymerases.⁴¹²

In tissue culture tumor cell lines and mouse tumor models, tezacitabine 45 has been examined as a candidate for treatment of colorectal and hematological solid cancers, wherein hydroxyurea and cytarabine 4 have limited antitumor activities.^413,414 Furthermore, the radiosensitizing effects of tezacitabine 45 in the presence of zidovudine (a nucleoside analog inhibitor of reverse-transcriptase) were tested in the colon cancer cell line WiDr.⁴¹⁵ In addition, tezacitabine 45 treatment was also shown to have antiangiogenic activity in xenograft models.⁴¹⁶

A phase I clinical study showed that tezacitabine 45 treatment caused myelotoxicity in 53% of the patient and 83% had febrile episodes.⁴¹² An additional Phase I study showed that tezacitabine 45 plus 5-FU 2 treatment had activity in patients with advanced esophageal and other gastrointestinal carcinomas.⁴¹⁷ However, tezacitabine 45 treatment failed to have significant activity as a stand-alone therapy.⁴¹⁷ In 2004 Chiron reported that it had discontinued development of tezacitabine 45 after it failed to meet expectations in a Phase II trial (ClinicalTrials.gov identifier: NCT00051688–terminated in 2004).

5.3.8. Tezacitabine Synthesis.

McCarthy et al. reported a synthesis of tezacitabine 45,⁴¹⁸ which proceeded in seven steps from uridine in 10% overall yield, requiring five column chromatography purifications. Two years later, an improved process was published that used cytidine as starting material, allowing for the preparation of multigram quantities of the compound in five steps with a 29% overall yield (Scheme 33).⁴¹⁹ A one-pot protection of the 3′-OH and 5′-OH of cytidine 96 by treatment with TIPSCl and of the N⁴-amino group by the addition of N,N-dimethylformamide dimethyl acetal gave intermediate 185. This derivative underwent swern oxidation to provide the keto intermediate 186, which was worked up by a nonaqueous silica gel plug filtration in order to avoid ketone hydration. A Horner–Wadsworth–Emmons reaction between 186 and the in situ-generated diethyl-1-fluoro-1-(phenylsulfonyl)-methylphosphonate produced a separable mixture of fluorovinyl sulfones (Z/E ratio 10:1, respectively). Deprotection of the N⁴-amino group with methanolic ammonia yielded the corresponding α-fluorovinyl sulfone 187, which readily crystallized from hexanes. Radical-mediated sulfone/stannane interchange provided 188 with retention of configuration, and treatment of the intermediate with cesium fluoride in refluxing methanol allowed for the simultaneously removal of the TIPS and tributyltin groups.

Scheme 33. Synthesis of Tezacitabine 45^a.

^aReagents and conditions: (a) TIPSiCl, pyr, 34 °C, 18 h; (b) (CH₃)₂NCH(OCH₃)₂, 37 °C, 4 h, 85% (over two steps); (c) (COCl)₂, DMSO, DCM, −75 °C, 0.5 h, then Et₃N, −75 °C to rt, 1 h, 90%; (d) PhSO₂CH₂F, ClP(O)(OEt)₂, 1 M LiHMDS, −60 °C to rt, 3 h; (e) NH₃/MeOH, rt, overnight, 66% (over two steps); (f) Bu₃SnH, AIBN, cyclohexane, reflux, 18 h, 83%; (g) CsF, MeOH, reflux, 24 h, 69%.

5.3.9. Troxacitabine Biology.

Troxacitabine 48, β-l-dioxolane-cytidine (Troxaty, SGX Pharmaceuticals), was originally studied as an anti-HIV agent, and later it was shown to have potential as an anticancer agent.⁴²⁰ Moreover, this agent is the first L-nucleoside studied for cancer.⁴²¹

Troxacitabine 48 enters cancer cells by passive diffusion, possibly due to the lack of 3′-OH making it not a good substrate for facilitated transporters.⁴²² The agent is not a substrate for cytidine deaminase.⁴²³ However, it is recognized by 2′-deoxycytidine kinase, suggesting more chiral specificity from the former enzyme and less from the latter.⁴²⁴

Troxacitabine-5′-monophosphate is further phosphorylated to the di- and triphosphate forms by cellular enzymes. In contrast to a majority of other anticancer nucleoside analogs, troxacitabine-5′-diphosphate is the predominant intracellular metabolite, and a large pool accumulates intracellularly.^311,425 Phosphoglycerate kinase converts troxacitabine-5′-diphosphate to troxacitabine-5′-triphosphate, the major intracellular active metabolite of troxacitabine 48.⁴²⁶

Troxacitabine-5′-triphosphate is incorporated into newly synthesized DNA, being a substrate for DNA polymerases α, β, δ, γ, and ε, leading to polymerase inhibition by an absolute DNA chain termination mechanism.^427,428 The incorporated troxacitabine-5′-monophosphate is not readily excised from the DNA chain by proof reading 3′ → 5′ exonuclease activities associated with DNA polymerases, due to the chiral specificity of the enzymes. However, removal of the incorporated troxacitabine-5′-monophosphate from DNA can occur by the APE-1 excision repair mechanism.⁴²⁰ Resistance to troxacitabine 48 exposure can arise by decreased expression of 2′-deoxycytidine kinase in vitro.⁴²⁹

Troxacitabine 48 initially showed promising anticancer activity against solid tumors and leukemias.^{424,427,430–432} Tumor cells with catalytically inactive p53 have increased sensitivity to troxacitabine 48 exposure.⁴²⁹ In addition, pancreatic cell lines were shown to have increased sensitivity to troxacitabine prodrugs bearing N⁴ single carbon chain amides to increase drug lipophilicity.⁴³³ However, no additional information is available addressing preclinical efficacy study models for this compound.

Unfortunately, despite the promising preclinical results, troxacitabine 48 therapy had limited success in clinical trials for leukemias and neoplasms (administered by intravenous infusion)^434,435 and ended in the termination of multiple trials (ClinicalTrials.gov identifiers: NCT00129948–terminated in 2006, NCT00104468–terminated in 2007, and NCT00104286–terminated in 2006). In 2007, results were reported for a Phase II study evaluating troxacitabine 48 therapy in relapsed or refractory lymphoproliferative neoplasms or multiple myeloma.⁴³⁶ All the patients had one adverse event, with 62% with one serious adverse event. Overall, troxacitabine 48 treatment had limited benefit in patients with these advanced disorders.⁴³⁶ Furthermore, the agent also exhibited significant toxicity issues, indicating that further development of the agent is unlikely.

5.3.10. Troxacitabine Synthesis.

An initial diastereoselective synthesis of troxacitabine 48 used l-ascorbic acid as starting material (Scheme 34).⁴³⁷ Acid 189 was condensed with benzyloxyacetaldehyde dimethyl acetal in the presence of tosylic acid, affording dioxolane 190 as a mixture of diastereomers. Oxidative degradation followed by benzyltriethylammonium chloride-catalyzed Wolfe oxidation produced a mixture of carboxylic acid products 192, which could be separated with flash chromatography. Oxidative decarboxylation followed by glycosylation, separation of the α and β isomers, and deprotection furnished troxacitabine 48.

Scheme 34. Synthesis of Troxacitabine 48^a.

^aReagents and conditions: (a) PhCH₂OCH₂CH(OMe)₂, TsOH, CH₃CN, 95%; (b) H₂O₂, K₂CO₃, EtOH; (c) RuCl₃ hydrate, NaOCl, DCE/MeCN/H₂O, BnEt₃NCI; (d) H⁺, 51% (over three steps); (e) Pb(OAc)₄, MeCN/DCM, pyr, 80%; (f) silylated N-acetylcytosine, TMSOTf, DCM, rt; (g) deprotection, 30%.

Kim et al. reported a gram-scale synthesis (Scheme 35)⁴³⁸ starting from 2,3:5,6-di-O-isopropylidene-L-gulofuranose 194, which is made by protecting and reducing L-gulono-6,3-lactone. Acid-mediated rearrangement of 194 to pyranose 195 followed by oxidative cleavage and reduction produced dioxolane 196. Protecting group manipulation, oxidation, and Pb(OAc)₄-mediated decarboxylation furnished intermediate 198. Glycosylation of benzoyl-protected cytosine with 198 followed by deprotection with methanolic ammonia afforded troxacitabine 48.

Scheme 35. Synthesis of Troxacitabine 48^a.

^aReagents and conditions: (a) aq HCl, reflux, 20 h, 60%; (b) NaIO₄, MeOH, H₂O, then NaBH₄; (c) TsOH, Me₂CO, 6 h, 62%; (d) BzCl, pyr, DCM, rt, 2 h; (e) TsOH, MeOH, rt, 2 h, 83% (over two steps); (f) NaIO₄, RuO₂·H₂O, CH₃CN:CCl₄:H₂O (1:1:1.5 v/v), rt, 5 h; (g) Pb(OAc)₄, pyr, THF, rt, 45 min, 63% (over two steps); (h) N⁴-Bzcytosine, HMDS, NH₄SO₄, DCE, reflux, 2.5 h, then TMSOTf, DCE, rt, 1.5 h; (i) NH₃/MeOH, 0 °C, rt, 72 h, 38% (over two steps).

5.3.11. Thiarabine Biology.

Thiarabine 47 was first prepared in the 1970s and showed significant anticancer activity. However, difficulty in its synthetic preparation limited further research.^439,440 Two decades later, Southern Research Institute drew upon this finding as well as the report that 4-thionucleo-sides showed resistance to purine nucleoside phosphorylase cleavage⁴⁴¹ to continue the development of this next-generation anticancer nucleoside analog.^442,443

Thiarabine 47 (T-ara C; 4′-thio-arabinofuranosylcytosine; OSI-7836) is phosphorylated to the monophosphate form via 2′-deoxycytidine kinase, although the phosphorylation rate of thiarabine 47 is 1% as compared to cytarabine 4.⁴⁴⁴ The thiarabine-5′-monophosphate exhibits reduced susceptibility to deaminase activity yet is still a good substrate for uridine/cytidine monophosphate kinase, leading to thiarabine-5′-diphosphate.^445,446

Thiarabine-5′-diphosphate is further converted to the triphosphate form (the active metabolite of thiarabine 47). Thiarabine-5′-triphosphate accumulates slowly in the cell but was shown to have long intracellular retention,⁴⁴⁷ and yet was still efficiently degraded to the nucleoside when using cellular extracts.⁴⁴⁶

Thiarabine-5′-triphosphate is a substrate for the nuclear DNA polymerase and is incorporated into DNA, inhibiting DNA synthesis and generating DNA lesions,^444,448 with the primary mechanism of action being disruption of DNA polymerase activity.⁴⁴⁹ In addition, HCT 116 colon carcinoma cells exposed to thiarabine 47 led to cleavage of caspase 3 and PARP that promoted cell death via apoptosis.^447,450 Thiarabine 47 was also reported to have antiangiogenic properties.⁴⁴²

Preclinical studies for thiarabine 47 were promising, having broad spectrum activities against human solid tumors and leukemia/lymphoma xenografts in mice.^17,446 In addition, cross-resistance in multidrug resistant cell lines did not occur for thiarabine 47, suggesting that it potentially could be used in combination drug therapy.^451,452 Previously, a Phase I/II clinical trial showed positive results when using clofarabine 12 and cytarabine 4, having a proposed mechanism of increased cytarabine-5′-triphosphate concentration in cells.⁴⁵³ These results led to a combination therapy of clofarabine 12 and thiarabine 47 study, which showed synergistic activity of 12 and 47 leading to delayed tumor growth in mice.¹⁷

A few clinical trials have been reported for thiarabine 47. In 2006, two Phase I trials were reported for thiarabine 47.^454,455 Study results indicated excessive fatigue reversible grade-3 lymphopenia. Changes in the schedule and duration of thiarabine 47 did not improve its tolerability. Development of the agent has stalled, although some have called for continued clinical evaluation of the agent for hematological and/or solid tumors.⁴⁴⁹

5.3.12. Thiarabine Synthesis.

The first synthesis of thiarabine 47 was accomplished by Whistler et al. (Scheme 36)^{439,456–459} through a long and arduous sequence. The protected glucanofuranose 199, made from glucose in two steps (isopropylidene installation followed by benzylation), was transformed into the protected tosyl sugar 200 in six steps after protecting group manipulation and tosylation. Subsequent treatment of 200 with sodium methoxide, sodium benzyloxide, and tosyl chloride furnished 201, which was further converted to intermediate 202 by exposure to potassium thioacetate. Isopropylidene removal followed by oxidative cleavage gave the ring-opened aldehyde 203, and subsequent cyclization, deprotection/protection, and bromination furnished 204 as a mixture of diastereomers. The glycosylation step afforded an anomeric mixture, with the desired β-adduct crystallizing from solution, and deprotection concluded the synthesis of thiarabine 47.

Scheme 36. Synthesis of Thiarabine 47^a.

^aReagents and conditions: (a) Na, ether, reflux, 16 h, then BnBr, 70 °C, 16 h; (b) Ac₂O, H₂O, 40 °C, 16 h; (c) AcCl, pyr, 25 °C, 16 h, 78% (over three steps); (d) Na, MeOH, 16 h; (e) BzCl, pyr, −15–25 °C, 20 h; (f) TsCl, pyr, CHCl₃, 40 °C, 36 h, 92% (over three steps); (g) NaOMe, CHCl₃, MeOH, −15 to 0 °C, 2 h, 87%; (h) NaOBn, BnOH, 40 °C; (i) TsCl; (j) KSC(O)CH₃, DMF, 115 °C, 73%; (k) HOAc, H₂O, 70 °C, 36 h, 78%; (l) NaIO₄, EtOH, H₂O, 30 °C, 0.5 h, 97%; (m) aq HCl, MeOH, reflux, 3 h, 41%; (n) Na, NH₃, ~100%; (o) BnCl; (p) HBr/HOAc; (q) bis(trimethylsilyl)-N-acetylcytosine, 130 °C, 3 h, 20%; (r) NH₃/MeOH, 0 °C, 2 days, 89%.

Secrist et al. developed an improved gram-scale synthesis of thiarabine 47 in order to access sufficient quantities for biological study (Scheme 37).^441,443 l-Xylose 205 was converted into the benzyl-protected dithioacetal 208 by selective methoxylation, O-benzylation, and dithioacetalization. Treatment of 208 with a triphenylphosphine-iodine-imidazole reagent resulted in a single inversion cyclization at C-4, affording the desired 4-thiosugar 209 in the d-arabino configuration.⁴⁴¹ Coupling of 209 with silylated uracil led to a separable anomeric mixture of 210; as compared to making the cytosine analog, this reaction has several advantages, including a higher β:α ratio, higher yield, and easier purification process. 210β was converted to 212 via C-4 carbonyl activation and amination. Subsequent benzyl deprotection gave thiarabine 47.⁴⁴³

Scheme 37. Synthesis of Thiarabine 47^a.

^aReagents and conditions: (a) HCl, MeOH, rt, 5 h, 95%; (b) BnBr, NaH, TBAI, THF, rt, 3 days, 87%; (c) BnSH, SnCl₄, DCM, rt, overnight, 57%; (d) Ph₃P, imidazole, I₂, PhMe, CH₃CN, 90 °C, 24 h, 83%; (e) uracil, BSA, NBS, MeCN, 50–55 °C, 18 h, column purification (α:β = 1:1.15), 36%; (f) TPSCl, Et₃N, DMAP, 5 °C to rt, overnight; (g) NH₄OH, MeCN, rt, overnight, 50%.

5.3.13. RX-3117 Biology.

RX-3117 29, a fluorocyclopentenyl-cytosine nucleoside analog, was shown to have anticancer activity^460,461 with an IC₅₀ range of 0.4 to >30 μM for 59 cell lines.⁴⁶² The agent was also shown to inhibit DNA and RNA synthesis and induce apoptotic cell death of tumor cells.³⁴⁹ Interestingly, preclinical studies using nude mice implanted with various tumors, including colon, lung, renal, and pancreatic, indicated that the ribose form and not the 2′-deoxyribose form was the active compound.⁴⁶³

Biological data were recently published on the metabolism and mechanism of action of RX-3117 29.⁴⁶² Cellular uptake of the agent occurs by hENT, and the agent is a poor substrate for cytidine deaminase, preventing its degradation within the cell.⁴⁶² Furthermore, uridine-cytidine kinase generates RX-3117–5′-monophosphate, which is subsequently phosphorylated to the di- and triphosphate forms by cellular kinases. RX-3117–5′-trisphophate is incorporated into RNA and inhibits RNA synthesis at high concentrations, but at IC₅₀ values it does not affect RNA structural integrity. RX-3117–5′-diphosphate is a substrate for ribonucleotide reductase, and the 2′-deoxy-RX-3117–5′-triphosphate can be incorporated into DNA, potentially providing another mechanism of cancer cell growth inhibition. The 2′-deoxy-RX-3117–5′-triphosphate also downregulates DNA methyltransferase-1 protein (at IC₅₀ values for specific cell line) and Cdc2 protein expression, which are both involved in the cell cycle division process.⁴⁶²

Preclinical and clinical studies of RX-3117 29 showed potent efficacy in xenograph models.⁴⁶⁴ In 2012, Rexahn Pharmaceuticals reported the completion of a Phase I clinical trial for RX-3117 29 in Europe and indicated that the drug had 56% orally bioavailable, had a plasma half-life of 14 h, and was well tolerated in cancer patients. Currently, a Phase Ib trial is ongoing to test the dose-finding and safety of RX-3117 29 as an oral monotherapy to treat advanced malignancies (ClinicalTrials.gov identifier: NCT02030067–recruiting patients).

5.3.14. RX-3117 Synthesis.

A synthesis of RX-3117 29 is illustrated in Scheme 38.⁴⁶³ Protection of d-ribose 213 followed by Wittig olefination, Swern oxidation, and vinyl Grignard addition afforded diene 217. Silyl ether deprotection and selective primary alcohol benzylation, followed by ring-closing metathesis and oxidative rearrangement via a [3,3]-sigmatropic shift yielded benzylated cyclopentenone 221. Subsequent iodination of the double bond, lactone reduction, and protection of the resulting lactol provided TBDPS ether 224. Electrophilic vinyl fluorination followed by desilylation gave fluorocyclopentenol 225, which was coupled with N³-benzoyluracil under Mitsunobu conditions, providing protected nucleoside analogue 226. Removal of the N³-benzoyl group by methanolic ammonia and the 5′-benzyl group by BBr₃ generated 227. Finally, triacetatyl intermediate 227 was converted to the cytidine analog RX-3117 29 by a sequence involving formation of a 4-triazole intermediate and nucleophilic displacement with ammonia.

Scheme 38. Synthesis of RX-3117 29^a.

^aReagents and conditions: (a) concd H₂SO₄, Me₂CO, then TBDPSCl, imidazole, 99%; (b) Ph₃PCH₃Br, t-BuOK, THF, 93%; (c) (COCl)₂, DMSO, DCM, −78 °C, 1 h, then Et₃N, rt, 96%; (d) CH₂=CHMgBr, THF, −78 °C, 84%; (e) n-Bu₄NF, THF, 97%; (f) Bu₂Sn(O), PhMe, 15 h, then TBAI, BnBr, 50 °C, 16 h, 76%; (g) Grubbs’ catalyst (2^nd generation), PhMe, 80 °C, 90%; (h) PDC, DMF, rt, 18 h, 59%; (i) I₂, pyr, THF, 55%; (j) NaBH₄, CeCl₃, MeOH, 93%; (k) TBDPSCl, imidazole, DMF, 97%; (l) NFSI, n-BuLi, THF, −78 °C; (m) n-Bu₄NF, THF, 63% (over two steps); (n) N³-benzoyluracil, DEAD, Ph₃P, THF, 87%; (o) NH₃/MeOH; (p) BBr₃, DCM, −78 °C, 50%; (q) Ac₂O, pyr, then POCl₃, Et₃N, 1,2,4-triazole; (r) NH₄OH, dioxane; (s) NH₃/MeOH, 40% (over three steps).

An alternative and slightly improved approach to RX-3117 29 uses a trityl protecting group at the 5′-position of intermediate 214, which avoids a deprotection/reprotection step, decreasing the total number of steps of the synthesis by two.⁴⁶⁵ However, both processes include long and laborious syntheses, producing only small quantities of the final product in low yield. A large-scale synthesis is most likely needed for significant clinical treatment of RX-3117 29.

5.4. Cytarabine and Cytarabine Prodrugs

5.4.1. Cytarabine Biology.

In the 1940s, Werner Bergmann isolated two nucleosides from the marine sponge Cryptotethya crypta, which he termed spongothymidine and spongouridine. These spongonucleosides were determined to have a 2′-OH in the arabino configuration. In the late 1950s, various unnatural nucleosides (including spongonucleosides) were prepared with the hypothesis that they would kill cancer cells by interfering with DNA replication, and indeed, many were found to inhibit DNA synthesis in vitro. In the 1960s, Seymour Cohen reported promising cytotoxic effects of a new spongonucleoside in cancer cells: cytarabine 4 (aka spongocytidine, cytosine arabinoside, or ara-C).⁴⁶⁶ The anticancer agent rapidly progressed through rodent models and clinical trials to gain FDA approval as a cancer therapy.

Cytarabine 4 is transported into the cell primarily by hENT-1.^467,468 Cytarabine 4 is monophosphorylated by 2′-deoxycyti-dine kinase, which is the rate-limiting step in the formation of the active form: cytarabine-5′-triphosphate.⁴⁶⁹ Cytarabine-5′-tri-phosphate competes with dCTP for incorporation into DNA and is a weak competitive inhibitor of DNA polymerases. Cytarabine-5′-triphosphate is also a substrate for DNA polymerase α and is incorporated into both the leading and lagging strands of DNA. This incorporation leads to delayed chain termination, allowing several additional nucleotides to be incorporated before the DNA polymerase stops extending and falls off the replication fork. Cell cycle arrest occurs with the DNA repair mechanism deployed. If the embedded cytarabine-5′-monophosphate is not removed, the cell will enter apoptosis.^470–472

Various factors decrease the effectiveness of cytarabine 4. The agent is deaminated by cytidine deaminase, and cytarabine-5′-monophosphate can be deaminated by deoxycytidylate deaminase, leading to the inactive uridine analogs.⁴⁷³ In addition, cytarabine-5′-triphosphate is a feedback inhibitor of 2′- deoxycytidine kinase, limiting the formation of cytarabine-5′-monophosphate.⁴⁷⁴

Resistance to cytarabine 4 exposure is afforded by several different mechanisms, including nucleotide transporter and 2′-deoxycytidine kinase gene mutations and decreased expression levels,⁴⁷⁵ and changes in cytosolic nucleotidase II,⁴⁷⁶ cytidine deaminase, and deoxycytidylate deaminase activities.⁴⁷⁷ Recently, a correlation was reported between increased expression of the ABC transporter MRP8 and a poor treatment response of individuals with acute myelogenous leukemia treated with cytarabine 4.⁴⁷⁸

Cytarabine 4 has been FDA-approved since 1969 for the treatment of acute myeloid leukemia,⁴⁷⁹ with the exception of acute promyelocytic leukemia.⁴⁸⁰ It is usually given in combination with a topoisomerase II inhibitor such as daunorubicin, idarubicin, or mitoxantrone.^481–483 Several clinical trials are underway using cytarabine 4 alone or in combination treatment studies. (ClinicalTrials.gov identifiers: NCT01191541–recruiting patients, NCT01289457–recruiting patients, and NCT01756677–recruiting patients).

5.4.2. Elacytarabine Biology.

Elacytarabine (51, CP-4055), an oleic acid ester of cytarabine 4, was designed to enter cancer cells independent of nucleoside transporters, which would also overcome the low expression level of hENT-1 at the plasma membrane or loss of function mutation in hENT-1.^484,485 The lipophilic prodrug diffuses through the phospholipid bilayer, being retained in a membrane fraction within the cell.⁴⁸⁶ After the action of an unidentified esterase, cytarabine 4 is released into the cytosol.³⁴⁹ In comparison to cytarabine 4, elacytarabine 51 exposure led to an increased intracellular level of active metabolite, which was also retained longer in cells after drug removal.⁴⁸⁶

Elacytarabine 51 showed promising results in both preclinical and clinical studies.^487–489 However, in April 2013, despite the initial success, Clavis Pharma announced that elacytarabine 51 did not perform better than the regular standard of care therapy in a pivotal Phase III trial³⁵⁴ (ClinicalTrials.gov identifier: NCT01147939–completed in 2013). Moreover, patients with relapsed/refractory acute myeloid leukemia had no meaningful benefit from elacytarabine 51 when directly compared to seven other treatment regiments.⁴⁹⁰ As a result, Clavis has suspended all further developmental work on elacytarabine 51.

5.4.3. MB 07133 Biology.

Ligand Pharmaceuticals Inc. developed MB 07133 34, a phosphoramidate prodrug of cytarabine 4. In 2003, MB 07133 34 entered a phase I/II clinical trial as an intravenous infusion agent for the treatment of unresectable hepatocellular carcinoma (ClinicalTrials.gov identifier: NCT00073736–completed–no results posted). MB 07133 34 was well tolerated with evidence of disease stabilization. In 2011, Ligand Pharmaceuticals granted the license to Chiva Pharmaceuticals to develop MB 07133 34 in China.

5.4.4. Cytarabine, Elacytarabine, and MB 07133 Syntheses.

In 1959, cytarabine 4 and 3-β-d-arabinofuranosyluracil were originally synthesized by the Dekker group (Scheme 39).⁴⁹¹ Cytidine 93 was treated with polyphosphoryic acid and purified by gradient elution chromatography to afford the diphosphate intermediate 228 as a crystalline cyclohexylammonium salt. Subsequent enzymatic dephosphorylation with prostatic phosphatase furnished cytarabine 4. In order to verify the hypothesis that the transformation proceeded via a cyclonucleoside-5′-diphosphate intermediate, O²-2′-cyclocyti-dine 229 was isolated by ion-exchange fractionation. The intermediate could further be converted to cytarabine 4 by enzyme-mediated dephosphorylation and hydrolysis.

Scheme 39. Synthesis of Cytarabine 4^a.

^aReagents and conditions: (a) polyphosphoric acid; (b) prostatic phosphatase; (c) OH^–.

Two different process syntheses of cytarabine 4 used 1-β-d-arabinfuranosyl uracil 231 as a key intermediate, which can be synthesized on a large-scale in two steps from uridine 98 via 2,2′-anhydro-uridine 230 (Scheme 40).⁴⁹² One process treated 231 with HMDS under high pressure and high temperature, leading to the intermediate 232, which upon treatment with MeOH provided cytarabine 4 on a 50-g-scale.⁴⁹³ The other process converted 231 to the O-acyl-substituted, 1,2,4-triazole derivative 233, and subsequent deprotection produced cytarabine 4 on a kilogram-scale.⁴⁹⁴

Scheme 40. Synthesis of Cytarabine 4^a.

^aReagents and conditions: (a) (PhCO)₂O, DMF, 100 °C, then K₂CO₃, 137 °C, 1.5 h, 78%; (b) aq HCl, 80 °C, 2 h, 68%; (c) HMDS, high pressure, 150 °C, 80 h; (d) MeOH, 5 °C, 1 h, 87% (over two steps); (e) Ac₂O, pyr, rt, 3 h, 91%; (f) POCl₃, 1H-1,2,4-triazole, Et₃N, CHCl₃, <8 °C then rt, overnight, 84%; (g) NH₄OH, NH₃, dioxane, H₂O, rt, overnight, 70%.

Elacytarabine 51 can be prepared in one step by enzyme-mediated esterification of the 5′-OH of cytarabine 4 with oleic anhydride in DMF,⁴⁹⁵ whereas the cyclic protide MB 07133 34 has been synthesized from cytarabine 4 in four steps (Scheme 41).^496,497 Selective 2′,3′-silyl protection followed by treatment with N,N-dimethylformamide dimethyl acetal generated intermediate 236.⁴⁹⁶ In parallel, trans-p-nitrophenylphosphate 239 was prepared in three steps from methyl 3-hydroxy-3-(pyridin-4-yl)-propanoate 237 via esterification (with l-N,N-dimethyl phenylalanine) and subsequent diastereomer separation, reduction with LiAlH₄, and coupling with p-nitrophenyl phosphorodichloridate. Condensation of protected nucleoside 236 with p-nitrophenyl phosphorochloridate 239 followed by TBS deprotection furnished MB 07133 34.⁴⁹⁷

Scheme 41. Synthesis of Elacytarabine 51 and MB 07133 34^a.

^aReagents and conditions: (a) oleic anhydride, DMF, Mucor miehei lipase (MML), 60 °C, 48 h, 88%; (b) TBSCl, imidazole, DMF, rt, overnight, 85%; (c) 50% TFA, THF, −10 °C, 18 h, 65%; (d) DMF-dimethyl acetal, pyr, rt, 16 h, 90%;(e) EDC, DMAP, l-N,N-dimethyl-phenylalanine, DCM, rt, 16 h, 83%; (f) LiAlH₄, ether, −20 °C, 1 h, 66%; (g) 4-nitrophenyl phosphorodichloridate, THF, pyr, 0 °C to rt, 3.5 h, then sodium 4-nitrophenoxide, 40 °C, 4 h, 63%; (h) t-BuMgCl, THF, rt, 16 h, 78%; (i) 70% TFA, 60 °C, 16 h, 83%.

6. PURINE ANALOGS

6.1. Deoxypurines

6.1.1. Cordycepin Biology.

In 1951, cordycepin 27 was isolated from the fungus Cordyceps militaris⁴⁹⁸ and exhibited anticancer, anti-inflammatory, and antifungal activities.^499,500 It is currently thought to be the active ingredient in various traditional Chinese medicines.⁵⁰¹ The mechanism of cellular uptake of this agent has not been fully elucidated. However, the 3′-OH has been shown to be required for efficient uptake of nucleosides and nucleoside analogs by hENTs, suggesting that cordycepin 27 does not efficiently use this mechanism for uptake.⁵⁰² Once in the cell, the agent is converted to the cordycepin-5′-monophosphate by adenosine kinase,⁵⁰³ and then to the corresponding di- and triphosphate forms by other cellular kinases. Cordycepin-5′-triphosphate is incorporated into the poly A tail of the growing mRNA chain, presumably causing chain termination due to its lack of a 3′-OH.⁵⁰⁴ Furthermore, cordycepin 27 can inhibit ribose-phosphate pyrophosphokinase and 5-phosphoribosyl-1-pyrophosphate amidotransferase in vitro.^505,506

Cordycepin 27 exposure was shown to cause DNA double-stranded break in breast cancer cells.⁵⁰⁷ A possible explanation for this observation is the inhibitory effect this agent has on poly(ADP-ribose) polymerase, an enzyme involved in the repair of DNA single-strand breaks.²⁸ Unrepaired DNA single-strand breaks could lead to more extensive DNA damage, such as double-stranded breaks. In addition, cordycepin 27 is a substrate for adenosine deaminase, which can lead to significant degradation in vivo. Other mechanisms of action for cordycepin 27 are binding A3 adenosine and DR3 receptors, activating caspase-8/−3 pathways for cell death, suppressing cyclin D1 and c-myc expression cell signal pathways, inhibiting cAMP formation, and decreasing glycogen synthase kinase-3 β/β-catenin activation.^508–511

In 2008, Oncovista initiated a Phase I/II clinical trial of cordycepin 27 in combination with pentostatin 6 (adenosine deaminase inhibitor; see Section 6.1.5) for the treatment of refractory TdT-positive leukemia (ClinicalTrials.gov identifier: NCT00709215–unknown status >2 years without an update). Currently, no study updates have been reported. Earlier in 2000, a Phase I trial was conducted using cordycepin 27 plus pentostatin 6 in treating patients with refractory acute lymphocytic or chronic myelogenous leukemia (ClinicalTrials. gov identifier: NCT00003005–completed in 2001–no results reported). Development of cordycepin 27 seemed to have stalled, until Rainforest Pharmacal announced plans to reactivate the IND in 2016 and start a Phase I/II trial in TdT-positive refractory leukemia patients in 2017. However, issues with deamination will have to be addressed before cordycepin 27 can progress significantly in clinical trials.

6.1.2. Cordycepin Synthesis.

In 1961, Lee et al. reported a pioneering synthesis of cordycepin 27, which involved desulfurization of 3′-ethylthioadenosine by treatment with Raney nickel in refluxing 2-methoxyethanol.⁵¹² Later, Hansske and Robins used 2-acetoxyisobutyryl bromide (Mattock’s bromide) to synthesize cordycepin (Scheme 42).⁵¹³ Thus, reaction of adenosine 240 with 2-acetoxyisobutyryl bromide afforded the corresponding trans-bromoacetates (241a and 241b). Cordycepin 27 was finally obtained after treatment with acidic resin to provide epoxide 242 and regioselective reduction with lithium triethylborohydride.

Scheme 42. Synthesis of Cordycepin 27^a.

^aReagents and conditions: (a) 2-acetoxyisobutyryl bromide, CH₃CN, H₂O, rt, 1 h; (b) Amberlite IRA-400 (OH^–) resin, MeOH, 92% (over two steps); (c) 1 M LiEt₃BH, THF, Me₂SO, 0 °C to rt, overnight, then 5% HOAc/H₂O, Et₃B, 98%.

Aman et al. used a similar approach for a process synthesis of cordycepin 27 (Scheme 43).⁵¹⁴ Initially, the authors thought to use the epoxide 242 as a key intermediate, but they had difficulty controlling the regioselectivity of the subsequent reduction. They noticed, however, that trans-bromoacetate 243b fortuitously precipitated selectively from solution. This intermediate was isolated and subsequently deprotected at the 2′- and 5′-positions in order to increase compound solubility and reactivity (decreased steric hindrance) in the final reduction reaction. The deprotection required a delicate balance of reaction temperature, acid concentration, and methanol concentration in order to fully hydrolyze 243b to 244, without leading to compound decomposition via depurination. Depurination also plagued the final dehalogenation step, and conditions had to be optimized. Finally, hydrogenation in the presence of 4–5 molar equivalent of sodium acetate in a 1:1 solvent mixture of ethanol and water effectively debrominated the hydrochloride salt of 244. Ethyl acetate was used to extract crude cordycepin 27 from the reaction mixture, which had a high salt load. Subsequent addition of methanol to the organic solution followed by cooled agitation precipitated cordycepin 27. The overall process required no chromatography and produced the product cordycepin 27 on a 100-g-scale at 99% purity.

Scheme 43. Synthesis of Cordycepin 27^a.

^aReagents and conditions: (a) 2-acetoxyisobutyryl bromide, CH₃CN, EtOAc, <35 °C, 6 h, 52% for 243b; (b) AcBr, MeOH, concd HCl, 20 °C, rt, 2–4 days, 70%; (c) AcONa, EtOH, H₂O then Pd/C, H₂ (50 psi), rt to 50 °C, 60%.

6.1.3. Cladribine Biology.

Inherited adenosine deaminase deficiencies were reported to cause lymphopenia, an immunodeficiency in humans.⁵¹⁵ Furthermore, the accumulation of deoxyadenosine nucleotides had been implicated in the pathogenesis of this disorder. An increase in dATP levels was shown to kill lymphocytes.⁵¹⁶ Experimentally, a high dADP level in lymphoid cells led to negative feedback inhibition of ribonucleotide reductase, and the subsequent decrease in dNTP pools resulted in DNA synthesis inhibition and cell death.⁵¹⁷ It is now known that inhibition of adenosine deaminase causes an increase in both dATP and dADP levels.

Carson et al. reasoned that deoxyadenosine analogs, which are resistant to adenosine deaminase activity, may be phosphorylated into their triphosphate forms in the cell. Moreover, a deoxyadenosine analog that inhibits adenosine deaminase activity might be successful for the treatment of lymphocytic malignancies.⁵¹⁸ In collaboration with John Montgomery, Carson studied 2-chloro-2′-deoxyadenosine (cladribine, 8), which was shown to have significant cytotoxicity in malignant T lymphocytes.⁵¹⁸ The presence of the C2 chlorine in the purine base was believed to make the agent resistant to adenosine deaminase activity.⁵¹⁹ Furthermore, cladribine 8 was shown to be converted to the cladribine-5′-triphosphate in vitro, and to exert significant cytotoxicity both in vitro and in vivo.^519–521

Cladribine 8 is transported by both hCNT and hENT into cells.⁵²² It is phosphorylated by cytosolic 2′-deoxycytidine kinase⁵¹⁶ and mitochondrial deoxyguanosine kinase to generate cladribine-5′-monophosphate. This phosphorylation is the rate-limiting step in the activation of the parent nucleoside analog.⁵²³ Subsequent phosphorylation generates cladribine-5′-diphosphate and cladribine-5′-triphosphate forms; both are active metabolites of cladribine 8. Cladribine-5′-triphosphate is readily incorporated into DNA.⁵²⁴ Chain extension by DNA polymerase continues past a single DNA incorporated cladribine-5′-monophosphate, but elongation is terminated when the polymerase encounters three successive incorporated cladribine molecules in the growing DNA strand.¹⁷ In addition, cladribine-5′-triphosphate inhibits ribonucleotide reductase activity.⁵²⁵ Collectively, these actions can promote cell death by dNTP imbalance and DNA double-stranded breaks.^526,527 Furthermore, cladribine 8 was shown to interfere with methyltransferases and may promote hypomethylation in vitro.^528–530 Other effects of cladribine 8 include the induction of caspase-3 activation, thereby promoting cell apoptosis in vitro.⁵³¹ Interestingly, the agent has also been shown to impede mononuclear cells from crossing the blood brain barrier and to inhibit T cells and B cells in the central nervous system.⁵³²

Resistance to cladribine 8 exposure can arise from compound metabolism and detoxification as well as increased egress of the agent from cancer cells in vitro.⁵³³ Interestingly, a recent report examining metabolism data for cladribine 8 treatment in animals and humans indicated the formation of 2-chlorodeoxyinosine and 2-chlorohypoxanthine (inactive metabolites), suggesting a sensitivity to adenosine deaminase degradation.⁵³⁴ Furthermore, cladribine 8 was shown to be a substrate for the ABCG2 transporter, leading to cell egress.⁵³⁵

Cladribine 8 gained FDA approval in 2004 for use as a first-line monotherapy for hairy cell leukemia^536–538 and has a favorable toxicity profile in comparison to other lymphocytic drugs.⁵³⁹ It has been used in combination treatment for chronic lymphocytic leukemia³⁵ and abnormal mast cells debulking.⁵⁴⁰ Cladribine 8 has also been reported to kill immature and mature monocyte-derived dendritic cells.⁵³⁷ In addition, the agent was shown to be effective for treatment of Langerhans cell histiocytosis.⁵⁴¹ Cladribine 8 therapy has also been investigated for relapsingremitting multiple sclerosis, but has not gained FDA approval for this disease due to lack of safety information.⁵⁴²

6.1.4. Cladribine Synthesis.

Interestingly, the first reported preparation of cladribine 8 was as an intermediate in a synthesis of 2′-deoxyguanosine.^543,544 In 1972, Christensen et al. reported a nonstereoselective fusion glycosylation synthesis of cladribine 8 from diprotected sugar 245, made by treatment of 2′-deoxyribose in methanol with a catalytic amount of acid followed by p-toluoyl protection (Scheme 44).⁵⁴⁵ The anomeric mixture of glycosylated product 246 was resolved by silica gel chromatography, and the β-anomer was treated with methanolic ammonia, to give cladribine 8.

Scheme 44. Synthesis of Cladribine 8^a.

^aReagents and conditions: (a) 2,6-dichloropurine, melt, then dichloroacetic acid, and fusion under vacuum; (b) NH₃/NaOMe.

In 2006, Robins reported a new synthesis of cladribine 8 that focused on improving the yield of the glycosylation step between N⁶-imidazolyl purine 248 and α−1′-chlorosugar 247, which is formed by treating 2′-deoxyribose with HCl in methanol followed by tosylation and stereoselective crystallization of the α-anomer (Scheme 45).^546,547 The modified purine base exhibited improved solubility in organic solvents as well as better glycosylation regioselectivity, due to steric hindrance of the N⁷ position of the base, leading to exclusive formation of the N⁹ adduct. A binary solvent system was employed in the glycosylation in attempts to increase the solubility of the modified purine base as well as improve the stereoselectivity of the reaction. The rational for this system drew on a previously reported observation that preferential solvation of reactants can lead to different free energies of activation, and therefore alter the ratios of two competitive pathways.⁵⁴⁸ Indeed, a judicious solvent choice of cold dichloromethane and acetonitrile (1:1 mixture) allowed for the selective formation of the desired β-nucleoside in 95% yield. Furthermore, the use of cold solvent led to decreased anomerization of chlorosugar 247. The synthesis was concluded by converting the imidazoyl group to an imidazolium group by treatment with in situ-generated benzyl iodide, and the resulting salt was heated in methanolic ammonia to give, after recrystallization, pure cladribine 8, in >90% from chlorosugar 247.

Scheme 45. Synthesis of Cladribine 8^a.

^aReagents and conditions: (a) NaH, CH₃CN, rt, 8 h; (b) then 247, 0 °C to rt, 22 h, 100% (over two steps); (c) BnI, CH₃CN, 60 °C, 1.5 h; (d) NH₃/MeOH, 60 °C, 11 h, >90% (over two steps).

More recently, a kilogram-scale synthesis has been reported for cladribine 8 (Scheme 46).⁵⁴⁹ The researchers chose to explore Vorbrüggen conditions for the glycosylation. Generally, when this reaction involves 2′-deoxy lactones, it usually gives a product ratio of roughly 50% α/β nucleosides. This was indeed found to be the case when the glycosylation of the 4-chlorobenzyl protected sugar 249 and silylated 2-chloroadenine was performed in solvents such as tetrahydrofuran and dichloromethane. However, when solvents such as acetonitrile or lithium hexamethyldisilazane were employed, the desired β-anomer fortuitously precipitated from solution. It was hypothesized that a dynamic equilibrium existed in the reaction, involving glycosylation and retroglycosylation of the α-anomer along with concomitant precipitation of the β-anomer, driving the equilibrium to the desired β-nucleoside product 253. The reaction was complete within 1 h; however, an aging step at 60 °C was used to further increase the yield. Furthermore, only a true catalytic amount of the Lewis acid trimethylsilyl trifluoromethanesulfonate was needed, which greatly simplified the purification of the product by obviating the need for a catalyst quench and aqueous workup. Isolation was accomplished by direct filtration. Deprotection with 10–20 mol % of sodium methoxide in MeOH led to in situ crystallization of high purity cladribine 8, in 99.8–99.9% HPLC purity with an overall yield of 43%.

Scheme 46. Synthesis of Cladribine 8^a.

^aReagents and conditions: (a) 25% NaOMe/MeOH.

6.1.5. Pentostatin Biology.

Pentostatin (6, [(R)-3-(2-deoxy-β-D-erythro-pento-furanosyl)-3,6,7,8-tetrahydro-imidazo-[4,5-d][1,3-diazepin-8-ol]) is an antibiotic isolated from a culture broth of Streptomyces antibioticus.⁵⁵⁰ It is an irreversible, stoichiometric inhibitor of mammalian adenosine deaminase, with a K_i in the nanomolar range.^551,552 Importantly, pentostatin 6 directly inhibits this enzyme in order to truly mimic adenosine deaminase deficiencies, which is in contrast to cladribine 8, which was designed as an adenosine deaminase-resistant antimetabolite (see Section 6.1.3.). Pentostatin 6 is unique among anticancer nucleoside analogs in that it is active without metabolism (cellular phosphorylation). It has been proposed that the tetrahedral carbon at position 8 of the diazepine ring of pentostatin 6 mimics the purported transition state tetrahedral carbon in the deamination of adenosine to inosine.⁵⁵³ In addition to this effect, pentostatin 6 also targets cellular methyltransferases, hindering the ability of the cell to methylate both DNA and mRNA.^554–556

Pentostatin 6 is used to treat hairy cell leukemia and is effective against lymphoid malignancies with high adenosine deaminase activities.^557,558 The agent has also been used to prevent and to treat both acute and chronic graft-versus-host disease.⁵⁵⁹ Pentostatin 6 is currently being tested in several clinical trials. An additional Phase I/II clinical trial uses pentostatin 6 and cyclophosphamide (DNA-alkylating agent) cotreatment for low-intensity stem cell transplantation with multiple lymphocyte infusions and to treat advanced kidney cancer (ClinicalTrials.gov identifier: NCT00923845–active status–primary objective of trial was not achieved). A Phase II trial combines pentostatin 6, cyclophosphamide (DNA alkylating agent), and rituximab (monoclonal antibody against B cell protein CD20) followed by lenalidomide (antiangiogenic/immunomodulatory agent) treatment for relapsed or refractory B cell chronic lymphocytic leukemia (ClinicalTrials.gov identifier: NCT00074282–unknown status–no updates in 2 years). Finally, a Phase I trial involves the combination of pentostatin 6, bendamustine (DNA-alkylating agent), and ofatumumab (monoclonal antibody against B cell protein CD20) for the treatment of chronic lymphocytic leukemia (ClinicalTrials.gov identifier: NCT01352312–active status–not recruiting patients).

6.1.6. Pentostatin Synthesis.

Parke-Davis developed a large-scale industrial production of pentostatin 6 using fermentation cultures of Streptomyces antibioticus NRRL 3238. To date, this approach is the major form of production of the compound, although a few other syntheses have been reported.⁵⁶⁰

In 1982, the first synthesis of pentostatin 6 was reported by chemists at Warner-Lambert/Parke-Davis (Scheme 47).^561,562 Knoevenagel condensation of azole 254 with benzaldehyde produced styrylimidazole 255, and subsequent benzylation formed regioisomers 256 and 257. The mixture was subjected to ozonolysis, and 258 was separated from its isomeric product by differential precipitation. Carbon homologation of 258 was followed by nitro reduction, ring closure, and debenzylation, producing heterocycle 261. Due to the instability of heterocycle 261, glycosylation conditions had to be optimized. The use of tin(IV) chloride as Lewis acid and BSTFA as silylating agent were found to produce the desired product in a 1:1 mixture of α/β diastereomers. The mixture can be separated by preparative HPLC on a small-scale and by fractional crystallization on a larger-scale. Deprotection of the β-anomer 262 in the presence of sodium methoxide followed by ketone reduction produced a separable 1:1 mixture of pentostatin 6 and its stereoisomer. It is noteworthy that Zhang et al. improved the conclusion of the synthesis by using a chiral ruthenium catalyst for asymmetric transfer hydrogenation of ketone 262.⁵⁶³ This reaction was completely diastereoselective and gave a yield of 80%.

Scheme 47. Synthesis of Pentostatin 6^a.

^aReagents and conditions: (a) PhCHO, piperidine, 95 °C, 21 h, 70%; (b) BnCl, K₂CO₃, DMF, 75 °C, 6 h, 96%; (c) O₃, DCM, −78 °C, 8 h, then HCO₂H, H₂O₂, 0 °C to rt, overnight, 67%; (d) CDI, THF, reflux, 1 h, then t-BuOK, CH₃NO₂, THF, 0 °C, 45 min, 68%; (e) SnCl₂, concd HCl, 60 °C, 2.5 h, then H₂S, rt, 75%; (f) H₂, Pd/C, MeOH, H₂O, rt, 16 h, 96%; (g) HC(OEt)₃, Me₂SO, 65 °C, 15 min, 81%; (h) N,N-bis(trimethylsilyl)trifluoroacetamide, pyr, CH₃CN, rt, overnight; (i) 2-deoxy-3,4-di-O-p-toluoyl-D-pentofuranosyl chloride, SnCl₄, DCE, CH₃CN, −50 °C, 45 min; (j) NaHCO₃, H₂O, rt, 0.5 h, 23% (over three steps); (k) NaOMe, MeOH, rt, 1 h; (l) NaBH₄, H₂O, MeOH, 25 °C, 30 min, 26% (over two steps).

Phiasivongsa and Redkar reported a ring expansion of protected 2′-deoxyinosine 263 using diazomethane.⁵⁶⁴ This was followed by deprotection and reduction to furnish a diastereomeric mixture of pentostatin 6, which was separated by following the Warner-Lambert/Parke-Davis procedure in Scheme 48.⁵⁶⁴ Unfortunately, no yields were reported.

Scheme 48. Synthesis of Pentostatin 6^a.

^aReagents and conditions: (a) (Me₃Si)₂NC(O)CF₃, pyr, MeCN, 12 h; (b) BF₃·Et₂O, DCM, 0 °C, 30 min; (c) TBAF, THF, 0 °C, 2 h; (d) NaBH₄, MeOH, H₂O, rt, 1 h.

6.1.7. Clofarabine Biology.

Clofarabine 12 is a second-generation purine nucleoside analog designed to overcome the limitations of fludarabine-5′-monophosphate 7 (see Section 6.2.3) and cladribine 8 (see Section 6.1.3),⁵⁶⁵ which are both susceptible to glycosidic bond cleavage.⁵²¹ Indeed, the introduction of a C2′-fluorine in the arabino configuration in clofarabine 12 significantly increased the stability of the glycosidic bond in acidic solution and toward phosphorolytic cleavage.^566,567 Chlorine substitution instead of fluorine at the 2-position of the adenine base was chosen in order to avoid production of the 2-fluoroadenine, a precursor to the toxic 2-fluoro-adenosine-5′-triphosphate, should the glycosidic bond be cleaved.⁵⁶⁸

Clofarabine 12 is usually administered intravenously; however, the increased stability of the glycosidic bond also allows for oral administration.^566,569,570 Clofarabine 12 enters the cell via hENT1, hENT2, and hCNT2, and to a much lesser extent by passive diffusion.⁴²⁹ 2′-Deoxycytidine kinase generates clofarabine-5′-monophosphate;⁵⁷¹ however, the enzyme has a lower K_m for clofarabine 12 than cladribine 8.⁵⁷² Moreover, clofarabine resistance arises from decreased 2′-deoxycytidine kinase activity in vitro.⁵⁷² Formation of clofarabine-5′-diphosphosphate is generated by purine nucleotide monophosphate kinase, and is the rate-limiting step in the activation of the agent.⁵⁷³ This is in contrast to many other nucleoside analogs, in which the rate-limiting step is the generation of clofarabine-5′-monophosphate.

Clofarabine-5′-diphosphate serves as an intracellular reservoir for the active-metabolite clofarabine-5′-triphosphate, which is retained in the cell longer than clofarabine-5′-triphosphate in cells from CLL and AML patients ex vivo.⁵⁷² Clofarabine-5′-triphosphate has several mechanisms of action in the cell. It inhibits ribonucleotide reductase (presumably by allosteric inhibition), which results in a decrease in dCTP and dATP concentrations.⁵²⁰ The reduction in the amount of dCTP likely leads to DNA synthesis inhibition, and a decrease in the concentrations of dATP leads to greater DNA incorporation of clofarabine-5′-triphosphate (self-potentiation).⁵⁶⁸ A low intracellular concentration of clofarabine-5′-triphosphate can be incorporated by DNA polymerases α and ε⁵⁷⁴ into DNA and can promote polymerase arrest at the replication fork, inhibiting DNA repair and inducing strand breaks in vitro.^573,574 This DNA damage leads to the induction of cytochrome c-mediated apoptosis in vitro.⁵⁷⁵ Studies have shown that clofarabine-5′-triphosphate can also be incorporated into RNA in cell lines.⁵⁷³

Clofarabine 12 is an FDA-approved drug for the treatment of relapsed or refractory pediatric acute lymphoblastic leukemias.^35,576 It is being evaluated for the treatment of acute myeloid leukemia in both children and the elderly.⁵⁷⁷ The agent was examined for treatment of non-Hodgkin’s lymphomas, myelodysplastic syndrome, and solid tumors.⁵²⁰ Oral clofarabine 12 has been studied in relapse/refractory non-Hodgkin’s lymphomas⁵⁷⁸ and in high-risk myelodysplastic syndrome patients.⁵⁷⁹ Currently clofarabine 12 in combination treatments is being investigated in over a dozen clinical trials for chronic lymphocytic leukemia, acute myelogenous leukemia, myelodysplastic syndrome, and mixed phenotype acute leukemia with nucleoside analogs such as cytarabine 4, gemcitabine, and fludarabine-5′-monophosphate 7 and with other chemotherapies such as but not limited to busulfan, idarubicin, etoposide, and mitoxantrone.

6.1.8. Clofarabine Synthesis.

In the early 1990s, Montgomery et al. utilized differentially protected sugar 272 in order to access various nucleoside analogs, including clofarabine 12 (Scheme 49).^566,580,581 The synthesis began with the stereoselective fluorination of diprotected tosylate 267, which can be formed by isopropylidene protection and tosylation of α-d-glucofuranose, followed by selective deprotection, leading to fluorinated sugar 268. Benzoylation of the 5′-OH followed by acid-mediated deprotection afforded lactol 269. Oxidative cleavage and subsequent rearrangement produced 270, which was eventually converted to bromosugar 272. Condensation of 2,6-dichloropurine with 272 occurred in refluxing dichloroethane, and chromatographic separation of the anomeric isomers produced nucleoside derivative 273. Three days of treatment of 273 in a steel bomb with ethanolic ammonia afforded a mixture of clofarabine 12 and the partially deprotected 5′-O-benzoyl derivative. Treatment of the mixture with LiOH in MeCN/H₂O removed the residual protecting group, and pure clofarabine 12 was produced by three crystallizations from water.

Scheme 49. Synthesis of Clofarabine 12^a.

^aReagents and conditions: (a) KF, acetamide, 210 °C, 62%; (b) MeOH–0.7% H₂SO₄ (1:1 v/v); (c) BzCl in DCM, pyr, −15 °C, 80% (over two steps); (d) Amberlite IR-120 (H⁺), H₂O/dioxane, 80 °C, 78%; (e) KIO₄, H₂O; (f) NaOMe, MeOH; (g) Ac₂O, pyr, 80% (over three steps); (h) HBr/HOAc, DCM; (i) 2,6-dichloropurine, DCE, 100 °C, 32% (β-anomer); then NH₃/EtOH, steel bomb, 3 days, solvent switch to MeCN/H₂O, LiOH, 42%.

Twelve years later, Bauta et al. reported an improved and scalable process synthesis of clofarabine 12 (Scheme 50).⁵⁸² Studies were performed on bromosugar 279,⁵⁸³ with the desire to increase β-anomer formation in the glycosylation step. Compound 279 was easily prepared by first conversion of 1-O-acetyl-2,3,5-tri-O-benzoyl-β-d-ribofuranose 274 to protected sugar 276⁵⁸⁴ via formation of the benzoxonium ion intermediate and solvolysis. Subsequent installation of the imidazolylsulfonate leaving group, stereoselective deoxyfluorination using excess KHF₂, and finally bromination under acidic conditions completed the synthesis of 279.⁵⁸³ Empirical studies of the glycosylation step of 2-chloroadenine found potassium tert-butoxide to be the optimum base, with the potassium counterion being particularly important for both yield and β-selectivity. Solvent choice was guided by the precedent that low dielectric constants of a solvent favor the formation of the β-anomer in glycosylations involving bromosugar 279. It is believed that this effect arises from suppression of the formation of the oxonium ion intermediate 280 and subsequent S_N1-type condensation, which would lead to both α- and β-glycosylation products. Nonpolar solvents would therefore be needed in order to favor a S_N2 reaction pathway. However, solubility in such solvents would also be an issue, particularly with the 2-chloroadenine base. In order to find a balance between the two conflicting requirements of anomeric selectivity and substrate solubility, a 1:2:1 mixture of MeCN/tert-amyl alcohol/DCE was eventually chosen after considerable experimentation. In addition, the additive CaH₂ was found to increase overall yield and anomeric ratio, due in part to removal of trace amounts of water from the solvent.

Scheme 50. Synthesis of Clofarabine 12^a.

^aReagents and conditions: (a) 0.5 M HBr, DCM, 0 °C; (b) H₂O, 81% (over two steps); (c) NaH, DMF, 0 °C, 0.5 h, then N,N′-sulfuryldiimidazole, −40 °C to rt, 2 h, 85%; (d) KHF₂, 2,3-butanediol, HF (50% in H₂O), 160 °C, 1 h, 63%; (e) HBr/HOAc, DCM, rt, 16 h, 98%; (f) 2-chloroadenine, CH₃CN, t-amyl alcohol, tert-BuOK, CaH₂, 50 °C, 40 min, then 279, DCE, 50 °C, 19 h, 50% (β/α = 80:1); (g) NaOMe, MeOH, 33 °C, 7 h, 64%.

A number of challenges were overcome to separate protected clofarabine 281 from the heterogeneous reaction mixture. Excess 2-chloroadenine was filtered off. A subsequent solvent exchange to n-butyl acetate dissolved the mixture to allow for heptane-induced precipitation of crude 281, which after two subsequent pH-maintained methanol slurries was isolated in a >50:1 β/α ratio. Finally, deprotection with a catalytic amount of sodium methoxide in methanol followed by recrystallization formed drug-pure clofarabine 12 on a 25-g-scale.

6.2. Arabinose Purine Analogs

6.2.1. Nelarabine Biology.

Arabinosylguanine (9-β-d-arabinofuranosylguanine; ara-G) was discovered in the 1960s, but was not significantly developed at that time. In the 1970s, studies on the rare autosomal disorder purine nucleoside phosphorylase deficiency rekindled interest in ara-G.^393,585 A deficiency in purine nucleoside phosphorylase, an enzyme that catalyzes the phosphate-mediated deribosylation of certain purine nucleosides, results in a profound dGTP-mediated imbalance in the endogenous nucleotide pools that leads to T cell lymphopenia without concomitant decreases in B cell populations.⁵⁸⁶ This sensitivity of human T cells (especially immature T cells) to purine nucleoside phosphorylase deficiency is hypothesized to arise from their relatively high ratio of kinase activity versus nucleotidase activity (in contrast to other cell types) which results in a high accumulation of intracellular dGTP.⁵⁸⁷ Deoxyguanosine, but not ara-G, is a substrate for purine nucleoside phosphorylase.^588,589

Nelarabine (13, 2-amino-9-β-d-arabinosyl-6-methoxy-9H-guanine; 506U78) is a prodrug of ara-G and has a 10-fold higher solubility as compared to ara-G.^590,591 Nelarabine 13 is converted to ara-G in the serum by adenosine deaminase.⁵⁹² Ara-G is phosphorylated by either cytosolic 2′-deoxycytidine kinase or mitochondrial deoxyguanidine kinase into ara-G-5′-monophosphate⁵⁹³ and, subsequently, into ara-G-5′-diphosphate and ara-G-5′-triphosphate by cellular kinases. The active metabolite ara-G-5′-triphosphate is a substrate for DNA polymerases and is incorporated into DNA. This incorporation leads to inhibition of the DNA polymerases, causing inhibition of DNA replication and resulting in cell death via apoptosis.⁵⁹⁴ Moreover, it was believed that ara-G-5′-triphosphate would accumulate to high intracellular levels and would promote cytotoxic effects similar to those created by high dGTP concentrations. When T cell lymphoma cell lines were exposed to nelarabine, this resulted in FasL-mediated cytotoxicity, whereas myeloid and B cells did not accumulate ara-G-5′-triphosphate, but were stopped in the S phase of the cell cycle.⁵⁹⁵ High concentrations of ara-G-5′-triphosphate accumulated in malignant T cells from human bone marrow when tested in a rodent model.⁵⁹⁶

In 2005, nelarabine 13 was approved by the FDA for the treatment of relapse T cell acute lymphocytic leukemia and relapse T cell lymphoblastic lymphoma.⁵⁹⁷ Additional clinical trials are being conducted studying pharmacokinetic and pharmacodynamics properties in addition to efficacy in combination with etoposide (topoisomerase inhibitor) and cyclophosphamide (DNA-alkylating agent) treatment in lymphomas (ClinicalTrials.gov identifiers: NCT01094860–recruiting patients and NCT00981799–terminated status for Phase I/II in 2015).

6.2.2. Nelarabine Synthesis.

An early synthesis of nelarabine 13, shown in Scheme 51,⁵⁹⁸ goes through a key enzyme catalyzed trans-glycosylation between 1-β-d-arabinosyl uracil 231 and 2-amino-6-methoxy-9H-purine 282. The reaction uses uridine phosphorylase and purine nucleoside phosphorylase and is completely regio- and stereoselective, furnishing only the N⁹, β-anomer nelarabine 13 (Scheme 51).

Scheme 51. Synthesis of Nelarabine 13^a.

^aReagents and conditions: (a) 10 mM K₃PO₄, n-PrOH, H₂O, pH 6.75, uridine phosphorylase, purine nucleoside phosphorylase, 37 °C, 26 days, 17%.

A more recent process synthesis for nelarabine 13 was reported by Zong et al. (Scheme 52).⁵⁹⁹ Glycosylation of 2-amino-6-chloropurine 61 with 2,3,5 tri-O-benzyl-arabifuranyl 283β (see Section 6.2.4 for preparation) followed by methoxy displacement of the chlorine at position 6 of the purine base furnished benzylated intermediate 285.^600,601 Debenzylation by catalytic transfer hydrogenation, using ammonium formate as the in situ hydrogen source, afforded nelarabine 13 on a 30-g-scale at 99.8% purity after HPLC.

Scheme 52. Synthesis of Nelarabine 13^a.

^aReagents and conditions: (a) NaH, MeCN; (b) MeONa, MeOH; (c) Pd/C, HCOONH₄, MeOH, reflux, 90 min, 91%.

6.2.3. Fludarabine and Fludarabine-5′-monophosphate Biologies.

Arabinofuranosyladenine (ara-A; vidarabine) has been around since the 1960s and was originally studied for its antiviral effects.⁶⁰² The agent was also tested for anticancer effects, but showed limited success due to its rapid degradation by adenosine deaminase. In order to avoid adenosine deaminase-mediated deactivation, a fluorine was installed at the 2-position of the purine base of ara-A, producing fludarabine 291.⁶⁰³ This agent, however, had limited solubility and difficulties with its formulation. Therefore, the fludarabine-5′-monophosphate prodrug 7 (Fludara or fludarabine phosphate) was subsequently explored in clinical trials, which led to its FDA approval in 1991. Fludarabine-5′-monophosphate 7 is a commonly used drug against chronic lymphoid leukemia and hairy cell leukemia.⁵³⁸

The negatively charged fludarabine-5′-monophosphate 7 is dephosphorylated in the plasma by ecto-5′-nucleotidase, and fludarabine 291 is transported into cells by nucleoside transporters hENT1, hENT2, and hCNT3.^429,592 Although fludarabine-5′-monophosphate is adenosine deaminase resistant,⁶⁰⁴ the drug is susceptible to glycosidic bond cleavage, which results in the formation of 2-fluoroadenine, a precursor to the toxic 2-fluoroadenosine-5′-triphosphate (see section 6.1.7).⁶⁰⁵ After entering the cell, fludarabine is initially phosphorylated by cellular 2′-deoxycytidine kinase, although it is a relatively poor substrate for the enzyme.^606,607 Deoxyguanosine kinase has also been reported to phosphorylate fludarabine.⁶⁰⁸ Fludarabine-5′-monophosphate 7 is then phosphorylated to the fludarabine-5′-diphosphate and fludarabine-5′-triphosphate by adenylate kinase and nucleoside diphosphate kinase, respectively, in the cell.

Fludarabine-5′-triphosphate is the active cytotoxic agent. It is a ribonucleotide reductase inhibitor, which leads to a decrease in cellular dNTP concentrations, and therefore negatively influences DNA synthesis. As the concentrations of dNTP diminish during cell division, this results in more fludarabine-5′-triphosphate being incorporated into the host DNA (self-potentiation).^572,592 Furthermore, the incorporated fludarabine-5′-monophosphate prevents further DNA chain elongation by restricting the addition of more dNTPs, leading to effective chain termination.⁶⁰³ Moreover, fludarabine-5′-triphosphate can inhibit DNA polymerase, DNA primase, and DNA ligase, causing DNA strand breaks, which may lead to apoptosis.^592,609 Fludarabine-5′-triphosphate was also shown to be incorporated into RNA and can inhibit RNA synthesis.^312,610 In addition, fludarabine 291 exposure can also suppress the excision repair enzyme ERCC1 and led to cytotoxic synergy with SJG-136 (DNA minor groove binding agent) in chronic lymphocytic leukemia cells.⁶¹¹ However, the side effects of fludarabine-5′-monophosphate 7 treatment are becoming an increasing concern, because secondary myelodysplastic syndrome and leukemia have been reported in at least 3% of persons treated with fludarabine-5′-monophosphate 7 (Fludara).⁶⁰⁹

Several different mechanisms have been reported that lead to fludarabine resistance in vivo. Changes in the expressions of miRNA-29a, miRNA-181a, and miRNA-221 have been reported.⁶¹² The amplification of MYC gene and its transcript level miRNAs can promote fludarabine resistance.⁶¹³ Differential overexpression of sulfatase was correlated to fludarabine resistance.⁶¹³

Fludarabine-5′-monophosphate 7, cyclophosphamide (DNA-alkylating agent), and rituximab (monoclonal antibody against B cell protein CD20) combination therapy is generally considered the standard treatment for younger fit patients with chronic lymphocytic leukemia.⁶¹⁴ A combination treatment of fludarabine-5′-monophosphate 7, bendamustine (DNA-alkylating agent), and rituximab for chronic lymphocytic leukemia is being investigated, in addition to the fludarabine-5′-monophosphate 7, clofarabine 12, SAHA (vorinostat, histone deacetylase inhibitor), and busulfan (antineoplastic alkylating agent) combination for acute leukemia (ClinicalTrials.gov identifiers: NCT01096992–active status–not recruiting patients yet, and NCT02083250–recruiting patients).

6.2.4. Fludarabine and Fludarabine-5′-Monophosphate Syntheses.

In 1969, Montgomery and Hewson reported the discovery synthesis of fludarabine 291 (Scheme 53).⁶¹⁵ Key chlorosugar 283 was prepared from d-ribose by first cyclization in methanolic sulfuric acid to methyl-β-d-ribofuranoside, followed by 3′,5′-O-benzylation, and subsequent hydrolysis, p-nitrobenzoylation, and chlorination of the anomeric position.^616,617 The glycosylation step between chlorosugar 283 and 2,6-dichloropurine 286 gave a column-chromatography-separable mixture of α/β diastereomers, with the desired β-anomer 287 formed in only 11%. Treatment with sodium azide followed by catalytic reduction furnished the diaminopurine intermediate 289. Due to solubility issues of 289, a mixed solvent system was employed for the modified Balz–Schiemann reaction, which introduced the N²-fluorine. Debenzylation concluded the synthesis, producing fludarabine 291.

Scheme 53. Synthesis of Fludarabine 291 and Fludarabine-5′-monophosphate 7^a.

^aReagents and conditions: (a) Hg(CN)₂, CaSO₄, MeNO₂, reflux, 3 h, 11%; (b) NaN₃, EtOH, H₂O, 1 h, 98%; (c) Pd/C, H₂, EtOH, rt, 6 h, 75%; (d) HBF₄, NaNO₂, CHCl₃, −10 °C, 40 min, 36%; (e) Na, NH₃, 34%; (f) POCl₃, PO(OEt)₃, 0 °C, 3.5 h, 96%.

Montgomery later reported a gram-scale preparation of fludarabine 291, which utilized the same basic route, although the glycosylation reaction was performed using a silylated or acylated derivative of 2,6-diaminopurine.⁶¹⁸ Soon thereafter, he reported a gram-scale preparation of fludarabine-5′-monophosphate 7 by treatment of fludarabine 291 with phosphoryl chloride and triethyl phosphate (Scheme 53).⁶¹⁹

Blumbergs et al. developed a kilogram-scale preparation of fludarabine-5′-monophosphate 7, which followed a similar route used by Montgomery (Scheme 54).^620,621 Yields were improved, allowing the synthesis of the drug in five steps from chlorosugar 283α (12% yield). Improvements included the use of 2,6-di(methoxyacetamido)purine 293 during the glycosylation, which increased base solubility and greatly reduced the volume of solvent (ethylene dichloride) required for the reaction. Furthermore, chlorosugar 283α, which was produced almost exclusively as the α-anomer by treating the p-nitrophenol intermediate 294 with hydrogen chloride in dichloromethane,⁶¹⁷ was used in the glycosylation step, which allowed for higher yield of the desired β-anomer of 295.

Scheme 54. Synthesis of Fludarabine 291 and Fludarabine-5′-monophosphate 7^a.

^aReagents and conditions: (a) methoxyacetic anhydride, pyr, 88 °C, 1 h, then methyl ethyl ketone, overnight, 90%; (b) HCl (g), DCM, 4–6 °C, 2 h, quantitative; (c) DIPEA, DCE, reflux, overnight, quantitative; (d) NaOMe, MeOH, 68%; (e) POCl₃, TMP, 0 °C, 18 h, 56%.

Another highlight of the synthesis involved the use of hydrochloric acid in the debenzylation step to avoid the partial-defluorination problem associated with Montgomery’s catalytic hydrogenation, and to simplify the purification procedure. Furthermore, the phosphorylation step yield was significantly increased when using scrupulously dried nucleoside as the starting material. Recently, another publication effected the debenzylation by transfer hydrogenation, using ammonia formate as an in situ hydrogen donor, to produce kilograms of fludarabine 291 at 99.8% purity in 90–95% yield.⁶²²

Farina et al. prepared fludarabine 291 through enzymatic transglycosylation involving Enterobacter aerogenes, 2-fluoroade-nine, and 9-β-d-arabinofuranosyl-uracil (Scheme 55).⁶²³ The crude product was treated with acetic anhydride, forming 2′,3′,5′-tri-O-acetyl-9-β-d-arabinofuranosyl-2-fluoroadenine, which was hydrolyzed and recrystallized to the pure fludarabine 291. Subsequent phosphorylation afforded fludarabine-5′-monophosphate 7.

Scheme 55. Synthesis of Fludarabine 291 and Fludarabine-5′-monophosphate 7^a.

^aReagents and conditions: (a) Enterobacter aerogenes, K₂HPO₄, H₂O, 60 °C, 24 h, 35%; (b) Ac₂O, 95 °C, 9 h, 87%; (c) NH₄OH, H₂O, MeOH, rt, 19 h, 74%.

6.3. Base Modified Purine Nucleosides

6.3.1. 8-Chloro-adenosine and Tocladesine Biologies.

In the 1980s, 8-chloro-adenosine 28 and tocladesine (50; 8-chloroadenosine 3′,5′-monophosphate; 8-Cl-cAMP) were reported to have potent anticancer activities.^624–627 Tocladesine 50 is converted extracellularly to 8-chloro-adenosine 28,⁶²⁸ which is then transported by hCNT1 and hCNT2 into the cell.⁶²⁹ Once in the cell, this agent is converted via adenosine kinase to 8-chloro-adenosine-5′-monophosphate, and ultimately to the active metabolite 8-chloro-adenosine-5′-triphosphate by cellular kinases. 8-Chloro-adenosine-5′-triphosphate accumulates to almost millimolar concentration in the cell.⁶³⁰ 8-Chloroadenosine is a poor substrate for adenosine deaminase, and therefore, deamination of the agent is minimal.⁶⁰³

The 8-chloro-adenosine-5′-triphosphate affects cellular energetics by decreasing the levels of intracellular ATP, leading to RNA synthesis inhibition. Furthermore, RNA polymerase II incorporates 8-chloro-adenosine-5′-triphosphate into mRNA, and inhibits polyadenylation of full-length mRNA transcripts.^631,632 A decrease in these short-lived mRNA transcripts can also lead to a reduction in protein levels, which promote growth and survival of tumor cells. 8-Chloro-adenosine 28 exposure has been found to deplete the cyclin E level in breast cancer cell lines,⁶²⁴ decrease the level of the receptor tyrosine kinase MET in multiple myeloma cell lines,⁶³³ and depress the level of Mcl-1 in chronic lymphocytic leukemia cells.⁶³⁰ Collectively, all of these mechanisms promote apoptosis.

Although many effects of 8-chloro-adenosine 28 are RNA-directed, 8-chloro-adenosine-5′-triphosphate has also been found to inhibit topoisomerase II and induce DNA double-stranded breaks in human myelocytic leukemia K562 cells.⁶³⁴ 8-Chloro-adenosine 28 exposure also inhibits the rate of DNA synthesis and decreases dATP pools in mantle cell lymphoma cell lines.⁶³⁵ Furthermore, decrease in cancer cell proliferation by 8-chloro-adenosine 28 exposure has been linked to AMP-activated protein kinase and p38 mitogen-activated protein kinase inhibitions in vitro.^636,637

Glucose-6-phosphate dehydrogenase has been linked to 8-chloro-adenosine 28 resistance in multiple myeloma and chronic lymphocytic leukemia cell lines.⁶³⁸ Also, a report indicated that the agent influences glucose consumption, leading to autophagy activation in multiple myeloma cells.⁶³⁹

In 2008, the M.D. Anderson Cancer Center sponsored a Phase I clinical trial of a 8-chloro-adenosine 28 therapy to study dosing tolerance in subjects with chronic lymphocytic leukemia (ClinicalTrials.gov identifier: NCT00714103–active status–not recruiting patients yet). In June 2015, the City of Hope Medical Center began a Phase I/II clinical trial evaluating 8-chloro-adenosine 28 in treating patients with relapsed or refractory acute myeloid leukemia (ClinicalTrials.gov identifier: NCT00004902–currently recruiting participants). Tocladesine 50 was evaluated in two clinical trials. In 2000, it was evaluated in a Phase II clinical trial for the treatment of persons with recurrent or refractory multiple myeloma (ClinicalTrials.gov identifier: NCT00004902–completed status). In 2001 the agent was studied in a Phase I trial for the treatment of progressive metastatic colorectal cancer (ClinicalTrials.gov identifier NCT00021268–unknown status–greater than 2 years since updated). No trial results have been reported.

6.3.2. 8-Chloroadenosine and Tocladesine Syntheses.

Synthesis of 8-chloroadenosine 28 has been achieved in one step from adenosine 240 by using either tetrabutylammonium iodotetrachloride as the chlorinating reagent⁶⁴⁰ or m-CPBA and acid (HCl or acetyl chloride) in aprotic solvents (Scheme 56).^641,642 Additionally, the synthesis of 8-chloroadenosine 28 has been achieved by direct coupling of 8-chloroadenosine 28 with 2,3,5-tri-O-benzoyl-d-ribofuranosyl bromide. However, the major product of the reaction is the N³ glycosylated product and not the desired N⁹ product.⁶⁴³ Tocladesine 50 can be prepared by treatment of 8-chloroadenosine 28 with phosphoryl chloride in triethyl phosphate.⁶⁴²

Scheme 56. Synthesis of 8-Chloroadenosine 28 and Tocladesine 50^a.

^aReagents and conditions: (a) AcCl, m-CPBA, DMF, rt, 20 min, 40%; (b) POCl₃, PO(OEt)₃, 0 °C, 6 h, then aq NaOH, 2 h, 23%.

6.3.3. Forodesine Hydrochloride Biology.

Forodesine hydrochloride (36, (1S)-1-(9-deazaguanin-9-yl)-1,4-dideoxy-1–4-imino-d-ribitol; BCX-1777; Immucillin H) was developed as a purine nucleoside phosphorylase inhibitor (see Section 6.2.1). Transition state analysis of the purine nucleoside phosphorylasecatalyzed phosphorolysis of inosine was used as a model system in the design of the agent (Figure 16).^33,34 The protonated nitrogen of the iminoribitol ring of forodesine exhibits a similar charge distribution to the ribosyl moiety oxocarbenium ion of the transition state. Furthermore, the 9-deazapurine of forodesine can be protonated at N7, which provides additional interaction with a residue of the purine nucleoside phosphorylase active site.⁶⁴⁴ In addition, instead of a natural glycosidic bond, forodesine has a carbon–carbon bond which is not readily cleaved by purine nucleoside phosphorylase.⁶³⁰

Figure 16.

Transition state model for the phosphorolysis of inosine by purine nucleoside phosphorylase and the protonated form of forodesine.

Forodesine was shown to be a potent purine nucleoside phosphorylase inhibitor. Complete inhibition of the homotrimeric complex of the enzyme at a concentration as low as 72 pM has been reported, with the agent exhibiting a high equilibrium binding constant for and a very slow dissociation rate from the enzyme.^337,630 Inhibition occurs at a 1:1 molar ratio of inhibitor to purine nucleoside phosphorylase homotrimeric enzyme complex, with just one of the three substrate sites needing to be occupied by the agent to block enzyme activity.³⁴

Forodesine hydrochloride 36 is administered by intravenous infusion. Like pentostatin 6 (see Section 6.1.5), the agent is active without metabolism, meaning it does not require phosphorylation. It blocks purine nucleoside phosphorylase, causing a high plasma concentration of deoxyguanosine, and it is not incorporated into DNA.⁶⁴⁵ Deoxyguanosine is further converted to dGMP, dGDP, and dGTP by cellular kinases. High cellular levels of dGDP inhibit the ribonucleotide reductase complex and lead to imbalances of the nucleotide pool concentrations, resulting in p53-based apoptosis.⁵⁹¹ Because of this, forodesine works best in malignant cells containing high deoxyguanosine activity.⁶⁴⁶

Forodesine hydrochloride 36 has undergone clinical trials for the treatment of T-cell malignancies, including T-cell acute lymphocytic leukemia.^35,647 Additionally, the agent is being examined for treatment of chronic lymphocytic leukemia B cells, which may require an additional nucleoside treatment for effectiveness.^35,648 Forodesine hydrochloride 36 did not inhibit the growth of colon cancer cell lines or normal human nonstimulated T cells, indicating some specificity of the drug toward normal versus malignant tissues.³³ In 2007 a phase IIb clinical trial of forodesine hydrochloride 36 in patients with leukemia/lymphoma was terminated due to manufacturing issues (ClinicalTrials.gov identifier: NCT00419081–terminated), indicating a need to improve the synthesis process in order to produce sufficient quantities of the agent for further clinical trials. In January 2013, Mundipharma began a Phase I/II clinical trial evaluating forodesine hydrochloride 36 in recurrent/refractory peripheral T cell lymphoma subjects in Japan (ClinicalTrials.gov identifier NCT01776411–active status–not recruiting patients).

6.3.4. Forodesine Synthesis.

The discovery synthesis of forodesine involved formation of the key iminoribitol intermediate 306 (see Scheme 57) followed by a 16-step linear construction of the base, producing only milligram quantities of the compound.⁶⁴⁹ Therefore, a different route was designed, which could produce the required kilogram quantities of the agent needed for drug development. This new synthesis involved the coupling of imine 307 and brominated 9-deazahypoxanthine 313 (Scheme 57). One metric ton of 306 was prepared from d-ribose 213 by first bromine oxidation, protection, lactone epimerization,⁶⁵⁰ and Fleet chemistry (five steps).⁶⁵¹ The same quantity of 9-deazahypoxanthine 310 was produced by condensing (ethoxymethylene)cyanoacetate 308 with diethyl aminomalonate followed by cyclization and acid-mediated decarboxylation.⁶⁵² Three more steps were required to produce the bromo hypoxanthine derivative 312: chlorination, protection and methoxy displacement, and finally bromination. At this stage, the condensation of lithiated heterocycle 313 (generated in-situ from 312) with imine 307 afforded the β-anomer of 314 exclusively, due to steric hindrance of the α-face by the isopropylidene protecting group. In order to avoid a hard-to-remove tarry byproduct, 314 was not converted directly to the product, but instead passed through a process of protection and sequential deprotection, eventually furnishing pure forodesine hydrochloride 36.⁶⁵³

Scheme 57. Synthesis of Forodesine Hydrochloride 36^a.

^aReagents and conditions: (a) Br₂, K₂CO₃, H₂O, 5 °C to rt, overnight; (b) concd H₂SO₄, Me₂CO, reflux, 4 h, then 30 °C, 8 h, 55% (over two steps); (c) MsCl, Et₃N, DCM, −20 °C, 1 h, then rt, 8 h; (d) aq KOH, <30 °C, 4 h, 59% (over two steps); (e) TBDMSCl, imidazole, DMF, 20 °C, 4 h, 91%; (f) LiBH₄, THF, −30 °C to rt, 24 h, 75%; (g) (MeSO₂)₂O, pyr, 0 °C to rt, 18 h, 88%; (h) NaN₃, DMF, 100 °C, 2 h, 42%; (i) H₂, Pd, NaOAc, dioxane, 20 °C, 72 h, 94%; (j) NCS, pentane, rt, 1 h, then lithium tetramethylpiperidide (LiTMP), THF, −78 °C, 85%; (k) H₂NCH(CO₂C₂H₅)₂, NaOMe, MeOH, reflux, 4 h, 68%; (l) formamidine acetate, C₂H₅OH, reflux, 27 h, 61%; (m) aq KOH, reflux, 40 h, 74%; (n) POCl₃, reflux, 2 h; (o) BOMCl, NaH, THF, 0 °C to rt, 1 h, then NaH, MeOH, 1 h; (p) NBS, DCM, 52% (over three steps); (q) n-BuLi, −78 °C; (r) 307, −78 °C to 0 °C; (s) (Boc)₂O, DCM, rt, 1 h, 76% (over three steps); (t) H₂, Pd/C, 18 h, then NH₄OH, H₂O, 1 h; (u) concd HCl, MeOH, reflux, 1.5 h, 85% (over two steps).

In order to avoid some of the cumbersome manipulations from the above-mentioned route, a different approach to forodesine hydrochloride 36 explored a novel route to prepare the aza-sugar core (Scheme 58).⁶⁵⁴ Lactam 317, which is prepared by esterification and reduction of l-pyroglutamic acid, was transformed into unsaturated bicycle 318 by treatment with benzaldehyde and catalytic acid followed by treatment with Meyer’s reagent (methyl phenolsulfonate).⁶⁵⁵ Osmium tetr-oxide-catalyzed dihydroxylation of compound 318 gave the desired diol stereochemistry, and subsequent acetal protection furnished the intermediate 319. The coupling of lithiated 9-deazahypoxanthine to lactam 319 led to the ring-opened product 320. Treatment of the intermediate 320 with BBr₃ gave the cyclized product 321, which was then reduced with BH₃·Me₂S and deprotected under acidic condition to give forodesine hydrochloride 36 as a 4:1 β:α mixture.

Scheme 58. Synthesis of Forodesine Hydrochloride 36^a.

^aReagents and conditions: (a) PhCHO, TsOH; (b) PhSO₂Me, 59%; (c) OsO₄, Me CO; (d) DMP/H⁺ 2, 63%; (e) lithiated 9-deazahypoxanthine, anisole, Et₂O, −20 °C, 55%; (f) BBr₃, DCM; (g) BH₃·Me₂S, DMF; (h) MeOH/H⁺, 90%.

In 2016, a 50-g-scale synthesis of forodesine hydrochloride 36 was reported.⁶⁵⁶ Differences from the original preparation (Scheme 57) include coupling of imine 307 with lithiated heterocycle 313 under milder condition, and the use of concentrated HCl instead of hydrogen for BOM removal in 315.

7. ENZYME INHIBITOR AND CELL SURFACE RECEPTOR ANTAGONIST NUCLEOSIDE ANALOGS

7.1. Triciribine and Triciribine-5′-Monophosphate Biologies

In 1971, Schram and Townsend synthesized triciribine 30, a tricyclic derivative of a purine nucleoside as a potential anticancer drug,⁶⁵⁷ which has also been studied as an antiretroviral agent against HIV-1 and HIV-2.⁶⁵⁸ Triciribine 30, however, is poorly soluble, and this suboptimal property led to the development of the highly soluble prodrug triciribine-5′-monophosphate 33.

Interestingly, triciribine 30 does not work by blocking host DNA or viral polymerases. It inhibits the phosphorylation of Akt1, Akt2, and Akt3,⁶⁵⁹ and appears to influence the PI3K/Akt/mTOR cell survival pathway. The PI3K/Akt/mTOR pathway regulates a variety of cellular pathways, including cell proliferation and survival.^660,661 Akt is dysregulated in a variety of malignancies.⁶⁶² Initially, triciribine 30 was reported to inhibit de novo purine nucleotide and DNA synthesis in cell lines.^425,663 However, this would not explain the antiretroviral activity in HIV-1 and HIV-2 infected cell lines, because the reverse transcription step for proviral DNA synthesis of the viral life cycle was completed. Moreover, viral replication was no longer influenced by cellular dNTP concentrations or nucleoside reverse transcriptase inhibitors, which are chain terminators. Based on current data, the antiviral and anticancer activities of triciribine metabolite appear to be associated with Akt inhibition (Figure 17).⁶⁶⁴ In addition to breast cancer, triciribine 30 may be effective when used as a combination therapy in bladder cancer patients prescreened for alterations in the Akt pathway.^665,666

Figure 17.

Proposed biological mechanisms of actions of triciribine 30 as an Akt inhibitor (see Section 7.1, Triciribine Biology); MLN4924 44 as a Cullins inhibitor (see Section 7.3, MLN4924 Biology); and acadesine 41 as an AMPK activator (see Section 8.5, Acadesine Biology).

Triciribine-5′-monophosphate 33 is currently being evaluated in solid tumors, such as breast cancer.^662,667,668 A Phase I/II study is currently underway in patients with metastatic and locally advanced breast cancer that uses triciribine-5′-monophosphate monohydrate with paclitaxel (antimitotic taxane), which is followed by dose-dense doxorubicin (DNA intercalating agent) and the DNA alkylating-agent cyclophosphamide (ClinicalTrials.gov identifier: NCT01697293–recruiting patients). The H. Lee Moffitt Cancer Center and Research Institute recently began a Phase I/II clinical trial for the treatment of ovarian cancer with a combination of triciribine 30 and the DNA alkylating agent carboplatin (ClinicalTrials.gov identifier: NCT01690468–recruiting patients).

7.2. Triciribine and Triciribine-5′-Monophosphate Syntheses

Triciribine 30⁶⁵⁷ and triciribine-5′-monophosphate 33⁶⁶⁹ were originally prepared from the intermediate 322 by Townsend and Schram (Scheme 59). Aryl chloride displacement of 322 followed by tricycle formation led to triciribine 30, and subsequent phosphorylation produced triciribine-5′-monophosphate 33.

Scheme 59. Early Synthesis of Triciribine 30 and Triciribine-5′-monophosphate 33^a.

^aReagents and conditions: (a) CH₃NHNH₂, EtOH, rt, 3 min; (b) H₂O, reflux, 16 h; (c) POCl₃, PO(OMe)₃, 0 °C, 16 h, 21%;

More recently Shen et al. developed a new approach to prepare triciribine 30 from commercially available intermediates in only 4 steps (Scheme 60).⁶⁵⁹ Thus, reaction of 1-O-acetyl-2,3,5-tri-O-benzoyl-β-d-ribofuranose with 6-chloro-7-iodo-7-deazapurine 324 under classical glycosylation conditions gave intermediate 325, which was then selectively cyanated by treatment with tributyltin cyanide in the presence of catalytic Pd(0) to give compound 326. Aryl chloride displacement with methyl hydrazine followed by deprotection and ring closure with sodium methoxide in refluxing methanol afforded triciribine 30 in a 37% overall yield.

Scheme 60. Synthesis of Triciribine 30^a.

^aReagents and conditions: (a) BSA, 1-O-acetyl-2,3,5-tri-O-benzoyl-β-D-ribofuranose, TMSOTf, CH₃CN, 80 °C, 80%; (b) Bu₃SnCN, Pd(PPh₃)₄, DMF, 95 °C, 6 h, 72%; (c) H₂NNHCH₃, C₂H₅OH, CHCl₃, rt, 3.5 h, 92%; (d) NaOMe, MeOH, rt, 1 h, then reflux, 18 h, 74%.

7.3. MLN4924 Biology

Whereas the function of most FDA-approved nucleoside analogs involves their incorporation into DNA or RNA after phosphorylation, MLN4924 (44; Pevonedistat; TAK-924) suppresses tumor development mainly through blocking the proteasomal degradation pathway. MLN4924 44 is a structural analog of adenosine-5′-monophosphate. MLN4924 44 tightly binds to NEDD8-activating enzyme (NAE), which is an ubiquitin-like protein involved in the ubiquitin-proteasome degradation system (Figure 18).³⁷ An initial high throughput screening of adenosine-5′-monophosphate analogs identified N⁶-benzyl adenosine as an inhibitor of NAE, and further structural elaboration lead to MLN4924 44.³⁷

Figure 18.

Structures of MLN4924 44 and adenosine-5′-monophosphate.

Functionally, MLN4924 44 was shown to selectively inhibit NAE, and did not inhibit transfer RNA synthetases or other ATP-consuming enzymes.³⁷ NAE is the enzyme that initiates the important pathway to maintain protein homeostasis, by forming a tight binding adduct with NEDD8.^670,671 This is a multistep enzymatic process in which Cullin-RING E3 ligases (CRLs) add the NEDD8 polypeptide onto target proteins.⁶⁷² MLN4924 44 was shown to block polypeptide addition to target proteins. Furthermore, in cancer cells MLN4924 44 blocks the important homeostatic balance between protein synthesis and degradation, which is essential for cell survival, cell division, and regulating both innate and adaptive immune responses.^673–675 As a result, CRLs are deactivated and consequently multiple known substrates of CRLs accumulate inside the target cell (e.g., DEPTOR and HIF-1α), leading to cell cycle arrest and apoptosis of the cancer cells (Figure 17).^37,670,676

Cells in the S phase are more sensitive to exposure to MLN4924 44, a unique feature of the agent.³⁷ A recent study using the A375 human melanoma-derived cell line indicated that MLN4924 44 induced cell cycle arrest might largely be triggered by a p21-mediated intra-S phase checkpoint.⁶⁷⁶ Parallel to this study, an in vivo study with diffuse large B cell lymphoma models indicated that MLN4924 44 might also suppress tumor growth via its inhibition of the NF-κB pathway.⁶⁷⁷ Studies in mice demonstrated good toleration for MLN4924 44, as well as potent antiproliferative activity against human tumor cell xenografts derived from human colorectal and lung carcinomas.³⁷ However, HLC116 cells, a human colorectal carcinoma cell line, exposed to MLN4924 44 developed a point mutation at A171T in the UBS3 subunit of NAE, which greatly decreased sensitivity to the drug.⁶⁷⁸ Another study attempted to address a similar issue and found that both HIF1-REDD1-TSC1-mTORC1 and DEPTOR might play important roles in MLN4924-induced autophagy, which is a survival strategy in cancer cells in response to chemotherapy-mediated DNA damage.⁶⁷⁹ This study suggested that cells exposed to an autophagy inhibitor may have a marked improvement in MLN4924 44 response.⁶⁷⁹

MLN4924 hydrochloride affords better solubility than the parent compound, MLN4924 44. However, clinical studies indicate that MLN4924 44 pevonedistat is being administered. Currently the safety and efficacy of MLN4924 44 therapy are being assessed in several Phase I clinical studies involving solid tumors, melanoma, acute myeloid leukemia, refractory or relapsed/refractory multiply myeloma, non-Hodgkin’s lymphoma, Hodgkin’s lymphoma, and cisplatin resistant ovarian cancers.^680,681 Current studies also include investigating fluconazole and itraconazole (antifungals often administered with cancer chemotherapeutics) effects on MLN4924 44 safety and efficacy in solid tumors. Phase I trials using MLN4924 44 in combination therapy with docetaxel (antimitotic taxane), gemcitabine hydrochloride 9, or a combination of carboplatin (DNA alkylating agent) and paclitaxel (antimitotic taxane) in solid tumors are underway (ClinicalTrials.gov identifiers: NCT02122770–recruiting patients, and NCT01862328–active status–not recruiting patients).

7.4. MLN4924 Synthesis

A preparation of MLN4924 44 is summarized in Scheme 61.^682,683 Starting from diene 217, ring closing metathesis led to the tertiary β-alcohol product 328, which had the required stereochemistry that allowed oxidative rearrangement to cyclopentenone 329.⁶⁸² Stereoselective catalytic hydrogenation, due to hydrogen incorporation from the convex face of the intermediate, followed by Luche reduction with CeCl₃, provided 330 as a single diastereomer. Regioselective cleavage of the acetonide group of 330 was followed by treatment with SOCl₂ and RuCl₃/NaIO₄ to furnish the cyclic sulfate 331. Subsequent condensation with N⁶-indanyl-7-deazaadenine in the presence of NaH afforded the nucleoside analogue 332. Standard radical-mediated deoxygenation and O-TBDPS deprotection provided the intermediate 333, which was then reacted with chlorosulfonamide and trifluoroacetic acid to furnish MLN4924 44.⁶⁸³

Scheme 61. Synthesis of MLN4924 44^a.

^aReagents and conditions: (a) Grubbs’ catalyst (2^nd generation), DCM, rt, 2 days, 95%; (b) PDC, DMF, rt, 2 days, 84%; (c) H₂, Pd/C, MeOH, rt, overnight, 100%; (d) NaBH₄, CeCl₃·H₂O, MeOH, 0 °C to rt, 0.5 h, 98%; (e) Me₃Al, DCM, 0 °C to rt, 2 days, 62%; (f) SOCl₂, Et₃N, 0 °C, 10 min, then RuCl₃·3H₂O, NaIO₄, CCl₄, CH₃CN, H₂O, 87%; (g) N⁶-indanyl-7-deazaadenine, NaH, 18-crown-6, THF, 80 °C, overnight, then concd HCl, 80 °C, 2 h, 65%; (h) PhOC(S)Cl, DMAP, DCM, rt, overnight, then Bu₃SnH, AIBN, PhMe, 110 °C, 1 h, 82%; (i) HF-pyridine, THF, pyr, rt, 1 h, 99%; (j) NH₂SO₂Cl, Et₃N, CH₃CN, 0 °C to rt, 1 h, 92%; (k) TFA, rt, 2 h, 90%.

In 2007, Langston et al. published an alternative synthesis of MLN4924 44 (Scheme 62).⁶⁸⁴ The starting diol 334,⁶⁸⁵ formed in a four step sequence from cyclopentadiene and glycolic acid involving a hetero-Diels–Alder and oxidation/reduction reactions, was oxidized and treated with p-anisaldehyde to produce the cyclic epoxide 335. Condensation with pyrrolopyrimidine 337 was followed by deoxygenation, deprotection, selective 5′-OH silylation, and 3′-OH acetylation. Finally, desilylation, sulfamate ester formation, and deacetylation led to MLN4924 44.

Scheme 62. Synthesis of MLN4924 44^a.

^aReagents and conditions: (a) m-CPBA, DCM, 4 h, 76%; (b) 1-(dimethoxymethyl)-4-methoxybenzene, PPTS, rt, overnight, 78%; (c) (S)-(+)-1-aminoindan, DIPEA, 1-butanol, reflux, 60 h, 80%; (d) NaH, DMF, 110 °C, 2 h, 69%; (e) DMAP, PhOC(S)Cl, DCM, rt, 1 h, 99%; (f) AIBN, Bu₃SnH, PhMe, reflux, 30 min, 79%; (g) AcOH, THF, H₂O, rt, 60 h, 98%; (h) 1H-imidazole, DMAP, TBDMSCl, DMF, rt, 2 h, then DMAP, Ac₂O, pyr, rt, overnight, 86%; (i) HF-pyridine, THF, pyr, rt, 1 h, 80%; (j) chlorosulfonamide, Et₃N, MeCN, rt, 1 h, 94%; (k) NH₃/MeOH, rt, 5 days, 90%.

In 2015, a 100-g-scale process synthesis of MLN4924 hydrochloride 350 was reported (Scheme 63).⁶⁸⁶ The key intermediate amino-diol 346 can be prepared from commercially available carboxylic acid 342 in four steps: bromolactonization, reductive ring opening, Boc deprotection, and hydrogenation. Subsequent coupling of 346 with acetaldehyde and (S)-(+)-1-aminoindane afforded diol 348. Selective sulfamoylation of diol 348 was carried out using a Burgess-type reagent 349, which was produced on a 100-kg-scale by reaction of chlorosulfonyl isocyanate with tert-butanol and subsequent treatment with DABCO. MLN4924 44 was then separated from the disulfamoylated byproduct by stepwise hydrolysis with increasing concentration of HCl, requiring no chromatography. Finally, salt formation and crystallization from ethanol furnished MLN4924 hydrochloride 350.

Scheme 63. Synthesis of MLN4924 Hydrochloride 350^a.

^aReagents and conditions: (a) Br₂, py, DME, H₂O, 0 °C, 68%; (b) LiBH₄, THF, H₂O, −5 °C; (c) 8.84 M HBr, i-PrOH, 55 °C; (d) H₂, Pd/C, DIPEA, MeOH, 25 °C, 80% (over three steps); (e) 2-(4,6-dichloropyrimidin-5-yl)acetaldehyde, Et₃N, i-PrOH, 75 °C, 78%; (f) (S)-(+)-1-aminoindane hydrochloride, DIPEA, 2-butanol, 130 °C, 81%; (g) 349, MeCN, 53 °C, 5 h, then HCl, 59%; (h) 1.25 M HCl, EtOH, crystallization, 93%.

7.5. EPZ-004777 and EPZ-5676 Biologies

Histone methyl transferases (HMT) use S-adenosyl-L-methionine as a cofactor to methylate histone amino acid side chains, leading to post-translational gene modification. An important cellular HMT is DOT1-like histone H3K79 methyltransferase (DOT1L histone methyltransferase). It targets residue Lys79 of histone 3 (H3K79) for methylation, as illustrated in Figure 19.⁶⁸⁷ The DOT1L histone methyltransferase is unique among HMTs for several reasons, making it an attractive therapeutic target.⁶⁸⁸ The enzyme has recently been implicated as a therapeutic target for acute leukemias bearing rearrangements in the mixed lineage leukemia (MLL) gene.^689,690

Figure 19.

DOT1L-catalyzed methylation of H3K79.

A translocation in the MLL gene occurs in 5–10% of acute leukemias and is especially common in childhood acute leukemias.^691,692 These cancers are particularly aggressive and have a relatively poor prognosis with a five-year survival of <40%.⁶⁹³ In MLL-rearranged leukemia cells, DOT1L histone methyltransferase has been shown to physically bind to a complex of the rearranged MLL protein fused with sequences from genes such as AF4, AF9, AF10, and ENL. This interaction recruits DOT1L, leads to hypermethylation of H3K79 and aberrant expression of leukemogenic genes such as Hoxa9, Hoxa7, and Meis1.⁵⁵⁹ Furthermore, preclinical models have shown that MLL-rearranged leukemias depend on abnormal DOT1L histone methyltransferase methylation of H3K79.³⁹

Epizyme Inc. recently developed a potent and selective inhibitor of DOT1L histone methyltransferase. The compound, EPZ-004777 351 (Figure 19) is a 7-deaza-5′-amino-alkylurea substituted analogue of adenosine. The agent inhibits S-adenosyl methionine in the enzyme-binding pocket. EPZ-004777 binding leads to a conformational change in DOT1L histone methyltransferase, which opens a hydrophobic pocket past the amino acid site and produces additional interaction surfaces.⁶⁹⁴ Consequently, 351 selectively binds to DOT1L histone methyltransferase with high affinity. Furthermore, the agent was shown to selectively inhibit H3K79 methylation in vitro and inhibit the proliferation of MLL-rearranged cells. It also blocked MLL fusion target gene expression and led to prolonged survival in a mouse xenograft model of MLL leukemia.⁶⁹⁵ However, the poor pharmacokinetic properties of EPZ-004777 351 led to weak in vivo activity and precluded clinical development.

To overcome the suboptimal pharmacokinetic profile of 351, Epizyme synthesized additional analogs. In 2013, the company reported promising preclinical data for EPZ-5676 43, which exhibited improved drug-like properties and potency.⁶⁹⁶ Structural modifications were made to the 5′-side chain in order to decrease the number of hydrogen bonds and conformational freedom. EPZ-5676 43 also occupies the S-adenosyl methionine binding pocket of DOT1L histone methyltransferase, and it was found to be a potent and selective inhibitor of the enzyme with a K_i of <0.08 nM, having >37,000-fold binding selectivity against 15 other human methyltransferases.⁶⁹⁶

EPZ-5676 43 demonstrated selective inhibition of MLL-rearranged versus non-MLL-rearranged leukemia cells. In particular, an IC₅₀ of 3.5 nM for the agent was demonstrated in the MV4–11 cell line following 14 days of exposure, with antiproliferative activity being most evident after 7 days.⁶⁹⁶ This delayed onset could be explained by a progressive series mechanism involving a decrease of H3K79 methylation, lowered MLL-fusion target gene expression, and inhibition of the expression of leukemogenic genes.⁶⁹⁶ EPZ-5676 43 produces a concentration-dependent decrease of both cellular methylated H3K79 levels as well as mRNA transcript levels of the MLL-fusion target genes HOXA9 and MEIS1. Continuous intravenous infusion of EPZ-5676 43 (70 mg/kg) for 21 days produced a complete regression of tumors in a rat model of MLL-rearranged leukemia. Little or no tumor regrowth was detected 30 days after treatment cessation. Evaluation of subcutaneous administration of EPZ-5676 43 has also been reported.⁶⁹⁷

However, EPZ-5676 43 still exhibits suboptimal pharmacokinetics with a short plasma half-life and with fecal excretion being the main route of excretion.^698,699 A carbocyclic analogue of EPZ-5676 43 was synthesized 352 in an attempt to produce increased metabolic stability, and this compound (352; see Figure 19) did indeed show increased stability in human plasma and liver microsomes while maintaining a selective DOT1L histone methyltransferase K_i of 1.1 nM.⁵⁵⁹

In September 2012, Epizyme began a phase I clinical trial of EPZ-5676 43 for subjects with MLL-rearranged leukemia. Currently there are two ongoing Phase I clinical studies examining the agent for dose escalation and the first-in-human study for malignancies with rearrangement of the MLL gene (ClinicalTrials.gov identifiers: NCT02141828–competed in June 2016–no results posted, and NCT01684150–competed in February 2016–no results posted).

7.6. EPZ-5676 Synthesis

The synthesis of EPZ-5676 43 is shown in Scheme 64.^700,701 Amine 353 is prepared from adenosine in four steps: 2′,3′-O-isopropylidine protection, installation of a 5′-O leaving group, 5′-azide displacement, and azide reduction. Reductive amination involving 5′-amino 353 and methyl 3-oxocyclobuganecarboxylate produced a cis/trans mixture (separated using chiral HPLC) furnishing intermediate 354. Subsequent N-alkylation, combined reduction, and Wittig olefination, alkene reduction, and saponification furnished intermediate 357. A two-step process was used to form the benzoimidazole ring of 359, without isolation of the intermediate amide 358. Finally, acid-mediated deprotection completed the synthesis of EPZ-5676 43.

Scheme 64. Synthesis of EPZ-5676 43^a.

^aReagents and conditions: (a) methyl 3-oxocyclobutanecarboxylate, Ti(O-iPr)₄, MeOH, 45 °C, 2 h, then NaCNBH₄, rt, overnight, 41%; (b) 2-iodopropane, K₂CO₃, CH₃CN, 95 °C, overnight, 86%; (c) DIBAL-H, DCM, −78 °C to rt, then ethyl 2-(diethoxyphosphoryl)acetate, DBU, LiCl, rt, 1 h, 83%; (d) Pd/C, MeOH, rt, overnight, 78%; (e) LiOH·H₂O, THF, MeOH, rt, overnight, 28%; (f) 4-(tert-butyl)benzene-1,2-diamine, EDC, HOBt, Et₃N, rt, overnight, 50%; (g) AcOH, 65 °C, overnight, 98%; (h) HCl/MeOH, rt, 2 h, then HPLC, 51%.

In 2014, a 700-g-scale process of EPZ-5676 trihydrate was reported (Scheme 65).⁷⁰² Cyclobutanone 366 can be prepared from 360 on kilogram-scale in a six-step sequence: protection, [2+2] cyloaddition, dechlorination, deprotection, acyl substitution, and iron-catalyzed reduction followed by condensation. 353 was subsequently converted to isopropyl amine 367, which was further converted to 368 as a mixture of diastereomers. Deprotection, followed by two crystallizations from MeCN/H₂O, afforded pure EPZ-5676 43.

Scheme 65. Synthesis of EPZ-5676 43^a.

^aReagents and conditions: (a) BnBr, K₂CO₃, TBAI, Me₂CO, rt, 2 d, 92%; (b) Zn(Cu), ClCOCCl₃, CH₃OCH₂CH₂OCH₃, Et₂O, 50 °C, 3 days, 96%; (c) Zn, AcOH, 50 °C, 2 h, 93%; (d) H₂, Pd/C, (CH₃)₂CHCOOCH₃, PhMe, rt, 20 h, then DCHA, 20 °C, 18 h, 85%; (e) dioxane, DMF, (COCl)₂, 20 °C, 18 h, then 4-tert-butyl-2-nitroaniline, dioxane, 20–40 °C, 5 h, then 20 °C, 18 h, 89%; (f) Fe, AcOH, 75 °C then rt, overnight, 95%; (g) STAB, Me₂CO, MeOH, AcOH, 20 °C, 2 h, 95%; (h) STAB, MeCN, 55 °C, 16 h, 83%; (i) HCl, MeOH, 45 °C, 9 h then rt, overnight, 93%; (i) MeCN/H₂O, crystallization, 65%.

7.7. CF102 Biology

In 2007, preclinical studies developed by Can-Fite Biopharma showed that CF102 (2-chloro-N6-(3-iodobenzyl)-adenosine-5′-N-methyl-uronamide) 42 had potent anticancer activity in hepatocellular carcinoma,⁷⁰³ activity against hepatitis B virus, and anti-inflammatory activity (demonstrated in pre-clinical animal models of liver inflammation).⁷⁰⁴

CF102 42 is as antagonist for the A3 adenosine receptor. Hepatocellular carcinoma cells treated with CF102 had Wnt and NF-κB signaling pathways deregulation, which promoted apoptosis.⁷⁰³ Preclinical studies in rats with N1S1 hepatocellular carcinoma tumors, which highly express the A3 adenosine receptor, showed that CF102 42 encouraged tumor apoptosis.⁷⁰³ Furthermore, A3 adenosine receptor expression appears to be inflammation induced.^705,706 The A3 adenosine receptor is also highly expressed on PBMCs in hepatocellular carcinoma patients and patients with rheumatoid arthritis, psoriasis, dry eye syndrome, and glaucoma, in addition to hepatocellular carcinoma and hepatitis.^707,708

Currently, a Phase I/II study for dose escalation using CF102 42 was reported in patients with unresectable hepatocellular carcinoma.⁷⁰⁹ Overall study results showed that CF102 42 was well tolerated and had favorable pharmacokinetic characteristics. A Phase II study examining the safety and efficacy of CF102 42 is being planned (ClinicalTrials.gov identifier: NCT02128958–recruiting patients).

7.8. CF102 Synthesis

Kim et al. reported the discovery synthesis of CF102 42 in 1994 (Scheme 66).⁷¹⁰ Protecting group manipulation of 1-O-methyl-β-D-riboside 370 produced the intermediate 371. Subsequent oxidation and esterification of the 5′-OH followed by methyl amide introduction, 2′,3′-O-benzoylation, and 1′-acetylation produced intermediate 374. Condensation of intermediate 374 with silylated N⁶-(3-iodobenzyl)-2-chloroadenine 376, made by reacting 2,6-dichloropurine 375 with 3-iodobenzylamine, followed by debenzoylation produced CF102 42 in milligram quantities in an overall yield of 1.7%.

Scheme 66. Synthesis of CF102 42^a.

^aReagents and conditions: (a) TBDPSCl, DMAP, DMF, rt, 18 h, 55%;(b) BzCl, py-DCM, 0 °C to rt, 17 h, 99%; (c) n-Bu₄NF, THF, rt, 2 h, 89%; (d) RuO₂, NaIO₄, CHCl₃, CH₃CN, H₂O, rt, 2.5 h, 90%; (e) EDC, DMAP, MeOH rt, 3 h, 72%; (f) MeNH₂, THF, 50 °C, 15 h; (g) BzCl, py-DCM, rt, 3 h, 72% (over two steps); (h) Ac₂O, H₂SO₄, AcOH, rt, 15 h, 34%; (i) 3-iodobenzylamine·HCl, Et₃N, EtOH, rt, 5 d, 60%; (j) HMDS, (NH₄)₂SO₄, reflux, 4 h; (k) TMSOTf, DCE, reflux, 62 h, 33%; (l) NH₃/MeOH, rt, 16 h, 69%.

Hou et al. recently reported an alternative preparation of CF102 42 starting from the commercially available 1,2,3,5-tetra-O-acetyl-β-d-ribofuranose 136β, which is easily made from D-ribose 213 (Scheme 67).⁷¹¹ The synthesis performed the glycosylation before constructing the 5′-methyl amide and furnished the target compound on a gram-scale in 27% overall yield.

Scheme 67. Synthesis of CF102 42^a.

^aReagents and conditions: (a) BSA, TMSOTf, CH₃CN, 2,6-dichloropurine, 50 °C, 18 h, 88%; (b) 3-iodobenzyl amine, Et₃N, EtOH, rt, 2 d, 85%; (c) NaOMe, MeOH, DCM, rt, 1 h, 92%; (d) 2,2-dimethoxypropane, TsOH, DMF, rt, 12 h, 95%; (e) PDC, DMF, rt, 18 h, 70%; (f) MeNH₂·HCl, HOBt, EDAC, DIEA, DCM, DMF, rt, 12 h, 71%; (g) 80% HCO₂H, rt, 12 h, 83%.

8. ANTIVIRAL NUCLEOSIDE ANALOGS BEING EXPLORED FOR CANCER TREATMENT

8.1. Valacyclovir Hydrochloride and Valganciclovir Hydrochloride Biologies

Valacyclovir hydrochloride 17 and valganciclovir hydrochloride 35 are l-valyl ester prodrugs of acyclovir and ganciclovir, respectively. They were FDA-approved in 2009 and 2010, for the treatments of cytomegalovirus and herpes infections, respectively. After oral administration, the parent drug is released in vivo.⁷¹² Importantly the monophosphorylated agents occur by the viral-specific thymidine kinases, making them less toxic to uninfected cells. The monophosphorylated agents are further converted to their active triphosphate forms to block viral DNA synthesis.⁷¹³ Since the viral thymidine kinase does not exist in human cells, gene therapy involving an adenoviral vector (disabled virus) designed to express the herpes simplex virus thymidine kinase gene in combination with valacyclovir hydrochloride 17 or valganciclovir hydrochloride 35 may lead to targeted cancer treatment (see Section 5.1.1.6).⁷¹⁴ Furthermore, bystander killing of cancer cells has been proposed to be one mechanism of action for ganciclovir and acyclovir.^715,716

Several clinical studies are evaluating valacyclovir hydrochloride 17. A Phase I trial is underway using combination therapy with adenovirus expressing viral-specific thymidine kinase (AD-VTK) and radiation for pediatric brain tumors (ClinicalTrials.gov identifier: NCT00634231–active status–not recruiting patients). Another Phase I trial is using intrapleural AD-VTK therapy with valacyclovir hydrochloride 17 in patients with malignant pleural effusion (ClinicalTrials.gov identifier: NCT01997190–active status–not recruiting patients). A Phase I/II trial of the 17 in combination with HSV-TK and brachytherapy for recurrent prostate cancer is ongoing (ClinicalTrials.gov identifier: NCT01913106–recruiting patients), as well as a Phase III trial with AD-VTK and valacyclovir hydrochloride 17 combination therapy for localized prostate cancer (ClinicalTrials.gov identifier: NCT01436968–recruiting patients). Currently, valganciclovir hydrochloride 35 is being used in a Phase IV trial for chronic lymphocytic leukemia, but the action of the agent is to inhibit cytomegalovirus (CMV) replication in an attempt to stop chronic lymphocytic leukemia propagation (ClinicalTrials.gov identifier: NCT01552369–recruiting patients). The agent has also been examined as an add-on treatment for controlling CMV in patients being treated for the malignant brain tumor glioblastoma. Overall, results showed that the addition of valacyclovir hydrochloride 17 to the standard treatment improved patient survival,⁷¹⁷ which has encouraged a discourse on the possibility that CMV infection influences the development of brain tumors.^718,719

8.2. Valacyclovir and Valganciclovir Syntheses

Valganciclovir hydrochloride 35 has been prepared almost exclusively from its parent compound ganciclovir 387, following the chemistry described in Scheme 68.⁷²⁰ Epichlorohydrin 383 was converted to dibenzyl-protected 384, which was subsequently subjected to chloromethylation by treatment with paraformaldehyde and HCl. Exposure of the resulting intermediate to potassium acetate in acetone afforded 385. Acid catalyzed condensation of 385 with diacetylguanine furnished a mixture of N⁷:N⁹ regioisomers, and the desired N⁹ isomer was crystallized from toluene. Debenzylation followed by deacetylation completed the synthesis of ganciclovir 387, on a 250-g-scale.

Scheme 68. Synthesis of Ganciclovir 387^a.

^aReagents and conditions: (a) BnOH, aq NaOH, rt, 16 h, 63%; (b) HCl (g), (CH₂O)_n, DCM, 0 °C, 16 h; (c) KOAc, Me₂CO, rt, 16 h, ~100%; (d) Diacetylguanine, TsOH, sulfolane, 95 °C, 5 days, 31% (N₉-isomer); (e) 20% Pd(OH)₂/C, cyclohexene–EtOH, reflux, 32 h; (f) NH₄OH–MeOH (1:1 v/v), rt, 16 h, 86% (2 steps).

In 2004, a 100-g-scale process synthesis of triacetyl-protected ganciclovir 389 was reported that significantly increased the yield of the base condensation step.⁷²¹ The reaction coupled diacetylguanine 60 and protected glycerol (388 in the presence of TsOH (Scheme 69). This step occurs under thermodynamic conditions, with the desired N⁹ isomer being the thermodynamic product. Recycling of the undesired N⁷ isomer was possible, allowing for increased yields of the desired N⁹ isomer (overall 70%). Compound 389 can be deprotected to furnish ganciclovir 387.

Scheme 69. Synthesis of 389^a.

^aReagents and conditions: (a) TsOH, DMF, 95–100 °C, 40 h; (b) Crystallization, 70% (N₉-isomer).

In 1998, Arzeno et al. reported a preparation of valganciclovir hydrochloride 35 from ganciclovir 387 (Scheme 70).⁷²² Treatment of 387 with trityl chloride and DMAP in DMF followed by coupling with Boc-L-valine furnished the intermediate 390. Subsequent treatment with a TFA/TFE mixture, in order to remove the trityl groups, and hydrogenolysis of the resulting Boc-protected intermediate with palladium on carbon in a methanol/HCl solution afforded kilograms of valganciclovir hydrochloride 35 with an HPLC purity of 98.4%.

Scheme 70. Synthesis of Valganciclovir Hydrochloride 35^a.

^aReagents and conditions: (a) TrCl, DMAP, Et₃N, DMF, 50 °C, overnight; (b) Boc-l-valine, DCC, DMAP, Et₃N, DCM, rt, 16.5 h; (c) TFA, TFE, 20 °C, 5 h; (d) H₂, Pd/C, then HCl/MeOH, 34% (over four steps).

Khanduri et al. reported a preparation of valganciclovir hydrochloride 35 directly from ganciclovir 387 (Scheme 71).⁷²³ Treatment of ganciclovir 387 with Cbz-l-valine, DCC, and DMAP in DMSO led to a mixture of ganciclovir 387, mono-l-valyl ester 391, and bis-l-valyl ester 392. Addition of water and ethyl acetate caused 387 and 391 to precipitate, while 392 was removed in the organic layer. Ganciclovir 387 was removed from the 387/391 mixture by treatment with TFA followed by extraction with dichloromethane. Mono-l-valyl ester 391 was then separated from the organic layer by recrystallization. Subsequent hydrogenolysis of 391 in acidic solution produced valganciclovir hydrochloride 35, which was purified by recrystallization to give the target compound on a 60-g-scale in greater than 99% HPLC purity. Both ganciclovir 387 and bis-l-valyl ester 392 were recovered and recycled.

Scheme 71.

Preparation of Valganciclovir Hydrochloride 35

The discovery synthesis of valacyclovir hydrochloride 17 was reported in 1989 (Scheme 72). Acyclovir 393, which is prepared in ways similar to those of ganciclovir 387, was condensed directly with Cbz-L-valine.⁷²⁴ Palladium-catalyzed deprotection of the intermediate followed by treatment with aqueous HCl produced 17. An alternative preparation used Boc-l-valine instead of Cbz-l-valine, furnishing the compound on a 50-g-scale.⁷²⁵

Scheme 72. Synthesis of Valacyclovir Hydrochloride 17^a.

^aReagents and conditions: (a) Cbz-L-val-OH, DCC, DMAP, DMF, 60 °C to rt, 12 h; (b) Pd/C, H₂, MeOH, THF, aq HCl, 55% (over two steps); (c) SO₂Cl₂, imidazole, MeCN; (d) l-valine, CuSO₄, K₂CO₃, MeCN; (e) Acyclovir, DCC, DMAP, DMF, 10–15 °C, 45 min, 82%; (f) Raney Ni, EtOH, 50–60 °C then aq HCl, 82%.

An alternative route utilized azido-intermediate 395 to avoid protection–deprotection steps that increase risks of racemization (Scheme 72).⁷²⁶ Sodium azide 394 was treated with sulfuryl chloride and imidazole followed by L-valine and copper sulfate, producing the intermediate 395. Condensation of 395 with acyclovir followed by reduction with Rainey nickel furnished valacyclovir hydrochloride 17 in three steps.

8.3. Ribavirin Biology

Ribavirin (1-β-d-ribofuranosyl-1,2,4-triazole-3-carboxamide) 40 was first discovered as an effective antiviral agent against influenza A and B viruses in the 1970s using cell lines and and in vivo models.^727,728 It was FDA-approved 15 years ago for the combination treatment of hepatitis C virus.⁷²⁹ The agent’s antiviral activity is primarily attributed to ribavirin-5′-triphosphate, which can inhibit RNA polymerase.⁷³⁰ Ribavirin-5′-monophosphate is also a potent inhibitor of inosine monophosphate dehydrogenase (IMPDH),⁷³¹ an enzyme involved in the rate-limiting step of purine de novo synthesis. It is an important factor in maintaining guanine nucleoside pools, and a virus is highly dependent on this process. IMPDH Type I is ubiquitously expressed in normal cells, whereas a high level of IMPDH Type II is also associated with many hematological cancers.⁷³² In addition, ribavirin 40 has a negative impact on eukaryotic translation initiation factor eIF4E activity, which is dysfunctional in 30% of cancers.^733–735 High levels of eIF4E have been correlated with clinical progression, increased angiogenesis, and poor prognosis in breast cancer,^736,737 and ribavirin 40 was shown to suppress growth in breast cell lines.⁷³⁸

Ribavirin 40 is currently being evaluated in a Phase I/II study for malignant solid tumors (ClinicalTrials.gov identifier: NCT01309490–unknown status–no updates for over 2 years) and in a Phase I trial using combination therapy for head and neck cancer and squamous cell cancer (ClinicalTrials.gov identifier: NCT01721525–active status–not recruiting patients). Ribavirin 40 is also being examined in combination with an inhibitor of hedgehog proteins, which regulate cell growth, differentiation, and survival, with or without azacytidine 11 in acute myelogenous leukemia patients (ClinicalTrials.gov identifier: NCT02073838–recruiting patients). The pharmacodynamic effects of ribavirin 40 therapy are being investigated for patients with tonsil and/or base of tongue squamous cell carcinoma (ClinicalTrials.gov identifier: NCT01268579–active status–not recruiting patients). Ribavirin 40 is being evaluated in combination with decitabine 14 in high-risk acute myelogenous leukemia patients (ClinicalTrials.gov identifier: NCT02109744–recruiting patients). Furthermore, a Phase I/II study of ribavirin 40 given as monotherapy in solid tumor cancer patients is ongoing (ClinicalTrials.gov identifier: NCT01309490–unknown status–no updates for over 2 years).

8.4. Ribavirin Synthesis

Witkowski et al. reported the discovery synthesis of ribavirin 40 in 1972 (Scheme 73).^727,739 The compound was prepared by acid-catalyzed fusion of acylated sugar 136β with 1,2,4-triazole-3-carboxylate 397, which was prepared by reacting aminoguanidine bicarbonate with oxalic acid and subsequent esterification of the resulting 1,2,4-triazole-3-carboxylic acid. Treatment of methyl ester 398 with methanolic ammonia effected concomitant deprotection and amide formation, completing the synthesis.

Scheme 73. Synthesis of Ribavirin 40^a.

^aReagents and conditions: (a) SnCl₄, DCM, 15–20 ° C, then reflux, 2 h, 70%; (b) NaOMe/MeOH, 10 °C, 3 h then NH₃/MeOH, 20 °C, 4 h, 79%.

In 2007, Banfi et al. reported a large-scale synthesis, which allowed for the preparation of ribavirin 40 on a 400-g-scale (Scheme 74).⁷⁴⁰ Improvements over other preparations that allowed for scale up include the following: (1) a glycosylation involving direct coupling of triazole 397, thereby avoiding the need for silylating reagents, (2) intermediates pure enough to be used directly in the following stages without difficult purifications, and (3) relatively mild reaction conditions.

Scheme 74. Large-Scale Synthesis of Ribavirin 40^a.

^aReagents and conditions: (a) TfOH, 135 °C, 2 h, vacuum, 92%; (b) NH₃/MeOH, 20 °C, 40 h, 83%.

Another recent preparation makes use of cheap nucleosides as starting materials.⁷⁴¹ Inosine, adenosine, and guanosine were cleaved to the tetra-acylated sugar 136β and the corresponding acetylated purine base by treatment with acetic acid and TFA as catalyst. The glycosylation between 136β and triazole 397 was catalyzed by TfOH, and after deprotection, ribavirin 40 was produced on a 30-g-scale.

Ribavirin 40 can also be synthesized enzymatically. It has been formed on gram-scale. A transglycosylation involving purine nucleoside phosphorylase, 7-methylguanosine and 1,2,4-triazole-3-carboxamide has been used to produce grams of the agent.⁷⁴² Furthermore, purine nucleoside phosphorylase has been used in conjunction with phosphopentomutase to couple ribose-5′-monophosphate 399 with the triazole 400, forming ribavirin 40 (Scheme 75).⁷⁴³

Scheme 75. Synthesis of Ribavirin 40^a.

^aReagents and conditions: (a) Phosphopentomutase, purine nucleoside phosphorylase, tris-buffer, 100%.

8.5. Acadesine Biology

Acadesine (AICA-Ribonucleotide (AICAR); 5-amino-1-β-d-ribofuranosyl-imidazole-4-carboxamide, 41) was previously under investigation for preventing ischemia-reperfusion injury in persons undergoing coronary artery bypass graft surgery.⁷⁴⁴ The agent was found to exert its cardio-protective effect via regulation of adenosine monophosphate activated protein kinase (AMPK) activity.⁷⁴⁵ Acadesine-5′-monophosphate is a highly potent activator of AMPK, and it can bind to the allosteric site of the enzyme as an AMP mimic. In a recent study, this activation was found to exert anti-proliferative and apoptotic effects on acute lymphocytic leukemia cells by regulating different signaling pathways (Figure 17).^746,747 For this reason, cell-specific cytotoxicity of acadesine 41 might ascertain its future usage in anti-cancer therapy.

Preclinical trials with using acadesine 41 and rituximab showed synergistic anticancer activity in tissue culture and animal models.⁷⁴⁸ Acadesine 41 therapy is currently being evaluated in a Phase I/II study for myelodysplastic syndromes and acute myelogenous leukemia (ClinicalTrials.gov identifier: NCT01813838–trial status suspended in 2015) and a Phase I/II open label dose escalation study for safety and tolerability in B cell chronic lymphocytic leukemia patients (ClinicalTrials.gov identifier: NCT00559624–completed in 2010). Clinical trial results indicated that acadesine 41 had a manageable toxicity profile and might be a valuable agent for the treatment of relapsed/refractory CLL patients.⁷⁴⁹

8.6. Acadesine Synthesis

Acadesine 41 is now produced industrially from the fermentation broth of glucose,^750–752 but Baddiley et al. reported the first synthesis of the agent in 1959 (Scheme 76).⁷⁵³ The key intermediate 403 was synthesized from d-ribose 213 in a five-step sequence involving cyclic acetal formation and benzoylation, followed by anomeric bromination, acetylation, and chlorination.⁷⁵⁴ In the presence of silver chloride, protected chlorosugar 403 was condensed with amide 405, formed from 4-nitro-5-styrylglyoxaline 404 in three steps via a carboxylic acid and a methylester intermediate.⁷⁵⁵ The resulting 1:1 mixture of regioisomers 406a and 406b was separated using descending paper chromatography. Hydrogenation of 406b using Adams’ catalyst afforded acadesine 41.

Scheme 76. Synthesis of Acadesine 41^a.

^aReagents and conditions: (a) cat. HCl, MeOH; (b) BzCl, pyr, DCM; (c) HBr/HOAc; (d) Ac₂O, pyr, 48 h, 57% (from D-ribose); (e) HCl/ether, rt, 7 days; (f) AgCl, xylene, reflux, 1 h; (g) NH₃/MeOH, 90 h, 19% (406a:406b = 1:1); (h) H₂ (1 atm), PtO₂, H₂O, 1 h, 12%.

Kohyama et al. reported a gram-scale synthesis of acadesine 41, which features a hydrolysis of N-MEM protected inosine 408 by treatment with aqueous ammonia in methanol (Scheme 77).⁷⁵⁶ The resulting deprotected triol was subjected to aqueous alkaline hydrolysis, furnishing the ring-opened product acadesine 41 in four steps from inosine.

Scheme 77. Synthesis of Acadesine 41^a.

^aReagents and conditions: (a) MEMCl, DIPEA, DCM, 0 °C, 65 min, 86%; (b) NH₄OH, MeOH, rt, 1 h; (c) aq NaOH, reflux, 1 h, 73%.

9. CONCLUSIONS

Nucleoside analogs are used to treat a high percentage of clinical cancers, usually in combination with other agents. They interact with various biological targets, such as cellular kinases, ribonucleotide reductase, and cellular polymerases in order to produce cytotoxic effects. Overall, a majority of cancers treated with nucleoside analogs are haematological. Newer nucleoside analogs, however, such as RX-3117 29, triciribine-5′-monphosphate 33, NUC-1031 32, and NUC-3373 37 are exhibiting promising activity against solid tumors in patients, but further testing is needed. With the exceptions of 6-MP 1 and cytarabine 4, nucleoside analogs usually do not produce cures as standalone agents.

As is the case with most classes of compounds explored in the clinic, attrition rates for nucleoside analogs remains high. Continuing efforts are directed toward improving the selectivity and targeting of anticancer nucleoside analog therapies. Nucleoside analogs interact with various biological targets, such as cell membrane transporters for cellular uptake, kinases to promote their phosphorylation, as well as DNA and RNA polymerases in order to produce their cytotoxic effects. Therefore, the tendency to increase chemical modifications to the parent nucleoside analog in hopes to design a better derivative should be considered with caution. Major efforts to improve analog biological efficacy currently focus on developing new prodrug strategies and using antimetabolites in combination treatments.

Prodrug strategies have emerged as promising approaches in the search for new anti-cancer nucleoside analogs. Next generation nucleoside analogs have been developed in an attempt to overcome cellular uptake deficiencies (e.g., elacytarabine 51, SGI-110 22, and sapacitabine 25) as well as overcome sensitivity to cytidine deamination and adenosine deamination (e.g., FdCyd 38, MB 07133 34, fludarabine-5′-monophosphate 7, clofarabine 12). Compared to the parent compound, nucleoside analog prodrug candidates generally offer enhanced efficacy and less toxicity, with the most successful being capecitabine 10.

Nucleoside analog prodrugs that have progressed to Phase III trials include elacytarabine 51, SGI-110 22, and sapacitabine 25. Elacytarabine 51 and CP-4126 49 have stalled in development, with the former failing to achieve end point efficacy in a Phase III clinical trial involving elderly patients with acute myeloid leukemia,⁴⁹⁰ indicating that 5′-elaidic acid ester conjugates are not successful clinical candidates. Elacytarabine 51 is still under clinical investigation, but questions remain about its clinical prospects as a monotherapy. The N⁴-alkylester moiety was not a successful approach for LY2333727 52, but it may be for sapacitabine 25, which is still being investigated in a Phase III clinical trial as a monotherapy. Furthermore, in June 2016, Cyclacel Pharmaceuticals announced positive results from a Phase I clinical trial involving sapacitabine 25 with seliciclib (CDK inhibitor) treatment for advanced solid tumors. SGI-110 22 is also currently being evaluated in a Phase III clinical trial in patients with AML. In addition, in May 2016 BioCryst Pharmaceuticals submitted regulatory paperwork to obtain forodesine hydrochloride 36 approval in Japan.

The phosphoramidate prodrug strategy has been successful in HCV treatment.⁷⁵⁷ Currently, ProTide analogs NUC-3373 37 and NUC-1031 32 are still being evaluated in Phase I clinical trials. Furthermore, the phosphoramidate prodrugs of many of the nucleoside analogs in this review (e.g., cladribine 8, fludarabine-5′-monophosphate 7, clofarabine 12, thiarabine 47, troxacitabine 48, etc.) have been claimed in patents, but no interesting biological effects have been reported.^433,494 Beyond prodrugs, parent nucleoside analogs such as L-nucleosides seem to have an uncertain future, given the poor performance of troxacitabine 48, which produced multiple terminated clinical trials.

There are several other prodrug strategies being evaluated in preclinical studies. SGI-110 22 employs a duplex linked drug of decitabine(3′−5′)deoxyguanosine. Additional duplex drug linking of 5FdU(5′−5′)TAS-106⁴⁰¹ and 5-FdU(3′−5′)TAS-106⁴⁰² has been investigated in vitro. They may provide a means of delivering two nucleoside analogs at once to allow targeting of different cellular pathways. Preliminary data of the duplex drug linking approach seem promising, but further evidence is needed to show synergy and clinical efficacy. Other unique ways to modify nucleoside analogs being tested in vitro are mesoporous silica nanoparticles,³⁶¹ 5′-mono-amino acid monoester nucleoside prodrugs,³⁶² and Hoechst nucleoside conjugates.³⁶³

A better understanding of cellular degradation pathways may greatly aid in maintaining high intracellular triphosphate forms of the anticancer nucleosides. SAMHD1 hydrolyzes dNTPs into 2′-deoxynucleoside and inorganic triphosphates.^49,50 The tight catalytic pocket of the enzyme excludes ribonucleoside-5′-triphosphates.⁷⁵⁸ Clofarabine-5′-triphosphate was shown to be a substrate for SAMHD1 in a biochemical assay.⁵¹ Therefore, we postulate that nucleoside analog triphosphates such as decitabine-5′-triphosphate, cladribine-5′-triphosphate, floxuri-dine-5′-triphosphate ara-C-5′-triphosphate, ara-A-5′-triphosphate, ara-G-5′-triphosphate, fludarabine-5′-triphosphate, cladribine-5′-triphosphate, and clofarabine-5′-triphosphate may be sensitive to dNTP hydrolysis activity by SAMHD1 both in vitro and in vivo. Cotreatment of these nucleoside analogs with an effective inhibitor of SAMHD1, which still has not been developed, may increase the intracellular antimetabolite concentrations. Moreover, decreasing SAMHD1 protein levels have been shown to promote dNTP imbalance both in vitro and in vivo,^759,760 and inhibition of the dNTP hydrolytic activity of the enzyme might lead to higher mutation rates in cancer cells. An alternative approach has been suggested to increase SAMHD1 expression in cancer cells. Methylation of the promoter of SAMHD1 has been shown to decrease mRNA and protein levels in certain cancers,^53,761,762 and gene dosage of SAMHD1 has been shown to influence cellular dNTP levels in a mouse model.⁷⁶³ The current concept is to increase the SAMHD1 protein level within cells in order to decrease cellular dNTPs, which, in turn, might inhibit cell proliferation.^764,765

Continued investment in basic synthetic organic chemistry provides the ability to access novel nucleoside analogs on both discovery and production scales. For example, the Jamison group’s syntheses of capecitabine 10 and doxifluridine 23 (see section 5.1.2.2) using continuous flow chemistry show great potential for the synthesis of nucleoside, nucleotide, and base analogs. Flow chemistry is widely used in academia and currently is attracting great interest in the pharmaceutical industry (e.g., establishment of the MIT-Novartis Center for Continuous Manufacture).⁷⁶⁶ This burgeoning technology offers great potential for both microscale synthetic drug discovery efforts and macroscale production of active pharmaceutical agents. On the laboratory scale, the microreactor allows for rigorously controlled reaction conditions, leading to better heat and mass transfer, while also requiring much smaller reaction volumes.⁷⁶⁷ This allows for much greater efficiency in comparison to batch systems—a reduction in reaction times (minutes versus hours), and a significant drop in waste (particularly solvent) production. On the macroscale, the significantly reduced reaction time has the potential to produce larger quantities of drug per unit time, in comparison to batch. Furthermore, the continuous flow process joins well with other enabling technologies (microwave irradiation, electrochemistry, 3D-printing, etc.), potentially leading to fully automated processes.⁷⁶⁸ Although questions remain about anomeric selectivity in glycosylations involving 2′-deoxysugars as well as phosphoramidate prodrug synthesis, application of this technology to nucleoside analog drug discovery and drug production seems promising.

The recent FDA approval of TAS-102 (15 + 16) illustrates that possibilities still exist for anticancer nucleoside analogs. The wealth of information available for these analogs produced from decades of research offers a strong foundation for future pursuits, including prodrug and combination strategies. In addition, further development of novel nucleoside analog chemical entities as well as continued refinement of biological applications of current nucleoside analogs will continue to contribute positively to the fight against cancer.

Biographies

Jadd Shelton received his B.S. in Biology (Chemistry minor) from Brigham Young University—Idaho in 2008, where he worked in the lab of Jason Hunt, studying mink glycogen metabolism as a result of hormonal treatment. Also as an undergraduate, and NIH-funded Idaho INBRE fellow in the laboratory of Ron Strohmeyer at Northwest Nazarene University, he researched pharmacologic manipulation of microglial cell inflammatory gene expression related to Alzheimer’s disease. In 2012, he received his Ph.D. in Chemistry from Brigham Young University, working in the lab of Matt Peterson, where his research focused on the synthesis of anticancer nucleoside analogs. From 2012 to 2015, Jadd worked as a postdoctoral researcher in the laboratory of Raymond F. Schinazi at Emory University, synthesizing and studying nucleoside analogs for antiviral applications. He currently works as an R&D chemist for Autoliv Inc. His research interests include the design and synthesis of energetic materials for pyrotechnic formulations.

Xiao Lu received his B.S. degree in Pharmaceutical Sciences in 2005 and his M.S. degree in Chemical Biology in 2008 from Peking University. He then earned his Ph.D. degree in Pharmaceutical and Biological Sciences from The University of Georgia in 2012. After finishing his doctoral studies, he joined Prof. Schinazi’s Laboratory of Biochemical Pharmacology (LOBP) as a postdoctoral fellow at Emory University, working on medicinal chemistry for viral infectious diseases. He is currently a postdoctoral researcher at The National Center for Advancing Translational Sciences (NCATS), where he is contributing to developing drugs that block malaria transmission.

Joseph A. Hollenbaugh received his doctorate from the Department of Cellular and Molecular Biology at University of Vermont in 2006. He was trained as a cellular immunologist under the mentorship Richard Dutton (Trudeau Institute in Saranac Lake, NY) and examined CD8+ T cell responses to EG7OVA tumor in C57BL/6 mice. He did postdoctoral training in immunology and virology in the laboratories of David Topham and Baek Kim (University of Rochester). He is currently an Instructor in the Pediatrics Department, in the Laboratory of Biochemical Pharmacology at Emory University School of Medicine. His research interests include viral biology, cancer biology, and nucleoside biosynthesis regulation at the cellular level. His recent focus has been on the newly identified cellular protein SAMHD1, which degrades nucleoside triphosphates, in order to understand their biological implications in diseases and to develop a nucleoside inhibitor for potential therapeutic applications.

Jong Hyun Cho was born in Jangyumyeon, Gimhae-si, South Korea, in 1967. In 2002, he received his Ph.D. in Organic Chemistry from Seoul National University, working on biologically active peptide mimetics under the direction of B. M. Kim. He then joined David C. K. Chu’s group at the Department of Pharmaceutical and Biomedical Sciences, College of Pharmacy, University of Georgia, where he studied nucleoside chemistry and medicinal chemistry from 2003 to 2006. He is currently working as an instructor research fellow for Raymond F. Schinazi at Emory University. His research interests include synthetic approaches to nucleosides and peptidomimetics with antiviral activities against DNA and RNA viruses.

Franck Amblard was born in Châteauroux, France. He studied chemistry at the University of Orléans (France), where he received his Ph.D. in 2004 under the guidance of Luigi A. Agrofoglio, working on the synthesis of new nucleoside analogs using metathesis and palladium-catalyzed reactions. In 2005, he moved to the USA to join Raymond F. Schinazi’s research group at Emory University (Atlanta, GA) and worked, as a postdoctoral fellow, on new nucleosides and nucleotides prodrugs. He is now Assistant Professor at the Department of Pediatrics, Emory University School of Medicine. His main research interests include the study of nucleosides analogs and small molecules as potential antiviral agents.

Raymond F. Schinazi is the Frances Winship Walters Professor of Pediatrics, Director of the Laboratory of Biochemical Pharmacology at Emory University, and Adjunct Professor at Georgia State University. He also serves as Director of the HIV Cure Scientific Working Group within the NIH-sponsored Emory University Center for AIDS Research. Dr. Schinazi is a world leader in the area of nucleoside chemistry and biology, having published more than 500 papers and having 100 US patents issued. He is the founder of several biotechnology companies focusing on antiviral drug discovery and development, including Pharmasset, Inc., Triangle Pharmaceuticals, Idenix Pharmaceuticals, and Cocrystal Pharmaceuticals, Inc. He is best known for his innovative and pioneering work on stavudine, lamivudine, emtricitabine, telbivudine, and sofosbuvir, all of which have been approved by the United States Food and Drug Administration and are widely used clinically. More than 94% of HIV-infected individuals take at least one of the antiretroviral agents he invented as a fixed dose combination.