Recent Advances in the Chemical Synthesis and Evaluation of Anticancer Nucleoside Analogues

Nucleoside analogues have proven to be highly successful chemotherapeutic agents in the treatment of a wide variety of cancers. Several such compounds, including gemcitabine and cytarabine, are the go-to option in first-line treatments. However, these materials do have limitations and the development of next generation compounds remains a topic of significant interest and necessity. Herein, we discuss recent advances in the chemical synthesis and biological evaluation of nucleoside analogues as potential anticancer agents. Focus is paid to 4′-heteroatom substitution of the furanose oxygen, 2′-, 3′-, 4′- and 5′-position ring modifications and the development of new prodrug strategies for these materials.


Introduction
A significant proportion of current chemotherapeutic treatments for cancer involve the use of anti-metabolites, particularly modified nucleoside analogues that possess a capability to mimic native purine or pyrimidine nucleosides which can disrupt metabolic and regulatory pathways [1]. These molecules can be taken up by nucleoside transporters and then phosphorylated to their mono-, di-and triphosphate forms where they are able to interfere with DNA/RNA synthesis and repair; for example, by acting as chain terminators [2] or ribonucleotide reductase inhibitors [3]. Other notable modes of action include epigenetic regulation, through inhibition of DNA regulatory proteins, such as DNA methyltransferase [4]. Selected current examples of anticancer nucleoside analogues approved for chemotherapeutic treatment regimens include capecitabine, gemcitabine 1, clofarabine 2 and cytarabine (Ara-C) 3 ( Figure 1). Therapeutic intervention using nucleoside analogues is not without its problems and their use is often limited by poor cellular uptake, low conversion to the active triphosphate metabolite, rapid degradation or clearance and development of resistance profiles in certain cell types [5,6]. Consequently, research continues to develop next generations of nucleoside analogues that overcome some of these limitations and provide new therapeutic options.
This class of antimetabolite also possess a proud history, and current frontline treatment, as antiviral [7][8][9][10][11][12] and, more recently, antibacterial agents [13]. Indeed, the development of nucleoside analogues has a symbiotic relationship between compound class and final therapeutic treatment. For example, gemcitabine 1 was developed as an antiviral, but was subsequently shown to be very toxic to leukaemia cells.
In this review, we survey developments from 2010 onwards for the chemical synthesis and evaluation of modified nucleoside analogues for anticancer research. Specifically, focusing on alterations to the native furanose ring ( Figure 2) and generally retaining native purine or pyrimidine nucleobases. Comprehensive reviews concerning hetero-base modifications and general trends in nucleotide synthesis have been covered elsewhere [14,15]. The review is divided into sections that systematically consider: i) furanose 4 -oxygen atom replacements with N, S, Se and C ii) 2 -, 3 -, 4 -or 5 -position furanose ring modifications and iii) new prodrug approaches to deliver nucleoside analogues.
Molecules 2020, 25, x 2 of 24 Therapeutic intervention using nucleoside analogues is not without its problems and their use is often limited by poor cellular uptake, low conversion to the active triphosphate metabolite, rapid degradation or clearance and development of resistance profiles in certain cell types [5,6]. Consequently, research continues to develop next generations of nucleoside analogues that overcome some of these limitations and provide new therapeutic options.
This class of antimetabolite also possess a proud history, and current frontline treatment, as antiviral [7][8][9][10][11][12] and, more recently, antibacterial agents [13]. Indeed, the development of nucleoside analogues has a symbiotic relationship between compound class and final therapeutic treatment. For example, gemcitabine 1 was developed as an antiviral, but was subsequently shown to be very toxic to leukaemia cells.
In this review, we survey developments from 2010 onwards for the chemical synthesis and evaluation of modified nucleoside analogues for anticancer research. Specifically, focusing on alterations to the native furanose ring ( Figure 2) and generally retaining native purine or pyrimidine nucleobases. Comprehensive reviews concerning hetero-base modifications and general trends in nucleotide synthesis have been covered elsewhere [14,15]. The review is divided into sections that systematically consider: i) furanose 4'-oxygen atom replacements with N, S, Se and C ii) 2′-, 3′-, 4′-or 5′-position furanose ring modifications and iii) new prodrug approaches to deliver nucleoside analogues.

Figure 2.
General scope for nucleoside analogues covered in this review. Base = purine or pyrimidine (i.e., C, U, T, A, G or close derivative thereof). X = heteroatom or carbon and Y and Z = ring functional group or modification of native D-ribo stereochemistry.

Azanucleosides
Azanucleosides were originally defined as nucleoside analogues where the furanosyl oxygen is replaced by nitrogen, however this group of analogues has been extended to include nucleosides where the resultant pyrrolidine core has been replaced by other nitrogen containing rings, including heterocycles, heterotricycles and acyclic nitrogen-containing nucleosides [16,17]. This class of compound have proven successful in the treatment of cancer [18] and have also been established as having antiviral and antibacterial properties [19].

Development of Forodesine
Purine nucleoside phosphorylases (PNPs) are responsible for the phosphorolytic metabolism of purine nucleosides to ribose/deoxyribose phosphate and the corresponding nucleobase. Patients with abnormally low levels of PNP possess little T-cell immunity due to a severely reduced degradation of deoxyguanosine, which results in the accumulation of the corresponding triphosphate (dGTP). This then reduces the activity of ribonucleotide reductase and induces apoptosis. As such, human PNP inhibitors are potential treatments for T-cell lymphomas [17]. Immucillin H (forodesine) 4 ( Figure 3) is a highly potent PNP inhibitor (IC50 = 0.48-1.57 nM) which is effective against T-cell malignancies and was found to have excellent oral bioavailability in mice (63%) [20].
Forodesine is a gradual onset drug which binds tightly to PNP with a high affinity (Ki = 0.023 nM) [21]. Whilst clinical development of 4 was discontinued in the US and Europe, it was recently approved for use in the treatment of relapsed/refractory peripheral T-cell lymphoma (PCTL) in Japan (April 2017) [22]. Forodesine is a guanosine analogue and a transition state inhibitor of PNP with 100-1000-fold higher potency than previously identified inhibitors [23]. Due to the combined substitution of the furanosyl oxygen with nitrogen and the C-glycosidic bond, 4 is not incorporated into DNA, acting only as a highly selective PNP inhibitor [22]. Also noteworthy is an adenosine mimetic of 4 which is currently under development as a broad-spectrum antiviral [24].
Forodesine was found to have low oral bioavailability (<11%) in primates, contrary to the case in mice (63% [20]) and was thus originally developed as an intravenous formulation [21]. In 2005, Morris, Jr et al. synthesised BCX-3040, the 2′-deoxy analogue of 4, and comparatively evaluated its  Forodesine is a gradual onset drug which binds tightly to PNP with a high affinity (K i = 0.023 nM) [21]. Whilst clinical development of 4 was discontinued in the US and Europe, it was recently approved for use in the treatment of relapsed/refractory peripheral T-cell lymphoma (PCTL) in Japan (April 2017) [22].
Forodesine is a guanosine analogue and a transition state inhibitor of PNP with 100-1000-fold higher potency than previously identified inhibitors [23]. Due to the combined substitution of the furanosyl oxygen with nitrogen and the C-glycosidic bond, 4 is not incorporated into DNA, acting only as a highly selective PNP inhibitor [22]. Also noteworthy is an adenosine mimetic of 4 which is currently under development as a broad-spectrum antiviral [24].
In 2000, Tyler et al. described a linear synthesis of 4 in a satisfactory 39% yield over 10 steps (Scheme 1) [25]. Starting from 5, synthesised in nine steps by known methods from d-gulonolactone [26], the pyrrolidine was treated with N-chlorosuccinimide (NCS), obtaining the 1-chloro anomeric glycoside which subsequently underwent elimination using lithium tetramethylpiperidine (LiTMP) to afford imine 6. The nucleobase was next assembled via addition of lithiated acetonitrile to afford 7, followed by protection of the furanosyl nitrogen giving 8 and treatment with Bredereck's reagent to afford enamine 9. Acid-catalysed hydrolysis of 9 delivered enol 10 which was reacted with ethyl glycinate to obtain enamine 11. Treatment of 11 with excess benzyl chloroformate and 1,8-diazabicyclo [5.4.0]undec-7-ene (DBU) revealed 12 and subsequent hydrogenolysis of the N-Cbz group provided pyrrole 13. Completion of the carbocyclic nucleobase was achieved via treatment of 13 with formamidine acetate and acidic removal of the silicon, nitrogen and isopropylidene protecting groups, to afford 4. Forodesine is a guanosine analogue and a transition state inhibitor of PNP with 100-1000-fold higher potency than previously identified inhibitors [23]. Due to the combined substitution of the furanosyl oxygen with nitrogen and the C-glycosidic bond, 4 is not incorporated into DNA, acting only as a highly selective PNP inhibitor [22]. Also noteworthy is an adenosine mimetic of 4 which is currently under development as a broad-spectrum antiviral [24].
Forodesine was found to have low oral bioavailability (<11%) in primates, contrary to the case in mice (63% [20]) and was thus originally developed as an intravenous formulation [21]. Forodesine was found to have low oral bioavailability (<11%) in primates, contrary to the case in mice (63% [20]) and was thus originally developed as an intravenous formulation [21]. In 2005, Morris, Jr et al. synthesised BCX-3040, the 2 -deoxy analogue of 4, and comparatively evaluated its oral pharmacokinetic and pharmacodynamic properties in an effort to maintain potency and deliver oral bioavailability [27]. This was hypothesised from 2 -deoxyguanosine exhibiting tight binding to PNP [28] and therefore a possible redundancy for the 2 -OH. Starting from vinyl bromide 14 (Scheme 2), the 9-deazapurine underwent bromine-lithium exchange and addition to imine 6, obtaining nucleoside 15 in 85% yield. Subsequent removal of 5 -O-TBS and 2 ,3 -O-isopropylidene groups gave 16, followed by 3 ,5 -OTIPDS protection to afford 17. The free 2 -hydroxyl group was then converted to thiocarbonate 18 and deoxygenated via treatment with 1,1 -azobis(cyclohexane-1-carbonitrile) in excellent yield (91%) to give 19. Deprotection was completed in two steps, first cleaving the CH 2 OCH 2 Ph (BOM) group, followed by acidic hydrolysis and hydrochloride salt formation to obtain BCX-3040 in 83% yield (from 19).
Molecules 2020, 25, x 4 of 24 oral pharmacokinetic and pharmacodynamic properties in an effort to maintain potency and deliver oral bioavailability [27]. This was hypothesised from 2′-deoxyguanosine exhibiting tight binding to PNP [28] and therefore a possible redundancy for the 2′-OH. Starting from vinyl bromide 14 (Scheme 2), the 9-deazapurine underwent bromine-lithium exchange and addition to imine 6, obtaining nucleoside 15 in 85% yield. In vitro biological evaluation of BCX-3040 confirmed it to be a potent PNP inhibitor, with near identical IC50 values to 4 (BCX-3040 IC50 = 3.1 ± 0.50 nM and 4 IC50 = 1.2 ± 0.21 nM). Administration of 5.0 mg of 4 had a 12.6-fold greater 2′-deoxyguanosine response than administration of 10.0 mg of BCX-3040, indicating a reduced bioavailability for BCX-3040. Furthermore, following IV administration of 5 mg/kg of BCX-3040, the plasma concentration of BCX-3040 dropped rapidly to 3.0 ± 0.31 µg/mL. Overall, due to the observed rapid clearance and reduced bioavailability of BCX-3040, it was concluded to be a poorer PNP inhibitor candidate in comparison to 4.

Thionucleosides
4′-Thiofuranoses are of known importance in biological systems, for example as chemical biology or biomedical tools [29], and possess chemotherapeutic activity [30,31]. Furthermore, the thioaminal moiety within 4′-thiofuranosyl nucleosides has been proven to be resistant to metabolic hydrolysis in comparison to native 4′-oxa analogues [32]. In the early 1990s, Secrist, Montgomery and co-workers stimulated interest in this class of molecule with their synthesis and biological evaluation In vitro biological evaluation of BCX-3040 confirmed it to be a potent PNP inhibitor, with near identical IC 50 values to 4 (BCX-3040 IC 50 = 3.1 ± 0.50 nM and 4 IC 50 = 1.2 ± 0.21 nM). Administration of 5.0 mg of 4 had a 12.6-fold greater 2 -deoxyguanosine response than administration of 10.0 mg of BCX-3040, indicating a reduced bioavailability for BCX-3040. Furthermore, following IV administration of 5 mg/kg of BCX-3040, the plasma concentration of BCX-3040 dropped rapidly to 3.0 ± 0.31 µg/mL. Overall, due to the observed rapid clearance and reduced bioavailability of BCX-3040, it was concluded to be a poorer PNP inhibitor candidate in comparison to 4.

Thionucleosides
4 -Thiofuranoses are of known importance in biological systems, for example as chemical biology or biomedical tools [29], and possess chemotherapeutic activity [30,31]. Furthermore, the thioaminal moiety within 4 -thiofuranosyl nucleosides has been proven to be resistant to metabolic hydrolysis in comparison to native 4 -oxa analogues [32]. In the early 1990s, Secrist, Montgomery and co-workers stimulated interest in this class of molecule with their synthesis and biological evaluation of 2 -deoxy-4 -thiopyrimidine nucleosides [30]. Since then, there has been a resurgence of interest in the synthesis and evaluation of these compounds as potential antiviral and chemotherapeutic agents.
Additionally Finally, thioglycosylation was successfully employed to access the corresponding 4 -thionucleoside mixture 33 (54% yield). Comparatively, the group found that using 32 as a donor and employing silyl-Hilbert-Johnson glycosylation conditions yielded 33α/β in higher yield (93%), but still as a mixture of anomers (2.9:1 α/β). Finally, the group synthesised 2'-deoxy-2'-difluoro-1'(4'-thia-D-ribofuranose)cytosine 37, a thionucleoside analogue of the potent chemotherapeutic agent gemcitabine 1 (Scheme 5). Intermediate 23 was again utilised and oxidised using Albright-Goldman conditions to obtain ketone 34 which was difluorinated at C2 using DAST. The 3-O-benzyl group was next removed and replaced with benzoate to afford 35 in 79% overall yield from 34. Oxidation to the sulfoxide afforded 36 which subsequently underwent Pummerer rearrangement-glycosylation, before the remaining protecting groups were removed to afford 37α/β (α/β = 2.4/1) in a moderate yield of 51% from 36. The antineoplastic activities of 28, 33 and 37 were evaluated and compared to arabinocytidine 3 (Ara-C) and 1-(2-deoxy-2-C-methylene-β-D-erythro-pentofuranosyl)cytosine (DMDC) ( Table 1). As expected, all the α-anomer forms were found to be inactive against T-cell leukemia CCRF-HSB-2 cells. However, the β-anomers showed considerable cytotoxic activity against the same cell line. Notably, 28β and 33β were highly potent against both T-cell leukemia CCRF-HSB-2 cells and human solid tumour KB cells, with IC50 values of 0.01 µg/mL (CCRF-HSB-2) and 0.05 µg/mL (CCRF-HSB-2) for 28β and 33β, respectively. Indeed, the activity of 28β was significantly higher than that of its native counterpart, DMDC, against both cell lines, with an IC50 value 2.4 times lower in CCRF-HSB-2 cells and 3.7 times lower in KB cells. Interestingly, 4′-thiogemcitabine analogue 37β had poorer antineoplastic activity compared to 1, which the authors suggest may be due to a reduction in the phosphorylation efficacy of 37β by deoxycytidine kinase, a key enzyme which converts 2′deoxycytidine analogues to their corresponding monophosphates. Finally, the group synthesised 2 -deoxy-2 -difluoro-1 (4 -thia-d-ribofuranose)cytosine 37, a thionucleoside analogue of the potent chemotherapeutic agent gemcitabine 1 (Scheme 5). Intermediate 23 was again utilised and oxidised using Albright-Goldman conditions to obtain ketone 34 which was difluorinated at C2 using DAST. The 3-O-benzyl group was next removed and replaced with benzoate to afford 35 in 79% overall yield from 34. Oxidation to the sulfoxide afforded 36 which subsequently underwent Pummerer rearrangement-glycosylation, before the remaining protecting groups were removed to afford 37α/β (α/β = 2.4/1) in a moderate yield of 51% from 36. Finally, the group synthesised 2'-deoxy-2'-difluoro-1'(4'-thia-D-ribofuranose)cytosine 37, a thionucleoside analogue of the potent chemotherapeutic agent gemcitabine 1 (Scheme 5). Intermediate 23 was again utilised and oxidised using Albright-Goldman conditions to obtain ketone 34 which was difluorinated at C2 using DAST. The 3-O-benzyl group was next removed and replaced with benzoate to afford 35 in 79% overall yield from 34. Oxidation to the sulfoxide afforded 36 which subsequently underwent Pummerer rearrangement-glycosylation, before the remaining protecting groups were removed to afford 37α/β (α/β = 2.4/1) in a moderate yield of 51% from 36. The antineoplastic activities of 28, 33 and 37 were evaluated and compared to arabinocytidine 3 (Ara-C) and 1-(2-deoxy-2-C-methylene-β-D-erythro-pentofuranosyl)cytosine (DMDC) ( Table 1). As expected, all the α-anomer forms were found to be inactive against T-cell leukemia CCRF-HSB-2 cells. However, the β-anomers showed considerable cytotoxic activity against the same cell line. Notably, 28β and 33β were highly potent against both T-cell leukemia CCRF-HSB-2 cells and human solid tumour KB cells, with IC50 values of 0.01 µg/mL (CCRF-HSB-2) and 0.05 µg/mL (CCRF-HSB-2) for 28β and 33β, respectively. Indeed, the activity of 28β was significantly higher than that of its native counterpart, DMDC, against both cell lines, with an IC50 value 2.4 times lower in CCRF-HSB-2 cells and 3.7 times lower in KB cells. Interestingly, 4′-thiogemcitabine analogue 37β had poorer antineoplastic activity compared to 1, which the authors suggest may be due to a reduction in the phosphorylation efficacy of 37β by deoxycytidine kinase, a key enzyme which converts 2′deoxycytidine analogues to their corresponding monophosphates. The antineoplastic activities of 28, 33 and 37 were evaluated and compared to arabinocytidine 3 (Ara-C) and 1-(2-deoxy-2-C-methylene-β-d-erythro-pentofuranosyl)cytosine (DMDC) ( Table 1). As expected, all the α-anomer forms were found to be inactive against T-cell leukemia CCRF-HSB-2 cells. However, the β-anomers showed considerable cytotoxic activity against the same cell line. Notably, 28β and 33β were highly potent against both T-cell leukemia CCRF-HSB-2 cells and human solid tumour KB cells, with IC 50 values of 0.01 µg/mL (CCRF-HSB-2) and 0.05 µg/mL (CCRF-HSB-2) for 28β and 33β, respectively. Indeed, the activity of 28β was significantly higher than that of its native counterpart, DMDC, against both cell lines, with an IC 50 value 2.4 times lower in CCRF-HSB-2 cells and 3.7 times lower in KB cells. Interestingly, 4 -thiogemcitabine analogue 37β had poorer antineoplastic activity compared to 1, which the authors suggest may be due to a reduction in the phosphorylation efficacy of 37β by deoxycytidine kinase, a key enzyme which converts 2 -deoxycytidine analogues to their corresponding monophosphates. Table 1. Antineoplastic activities of 2 -modified-4 -thionucleosides.

Selenonucleosides
A first synthesis of pyrimidine 4'-selenonucleosides was reported by Jeong and colleagues in 2008 (Scheme 8), [42] starting from lyxose derivative 55, synthesised from D-gulonic-γ-lactone in four steps using established procedures [43]. Selective protection of the primary alcohol in 55 was achieved by reaction with TBDPSCl, giving hemiacetal 56 which was reduced with NaBH4, furnishing diol 57. Following double mesylation of 57 to give 58, cyclisation to give 4-selenosugar 59 was achieved by treatment with selenium in the presence of NaBH4. Oxidation of 59 to a diastereomeric selenoxide mixture 60 then enabled either a uracil or cytosine nucleobase to be installed via a Pummerer-type glycosylation, furnishing the βanomers 61 or 63. 4'-selenouridine 62 was obtained after global deprotection of 61 with 50% aqueous TFA in an overall yield of 12% over 11 steps. 4'-selenocytidine 64 was similarly obtained in an overall yield of 9% over 11 steps. The crystal structure of 62 revealed the non-native ring adopted an unusual C2'-endo/C3'-exo twist (Southern confirmation), contrary to uridine, which shows a C2'-exo/C3'-endo twist (Northern conformation). This difference was explained by the introduction of the bulky selenium, whereby stereoelectronic effects observed in 4'-oxanucleosides are outweighed by the size of the heteroatom.    Selective protection of the primary alcohol in 55 was achieved by reaction with TBDPSCl, giving hemiacetal 56 which was reduced with NaBH 4 , furnishing diol 57. Following double mesylation of 57 to give 58, cyclisation to give 4-selenosugar 59 was achieved by treatment with selenium in the presence of NaBH 4 . Oxidation of 59 to a diastereomeric selenoxide mixture 60 then enabled either a uracil or cytosine nucleobase to be installed via a Pummerer-type glycosylation, furnishing the β-anomers 61 or 63. 4 -selenouridine 62 was obtained after global deprotection of 61 with 50% aqueous TFA in an overall yield of 12% over 11 steps. 4 -selenocytidine 64 was similarly obtained in an overall yield of 9% over 11 steps. The crystal structure of 62 revealed the non-native ring adopted an unusual C2 -endo/C3 -exo twist (Southern confirmation), contrary to uridine, which shows a C2 -exo/C3 -endo twist (Northern conformation). This difference was explained by the introduction of the bulky selenium, whereby stereoelectronic effects observed in 4 -oxanucleosides are outweighed by the size of the heteroatom.

Selenonucleosides
A first synthesis of pyrimidine 4'-selenonucleosides was reported by Jeong and colleagues in 2008 (Scheme 8), [42] starting from lyxose derivative 55, synthesised from D-gulonic-γ-lactone in four steps using established procedures [43]. Selective protection of the primary alcohol in 55 was achieved by reaction with TBDPSCl, giving hemiacetal 56 which was reduced with NaBH4, furnishing diol 57. Following double mesylation of 57 to give 58, cyclisation to give 4-selenosugar 59 was achieved by treatment with selenium in the presence of NaBH4. Oxidation of 59 to a diastereomeric selenoxide mixture 60 then enabled either a uracil or cytosine nucleobase to be installed via a Pummerer-type glycosylation, furnishing the βanomers 61 or 63. 4'-selenouridine 62 was obtained after global deprotection of 61 with 50% aqueous TFA in an overall yield of 12% over 11 steps. 4'-selenocytidine 64 was similarly obtained in an overall yield of 9% over 11 steps. The crystal structure of 62 revealed the non-native ring adopted an unusual C2'-endo/C3'-exo twist (Southern confirmation), contrary to uridine, which shows a C2'-exo/C3'-endo twist (Northern conformation). This difference was explained by the introduction of the bulky selenium, whereby stereoelectronic effects observed in 4'-oxanucleosides are outweighed by the size of the heteroatom.
Molecules 2020, 25, x 10 of 24 A 2'-fluoroarabino analogue was obtained via a 2'-position DAST fluorination of intermediate 69, synthesised from 4'-selenouridine 61 using a four-step process [44]. At first attempt, DAST fluorination of 69 led to the formation of the desired 2'-fluoro product 70 as the minor product in 23% yield, along with a major product, 2,2'-O-anhydro nucleoside 71, in 60% yield (Scheme 9). This ratio was subsequently improved by instead fluorinating an N 3 -benzoyl derivative 72, giving 73 in 45% yield, now as the major product, and 71 in reduced 30% yield.  Table 2). From this study, it was found that 2'-fluoro-4'-selenoarabinocytidine 65 was the most potent analogue, showing even greater potency than 3, in the majority of cell lines tested. The group have also reported further 2'-substituted-4'-selenoribofuranosides [46], synthesising a series of pyrimidine 2'-substituted analogues and utilising a 2,2'-O-anhydro intermediate to enable regioselective nucleophilic ring opening to the desired ribo-configured products (Figure 6). Only a 2'fluoro riboselenocytosine derivative showed significant activity (uracil, thymine and 5-halo uracil derivatives showed no activity up to 100 µM), but this was less potent than the arabino-configured analogue previously described [45]. Following global deprotection and uracil to cytosine conversion, four nucleoside analogues 65-68 were evaluated against several human cancer cell lines (HCT116, A549, SU638, T47D, PC-3 and K562) and compared to the established anticancer nucleosides 1 and 3 (Table 2). From this study, it was found that 2 -fluoro-4 -selenoarabinocytidine 65 was the most potent analogue, showing even greater potency than 3, in the majority of cell lines tested. The group have also reported further 2 -substituted-4 -selenoribofuranosides [46], synthesising a series of pyrimidine 2 -substituted analogues and utilising a 2,2 -O-anhydro intermediate to enable regioselective nucleophilic ring opening to the desired ribo-configured products (Figure 6). Only a 2 -fluoro riboselenocytosine derivative showed significant activity (uracil, thymine and 5-halo uracil derivatives showed no activity up to 100 µM), but this was less potent than the arabino-configured analogue previously described [45].

Carbocyclic Nucleosides
In carbocyclic nucleosides the furanosyl oxygen is replaced by CH2, forming a cyclopentane ring. The lack of a hemiaminal linkage between the nucleobase and the sugar leads to an increased chemical stability. Furthermore, due to the lack of a glycosidic bond, these nucleosides show an enhanced resistance towards phosphorylases [47]. Although considered a second generation of nucleoside analogues, there are two naturally occurring carbocyclic nucleosides, aristeromycin 74 and neplanocin A 75 (Figure 7), both of which exhibit substantial biological activity as antitumor agents [47].  Fluorocyclopentenyl cytosine 76 was synthesised from cyclopentenone 80 (Scheme 10), which was obtained from D-ribose in nine steps using established procedures [52,53]. Iodination of 80 was accomplished by treatment with I2 and pyridine in THF, to give 81. Stereo-and regioselective reduction of 76 was achieved using Luche conditions to give 82, followed by TBDPS protection of the resulting alcohol to deliver 83. Following lithium-halogen exchange of 83 using n-BuLi, electrophilic fluorination was achieved by treatment with N-fluorobenzenesulfonimide (NFSI). Anomeric desilylation with TBAF then gave 84 to which a uracil nucleobase was installed via condensation with N 3 -benzoyluracil under Mitsunobu conditions. Finally, global deprotection was achieved by sequential treatment with methanolic ammonia and BBr3, providing uracil derivate 85 which was converted to 76 via a standard four-step process.

Carbocyclic Nucleosides
In carbocyclic nucleosides the furanosyl oxygen is replaced by CH 2 , forming a cyclopentane ring. The lack of a hemiaminal linkage between the nucleobase and the sugar leads to an increased chemical stability. Furthermore, due to the lack of a glycosidic bond, these nucleosides show an enhanced resistance towards phosphorylases [47]. Although considered a second generation of nucleoside analogues, there are two naturally occurring carbocyclic nucleosides, aristeromycin 74 and neplanocin A 75 (Figure 7), both of which exhibit substantial biological activity as antitumor agents [47].

Carbocyclic Nucleosides
In carbocyclic nucleosides the furanosyl oxygen is replaced by CH2, forming a cyclopentane ring. The lack of a hemiaminal linkage between the nucleobase and the sugar leads to an increased chemical stability. Furthermore, due to the lack of a glycosidic bond, these nucleosides show an enhanced resistance towards phosphorylases [47]. Although considered a second generation of nucleoside analogues, there are two naturally occurring carbocyclic nucleosides, aristeromycin 74 and neplanocin A 75 (Figure 7), both of which exhibit substantial biological activity as antitumor agents [47].  Fluorocyclopentenyl cytosine 76 was synthesised from cyclopentenone 80 (Scheme 10), which was obtained from D-ribose in nine steps using established procedures [52,53]. Iodination of 80 was accomplished by treatment with I2 and pyridine in THF, to give 81. Stereo-and regioselective reduction of 76 was achieved using Luche conditions to give 82, followed by TBDPS protection of the resulting alcohol to deliver 83. Following lithium-halogen exchange of 83 using n-BuLi, electrophilic fluorination was achieved by treatment with N-fluorobenzenesulfonimide (NFSI). Anomeric desilylation with TBAF then gave 84 to which a uracil nucleobase was installed via condensation with N 3 -benzoyluracil under Mitsunobu conditions. Finally, global deprotection was achieved by sequential treatment with methanolic ammonia and BBr3, providing uracil derivate 85 which was converted to 76 via a standard four-step process.

Carbocyclic Nucleosides
In carbocyclic nucleosides the furanosyl oxygen is replaced by CH2, forming a cyclopentane ring. The lack of a hemiaminal linkage between the nucleobase and the sugar leads to an increased chemical stability. Furthermore, due to the lack of a glycosidic bond, these nucleosides show an enhanced resistance towards phosphorylases [47]. Although considered a second generation of nucleoside analogues, there are two naturally occurring carbocyclic nucleosides, aristeromycin 74 and neplanocin A 75 (Figure 7), both of which exhibit substantial biological activity as antitumor agents [47].  Fluorocyclopentenyl cytosine 76 was synthesised from cyclopentenone 80 (Scheme 10), which was obtained from D-ribose in nine steps using established procedures [52,53]. Iodination of 80 was accomplished by treatment with I2 and pyridine in THF, to give 81. Stereo-and regioselective reduction of 76 was achieved using Luche conditions to give 82, followed by TBDPS protection of the resulting alcohol to deliver 83. Following lithium-halogen exchange of 83 using n-BuLi, electrophilic fluorination was achieved by treatment with N-fluorobenzenesulfonimide (NFSI). Anomeric desilylation with TBAF then gave 84 to which a uracil nucleobase was installed via condensation with N 3 -benzoyluracil under Mitsunobu conditions. Finally, global deprotection was achieved by sequential treatment with methanolic ammonia and BBr3, providing uracil derivate 85 which was converted to 76 via a standard four-step process.  Fluorocyclopentenyl cytosine 76 was synthesised from cyclopentenone 80 (Scheme 10), which was obtained from d-ribose in nine steps using established procedures [52,53]. Iodination of 80 was accomplished by treatment with I 2 and pyridine in THF, to give 81. Stereo-and regioselective reduction of 76 was achieved using Luche conditions to give 82, followed by TBDPS protection of the resulting alcohol to deliver 83. Following lithium-halogen exchange of 83 using n-BuLi, electrophilic fluorination was achieved by treatment with N-fluorobenzenesulfonimide (NFSI). Anomeric desilylation with TBAF then gave 84 to which a uracil nucleobase was installed via condensation with N 3 -benzoyluracil under Mitsunobu conditions. Finally, global deprotection was achieved by sequential treatment with methanolic ammonia and BBr 3 , providing uracil derivate 85 which was converted to 76 via a standard four-step process. Of the four nucleoside analogues evaluated (Figure 8), derivative 76 showed significant potency against several human cancer cell lines (Table 3) [51,54,55]. Furthermore, Jeong reported that 76 also showed significant antitumour activity in a nude mouse xenograft model implanted with A549 human lung cancer cells, wherein after 38 days the inhibition of tumour growth was 32% and 58% at 3 and 10 mg/kg doses, respectively [54]. Human cancer cell tissue type a colon; b breast; c pancreatic; d hormone-dependent breast; e lung; f stomach; g brain.
Carbocyclic nucleoside 76 has now been evaluated in more than 100 different cell lines, as well as several xenograft models, showing high potencies against numerous types of cancer, including gemcitabine resistant cell lines [56,57]. Pharmacokinetics and oral bioavailability for 76 were investigated in phase 0 clinical trials, wherein a small cohort of patients were administered a single oral dose (50 mg or 100 mg) of 76, or a single intravenous dose (20 mg, Table 4) [58]. This study found that the absolute bioavailability for 76 was 56% and 33% for 50 and 100 mg doses, respectively, suggesting it not to be dose-proportional. However, t1/2 was found to be 14 h and 21 h for 50 and 100 mg doses, respectively. This may suggest that 76 does exhibit some dose proportionality in some parameters, but not in others and it was noted that this result may be due to the small patient sample size. Analogue 76 is currently in phase II clinical trials for metastatic pancreatic cancer and advanced bladder cancer. Of the four nucleoside analogues evaluated (Figure 8), derivative 76 showed significant potency against several human cancer cell lines (Table 3) [51,54,55]. Furthermore, Jeong reported that 76 also showed significant antitumour activity in a nude mouse xenograft model implanted with A549 human lung cancer cells, wherein after 38 days the inhibition of tumour growth was 32% and 58% at 3 and 10 mg/kg doses, respectively [54]. Carbocyclic nucleoside 76 has now been evaluated in more than 100 different cell lines, as well as several xenograft models, showing high potencies against numerous types of cancer, including gemcitabine resistant cell lines [56,57]. Pharmacokinetics and oral bioavailability for 76 were investigated in phase 0 clinical trials, wherein a small cohort of patients were administered a single oral dose (50 mg or 100 mg) of 76, or a single intravenous dose (20 mg, Table 4) [58]. This study found that the absolute bioavailability for 76 was 56% and 33% for 50 and 100 mg doses, respectively, suggesting it not to be dose-proportional. However, t 1/2 was found to be 14 h and 21 h for 50 and 100 mg doses, respectively. This may suggest that 76 does exhibit some dose proportionality in some parameters, but not in others and it was noted that this result may be due to the small patient sample size. Analogue 76 is currently in phase II clinical trials for metastatic pancreatic cancer and advanced bladder cancer.

Norbornane-Derived (C2 ,C4 -bridged) Carbocyclic Nucleosides
Nencka and colleagues reported the synthesis of norborane, C2 ,C4 -bridged carbocyclic nucleosides [59], with a hypothesis of locking the nucleoside analogue in the 2 -exo or North conformation which has garnered attention in the field of carbocyclic antiviral nucleoside analogue development [60]. To access these compounds their synthesis started from known hydroxy ester 86 which was stereochemically inverted at the C1 position via an oxidation-reduction sequence to deliver 87 (Scheme 11). This enabled the desired stereochemistry to be attained at C1 when subsequently inserting an azide by nucleophilic displacement of a mesylate to afford 88 which was reduced to give 89. With scaffold 89 in hand, the group then elaborated the amine at C1 to several purine and pyrimidine nucleobase forms (including C6 purine analogues) and evaluated their cytotoxic potential.

C3',C5'-Bridged Carbocyclic L-Nucleosides
Tănase and colleagues reported the synthesis of an alternative carbocyclic system based on a similar bicyclo[2.2.1]heptane fragment, accessing a small series of 3'-5'-linked L-nucleoside analogues containing C6-amino modifications (Scheme 12) [61]. Their synthesis started from known alcohol 92 and proceeded through azide incorporation (to give 93) and reduction steps in good yields to deliver amine 94. The 6-chloropurine ring was then introduced using standard methods and elaborated at C6 with a series of amines via nucleophilic aromatic substitution. This small library was then screened in vitro at a single high dose (10 −5 M) in the full NCI 58 human tumor cell screen panel. Phenethylamine derivative 95 exhibited growth inhibition of 74% on SK-MEL-5 melanoma and UO-31 renal cancer cell lines, but was not deemed sufficiently cytotoxic for studies to proceed further. The group followed up this report with further synthesis of 6-position carbocyclic analogues [62], derived from 94, noting that a 6-(4-methoxy-phenethyl)amino group was active, but again the analogue was not progressed further.

C3',C5'-Bridged Carbocyclic L-Nucleosides
Tănase and colleagues reported the synthesis of an alternative carbocyclic system based on a similar bicyclo[2.2.1]heptane fragment, accessing a small series of 3'-5'-linked L-nucleoside analogues containing C6-amino modifications (Scheme 12) [61]. Their synthesis started from known alcohol 92 and proceeded through azide incorporation (to give 93) and reduction steps in good yields to deliver amine 94. The 6-chloropurine ring was then introduced using standard methods and elaborated at C6 with a series of amines via nucleophilic aromatic substitution. This small library was then screened in vitro at a single high dose (10 −5 M) in the full NCI 58 human tumor cell screen panel. Phenethylamine derivative 95 exhibited growth inhibition of 74% on SK-MEL-5 melanoma and UO-31 renal cancer cell lines, but was not deemed sufficiently cytotoxic for studies to proceed further. The group followed up this report with further synthesis of 6-position carbocyclic analogues [62], derived from 94, noting that a 6-(4-methoxy-phenethyl)amino group was active, but again the analogue was not progressed further.

2'-Furanose Modifications
Clofarabine 2 ( Figure 1) is a purine nucleoside analogue which, in its corresponding 5'-Otriphosphate form, inhibits both ribonucleotide reductase and DNA polymerases and prevents effective DNA synthesis. Ultimately, this leads to cell apoptosis, particularly in rapidly proliferating and dormant cancer cells. The nucleoside analogue exhibits excellent cytotoxic activity in vitro, with an IC50 range of 0.028-0.29 μM across a variety of solid tumour and leukaemia cell lines [63], alongside substantial tissue distribution and a half-life of at least 24 h for the active triphosphate metabolite [64]. The use of 2 for treatment of paediatric patients with relapsed or refractory acute lymphoblastic leukaemia was approved by the food and drug administration (FDA) in 2004, the first nucleoside analogue of its kind to be approved in over a decade [65].
In 2010, Sauve and colleagues reported an improved, stereoselective synthesis of 2 [66], in an overall yield of 38% (Scheme 13). This compared favourably to a prior report by ILEX Products Inc., which detailed an overall yield of 14% in six steps, starting from a fully protected ribose derivative. Sauve's work began from commercially available lactone 96 with 3,5-O protection using TIPS affording 97, which was then diastereoselectively fluorinated at the 2-position, obtaining the arabinoconfigured 98 exclusively in 72% yield. Conversion to the anomeric chloride (via hemi-acetal 99) delivered 100 which was condensed with 2,6-dichloropurine to afforded 101 in 66% yield over two steps and a β/α ratio of 3.5:1. Access to 2 was gained following diastereomeric separation of 101, 4position ammonolysis and deprotection.

2 -Furanose Modifications
Clofarabine 2 ( Figure 1) is a purine nucleoside analogue which, in its corresponding 5'-Otriphosphate form, inhibits both ribonucleotide reductase and DNA polymerases and prevents effective DNA synthesis. Ultimately, this leads to cell apoptosis, particularly in rapidly proliferating and dormant cancer cells. The nucleoside analogue exhibits excellent cytotoxic activity in vitro, with an IC 50 range of 0.028-0.29 µM across a variety of solid tumour and leukaemia cell lines [63], alongside substantial tissue distribution and a half-life of at least 24 h for the active triphosphate metabolite [64]. The use of 2 for treatment of paediatric patients with relapsed or refractory acute lymphoblastic leukaemia was approved by the food and drug administration (FDA) in 2004, the first nucleoside analogue of its kind to be approved in over a decade [65].
In 2010, Sauve and colleagues reported an improved, stereoselective synthesis of 2 [66], in an overall yield of 38% (Scheme 13). This compared favourably to a prior report by ILEX Products Inc., which detailed an overall yield of 14% in six steps, starting from a fully protected ribose derivative. Sauve's work began from commercially available lactone 96 with 3,5-O protection using TIPS affording 97, which was then diastereoselectively fluorinated at the 2-position, obtaining the arabino-configured 98 exclusively in 72% yield. Conversion to the anomeric chloride (via hemi-acetal 99) delivered 100 which was condensed with 2,6-dichloropurine to afforded 101 in 66% yield over two steps and a β/α ratio of 3.5:1. Access to 2 was gained following diastereomeric separation of 101, 4-position ammonolysis and deprotection.

2'-Furanose Modifications
Clofarabine 2 ( Figure 1) is a purine nucleoside analogue which, in its corresponding 5'-Otriphosphate form, inhibits both ribonucleotide reductase and DNA polymerases and prevents effective DNA synthesis. Ultimately, this leads to cell apoptosis, particularly in rapidly proliferating and dormant cancer cells. The nucleoside analogue exhibits excellent cytotoxic activity in vitro, with an IC50 range of 0.028-0.29 μM across a variety of solid tumour and leukaemia cell lines [63], alongside substantial tissue distribution and a half-life of at least 24 h for the active triphosphate metabolite [64]. The use of 2 for treatment of paediatric patients with relapsed or refractory acute lymphoblastic leukaemia was approved by the food and drug administration (FDA) in 2004, the first nucleoside analogue of its kind to be approved in over a decade [65].
In 2010, Sauve and colleagues reported an improved, stereoselective synthesis of 2 [66], in an overall yield of 38% (Scheme 13). This compared favourably to a prior report by ILEX Products Inc., which detailed an overall yield of 14% in six steps, starting from a fully protected ribose derivative. Sauve's work began from commercially available lactone 96 with 3,5-O protection using TIPS affording 97, which was then diastereoselectively fluorinated at the 2-position, obtaining the arabinoconfigured 98 exclusively in 72% yield. Conversion to the anomeric chloride (via hemi-acetal 99) delivered 100 which was condensed with 2,6-dichloropurine to afforded 101 in 66% yield over two steps and a β/α ratio of 3.5:1. Access to 2 was gained following diastereomeric separation of 101, 4position ammonolysis and deprotection.
The synthesis of 105 and 106 derived from a common 4-disubstituted acetate 102, available using established chemistry from diacetone-d-glucose (Scheme 14) [68,69]. The group used Vorbrüggen glycosylation of either 2,6-diaminopurine or 2,6-dicholoro-9H-purine to afford 103 or 104 respectively, with the substitution proceeding in good yields (100% for 103 and 64% for 104), but noting a requirement to control the amount of N,O-bis(trimethylsilyl)acetamide (BSA) used to silylate the purine to 2.5 equivalents. 2 ,4 -bridged compound 105 was constructed via base mediated cyclisation between C2 and C4 , followed by per-benzoylation of purine nitrogen and silicon protecting group removal. 2'-endo (South) systems [67]. These compounds were evaluated for their antiviral, antitumour and antibacterial properties, with 2',4'-bridged purines 105 and 106 showing µM inhibitory activity against CEM (IC50 = 0.36 µM for 105 and 7.6 µM for 106) and Raji (IC50 = 0.25 µM for 105 and 5.8 µM for 106) cancer cell lines. The synthesis of 105 and 106 derived from a common 4-disubstituted acetate 102, available using established chemistry from diacetone-D-glucose (Scheme 14) [68,69]. The group used Vorbrüggen glycosylation of either 2,6-diaminopurine or 2,6-dicholoro-9H-purine to afford 103 or 104 respectively, with the substitution proceeding in good yields (100% for 103 and 64% for 104), but noting a requirement to control the amount of N,O-bis(trimethylsilyl)acetamide (BSA) used to silylate the purine to 2.5 equivalents. 2',4'-bridged compound 105 was constructed via base mediated cyclisation between C2' and C4', followed by per-benzoylation of purine nitrogen and silicon protecting group removal. To access allyl substituted 2',4'-system, Stille couplings were completed at positions 2 and 6 of the purine, followed by a similar base-mediated C2' deacetylation and intramolecular cyclisation to afford, after C5'-protecting group removal, 106. Isomerisation of the allyl group at C6 of the purine ring was observed during the palladium and base-mediated cyclisation steps. The antitumour activities of 105 and 106 were considerably lower than for the known anticancer nucleoside cladribine (CEM IC50 = 0.5 nM and Raji IC50 = 9.0 nM), as was the activity (IC50 >10 µM) of the 2'4'-bridged analogue of cladribine, suggesting that the inclusion of this conformational restraint or the addition of an extra CH2O unit was enough to remove antitumour properties.

3'-Modified Nucleosides
Cheng and co-workers reported an asymmetric synthesis of 2',3'-dideoxy-3'-boronic acid pyrimidine nucleosides [70]. The group used two highly diastereoselective reactions of chiral boronic esters with (dihalomethyl)lithium reagents to install the ultimate stereochemistry required at C3' and C4' (Scheme 15). Starting from (S,S)-1,2-dicyclohexyl-1,2-ethanediol (DICHED) derived boronate ester 107, homologation was completed with (dichloromethyl)lithium to afford 108, the newly formed stereochemistry of which was inverted to give boronate ester 109. The chiral auxiliary component was next switched from DICHED to a pinanediol (derived from (+)-pinene), delivering 110. A second diastereoselective homologation was completed with (dibromomethyl)lithium, affording bromo boronic ester 111. Nucleophilic displacement and inversion of this bromide with allylmagnesium bromide followed by oxidative cleavage of the alkene afforded aldehyde 112. To access allyl substituted 2 ,4 -system, Stille couplings were completed at positions 2 and 6 of the purine, followed by a similar base-mediated C2 deacetylation and intramolecular cyclisation to afford, after C5'-protecting group removal, 106. Isomerisation of the allyl group at C6 of the purine ring was observed during the palladium and base-mediated cyclisation steps. The antitumour activities of 105 and 106 were considerably lower than for the known anticancer nucleoside cladribine (CEM IC 50 = 0.5 nM and Raji IC 50 = 9.0 nM), as was the activity (IC 50 >10 µM) of the 2 4 -bridged analogue of cladribine, suggesting that the inclusion of this conformational restraint or the addition of an extra CH 2 O unit was enough to remove antitumour properties.
group accessed C2-aryl iodide derivative 114 which was oxidised at C5' using KMnO4, followed by amide coupling to install the C5′-N-cyclopropylcarboxamido group. Following 2',3'-O-acetonide removal, the C2-iodide underwent a series of divergent Sonogoshira couplings to deliver a library of seven purine analogues. From this library, compounds 115 and 116 showed in vitro cytotoxic effects comparable to doxorubicin against human breast (MDA-MB-2312) and human colon (Caco2) cell lines.

5 -β-Hydroxyphosphonate Analogues
In 2014, Peyrottes and co-workers reported their synthesis of a series of β-hydroxyphosphonate nucleosides, targeting 2 and 3' hydroxyl group stereochemistry changes and the introduction of 5-position substituents to the nucleobase (Figure 9) [73]. This report built upon the group's earlier work developing this class of compound as 5'-nucleotidase inhibitors (specifically the cytosolic cN-II 5'-nucleotidase II) [74]. These enzymes catabolise nucleoside 5'-monophosphates and their expression level is of crucial interest for patients undergoing nucleoside analogue chemotherapy, with a higher expression level often associated with a worsened clinical outcome [75].

5'-β-Hydroxyphosphonate Analogues
In 2014, Peyrottes and co-workers reported their synthesis of a series of β-hydroxyphosphonate nucleosides, targeting 2' and 3' hydroxyl group stereochemistry changes and the introduction of 5position substituents to the nucleobase (Figure 9) [73]. This report built upon the group's earlier work developing this class of compound as 5'-nucleotidase inhibitors (specifically the cytosolic cN-II 5'nucleotidase II) [74]. These enzymes catabolise nucleoside 5'-monophosphates and their expression level is of crucial interest for patients undergoing nucleoside analogue chemotherapy, with a higher expression level often associated with a worsened clinical outcome [75]. The group completed the synthesis of a library of 32 β-hydroxyphosphonate analogues which included D-allo, D-altro, D-manno and D-gluco configurations of the furanose 2'-and 3'-positions along with C2'-C3' ring opened derivatives. They also installed alkynyl, aryl or ethenylaryl groups at C5 of the cytidine or uridine nucleobase using transition-metal cross-couplings of the corresponding C5-iodide. Biological activity data was obtained using recombinant, purified cN-II with inosine monophosphate as substrate. From the resultant SAR study, cytosine-based analogues were generally more active than their uracil counterparts, but the configurational isomers at C2' or C3' were generally less active than the parent compounds (allofuranose). Modification of the nucleobase was generally well tolerated with an observed Ki value of 1.14 mM for phenylethenyl substituted derivative 117 (Figure 9), noting that high substrate concentrations (in the mM range) are required for activity with cN-II.

Ferronucleosides
Tucker and colleagues recently described their synthesis of ferronucleosides, an important new development in the field of medicinal bioorganometallic chemistry [76]. Here the furanose ring was exchanged for the five-membered cyclopentadienyl ring of a ferrocene unit, but retained the nucleobase and hydroxymethyl group as key components appended in a 2,3 relationship to the ferrocene core (compounds 118 and 119, Figure 10). Using thymine or adenine as the nucleobase, the compounds were tested with a control series (where the hydroxymethyl or base component were The group completed the synthesis of a library of 32 β-hydroxyphosphonate analogues which included d-allo, d-altro, d-manno and d-gluco configurations of the furanose 2 -and 3'-positions along with C2 -C3' ring opened derivatives. They also installed alkynyl, aryl or ethenylaryl groups at C5 of the cytidine or uridine nucleobase using transition-metal cross-couplings of the corresponding C5-iodide. Biological activity data was obtained using recombinant, purified cN-II with inosine monophosphate as substrate. From the resultant SAR study, cytosine-based analogues were generally more active than their uracil counterparts, but the configurational isomers at C2 or C3' were generally less active than the parent compounds (allofuranose). Modification of the nucleobase was generally well tolerated with an observed K i value of 1.14 mM for phenylethenyl substituted derivative 117 (Figure 9), noting that high substrate concentrations (in the mM range) are required for activity with cN-II.

Ferronucleosides
Tucker and colleagues recently described their synthesis of ferronucleosides, an important new development in the field of medicinal bioorganometallic chemistry [76]. Here the furanose ring was exchanged for the five-membered cyclopentadienyl ring of a ferrocene unit, but retained the nucleobase and hydroxymethyl group as key components appended in a 2,3 relationship to the ferrocene core (compounds 118 and 119, Figure 10). Using thymine or adenine as the nucleobase, the compounds were tested with a control series (where the hydroxymethyl or base component were absent) for their cytostatic activity and compared to established chemotherapeutic agents 5-fluorouracil (5F-U) and cisplatin. In a proliferation activity assay on three tumour cell lines (L1210, HeLa and CEM), both 118 and 119 had activities in the low µM range, with 118 and 119 20 to 50-fold more active than 5F-U in CEM cell cultures (0.9 µM for 118, 0.35 µM for 119 versus 18 µM for 5F-U). The compounds also performed promisingly in cell growth inhibition (oesophageal cancer cell line) and cellular viability (MTT assay) studies, with the data indicating that both functional groups appended to the ferrocene component were required for optimum cytostatic activity.

Phosphorodiamidate Prodrugs
In 2018, Slusarczyk's team reported the synthesis of phosphorodiamidates for a series of established anticancer nucleoside analogues [77]. This was envisioned following the successful ProTide technology developed by McGuigan [78], to deliver nucleoside monophosphate analogues into cells and recent reports of phosphorothioamidate systems [79]. A lack of chirality at phosphorous in a phosphorodiamidate was envisioned to confer advantages in not having to resolve diastereomeric mixtures, often a requirement in earlier generations of ProTides, as SP and RP diastereoisomers exhibited markedly different biological activities [80]. Accordingly, seven different anticancer nucleosides (1, FUdR, 8-chloroadenosine, fludarabine, AraG, thioinosine and thioguanosine) were converted to their phosphorodiamidate form using a one-pot, two-stage strategy (Scheme 17). First the 5'-OH was converted to a phosphorodichloridate intermediate 120, followed by double phosphoramidation using an appropriate amino acid ester salt. The panel of analogues were then evaluated in vitro against a wide range of solid tumour and haematological cell lines with the potential for this approach confirmed for FUdR and 8-chloroadenosine, where similar or improved inhibitory activities compared to the parent nucleosides were observed. FUdR and its phosphorodiamidate prodrug showed activity in the sub-micromolar range with IC50 values of 0.0046-0.073 µM for FUdR and 0.01-0.40 µM for the FUdR phosphorodiamidates against the wildtype cell lines. Also of note was the inactivity of arabino configured analogues, suggesting conversion to the monophosphate might be prevented through their being poor substrates for phosphoramidase activity [81]. Enzymatic studies were undertaken to investigate the bioactivation pathway of this class of nucleoside prodrug with a carboxypeptidase Y assay and 31 P-NMR confirming the hydrolysis of both esters followed by loss of one phosphoramidate group.

Phosphorodiamidate Prodrugs
In 2018, Slusarczyk's team reported the synthesis of phosphorodiamidates for a series of established anticancer nucleoside analogues [77]. This was envisioned following the successful ProTide technology developed by McGuigan [78], to deliver nucleoside monophosphate analogues into cells and recent reports of phosphorothioamidate systems [79]. A lack of chirality at phosphorous in a phosphorodiamidate was envisioned to confer advantages in not having to resolve diastereomeric mixtures, often a requirement in earlier generations of ProTides, as S P and R P diastereoisomers exhibited markedly different biological activities [80]. Accordingly, seven different anticancer nucleosides (1, FUdR, 8-chloroadenosine, fludarabine, AraG, thioinosine and thioguanosine) were converted to their phosphorodiamidate form using a one-pot, two-stage strategy (Scheme 17). First the 5'-OH was converted to a phosphorodichloridate intermediate 120, followed by double phosphoramidation using an appropriate amino acid ester salt. The panel of analogues were then evaluated in vitro against a wide range of solid tumour and haematological cell lines with the potential for this approach confirmed for FUdR and 8-chloroadenosine, where similar or improved inhibitory activities compared to the parent nucleosides were observed. FUdR and its phosphorodiamidate prodrug showed activity in the sub-micromolar range with IC 50 values of 0.0046-0.073 µM for FUdR and 0.01-0.40 µM for the FUdR phosphorodiamidates against the wild-type cell lines. Also of note was the inactivity of arabino configured analogues, suggesting conversion to the monophosphate might be prevented through their being poor substrates for phosphoramidase activity [81]. Enzymatic studies were undertaken to investigate the bioactivation pathway of this class of nucleoside prodrug with a carboxypeptidase Y assay and 31 P-NMR confirming the hydrolysis of both esters followed by loss of one phosphoramidate group. 0.0046-0.073 µM for FUdR and 0.01-0.40 µM for the FUdR phosphorodiamidates against the wildtype cell lines. Also of note was the inactivity of arabino configured analogues, suggesting conversion to the monophosphate might be prevented through their being poor substrates for phosphoramidase activity [81]. Enzymatic studies were undertaken to investigate the bioactivation pathway of this class of nucleoside prodrug with a carboxypeptidase Y assay and 31 P-NMR confirming the hydrolysis of both esters followed by loss of one phosphoramidate group. Scheme 17. Reagents and conditions: (i) (Me)3PO4, POCl3 and (ii) Amino acid ester salt, DIPEA. X and Y represent different functional groups for analogue classes e.g. X = Y = F, Base = C = gemcitabine phosphorodiamidate. See Reference [77] for full list of analogues synthesised and evaluated.

Vitamin E Phosphate Prodrugs
Nucleoside analogue therapy can suffer from inducive and constitutive resistance, which limits the efficacy of the treatment. Isoforms of vitamin E, in particular δ-tocopherol and tocotrienols have displayed anticancer activity. As such, vitamin E phosphate nucleoside analogue prodrugs were envisaged to combat two mechanisms of resistance: i) downregulation of metabolic kinases (deoxycytidine kinase, dCK, in the case of 1) and ii) nucleoside transport [82,83]. Accordingly,

Vitamin E Phosphate Prodrugs
Nucleoside analogue therapy can suffer from inducive and constitutive resistance, which limits the efficacy of the treatment. Isoforms of vitamin E, in particular δ-tocopherol and tocotrienols have displayed anticancer activity. As such, vitamin E phosphate nucleoside analogue prodrugs were envisaged to combat two mechanisms of resistance: i) downregulation of metabolic kinases (deoxycytidine kinase, dCK, in the case of 1) and ii) nucleoside transport [82,83]. Accordingly, Daifuku and co-workers synthesised and evaluated four isoforms of vitamin E conjugated to 1 (Figure 11, compounds 121-124) [84]. Daifuku and co-workers synthesised and evaluated four isoforms of vitamin E conjugated to 1 ( Figure 11, compounds 121-124) [84]. The in vitro GI50 of 1, vitamin E phosphate (VEP) isoforms and compounds 121-124 showed these VEP-gemcitabine prodrugs to exhibit anticancer activity, consistent with their catabolism to vitamin E and gemcitabine monophosphate. Conjugate 122 displayed the best activity with GI50 values <5 μM in breast MDA (MB-231), non-small cell lung (NCI-H460) and colon (HCT-116) cell lines. The authors suggested this was due to steric hindrance, from the methyl groups proximal to the vitamin E-phosphate bond, reducing the rate of enzymatic cleavage to the monophosphate counterpart.
Prodrugs 122 and 123 were then evaluated in the presence of an inhibitor of nucleoside transport, dipyridamole (DP, Table 5) with the data indicating that both were largely unaffected by the presence of DP in comparison to 1. This suggested that these prodrugs bypass nucleoside membrane transporters and may thus be beneficial in the treatment of patients with gemcitabine resistant cells.  The in vitro GI 50 of 1, vitamin E phosphate (VEP) isoforms and compounds 121-124 showed these VEP-gemcitabine prodrugs to exhibit anticancer activity, consistent with their catabolism to vitamin E and gemcitabine monophosphate. Conjugate 122 displayed the best activity with GI 50 values <5 µM in breast MDA (MB-231), non-small cell lung (NCI-H460) and colon (HCT-116) cell lines. The authors suggested this was due to steric hindrance, from the methyl groups proximal to the vitamin E-phosphate bond, reducing the rate of enzymatic cleavage to the monophosphate counterpart.
Prodrugs 122 and 123 were then evaluated in the presence of an inhibitor of nucleoside transport, dipyridamole (DP, Table 5) with the data indicating that both were largely unaffected by the presence of Molecules 2020, 25, 2050 20 of 25 DP in comparison to 1. This suggested that these prodrugs bypass nucleoside membrane transporters and may thus be beneficial in the treatment of patients with gemcitabine resistant cells. In particular, as conjugate 123 displayed significant potency against the three DP(-) cell lines and moderate potency against the DP-dosed cell lines (Table 5), it was selected to further compare activity against 1 in in vitro wild-type leukemic CEM cells and CEM cells deficient in dCK. In dCK(−) cells 1 is not phosphorylated, with GI 50 values increasing from 0.002 µM in wild-type to 124.5 µM in dCK(−). For 123 the GI 50 increase was significantly lower than that of 1 (from 0.59 µM to 19.2 µM). Furthermore, the half-life of 123 in mice was shown to be 4 h, a 13.9-fold increase to that of 1 (0.3 h [85]). Overall, this study showed an interesting proof of concept for VEP-1 prodrugs against resistant cancer cell lines. However, further optimisation will be required to obtain promising candidate compounds with increased potencies.

Conclusions
Nucleoside analogues are an historically accomplished class of drugs with highly diversifiable scaffolds and proven potential to treat a wide range of cancer cell types. Recent synthetic trends have focused on furanose oxygen substitution with heteroatoms or carbon and fluorocyclopentenyl cytosine is proving a promising new clinical candidate in this regard. Alongside this, examples of templating such modifications onto established nucleoside analogue scaffolds, such as Ara-C and forodesine, are emerging. As further pharmacokinetic and pharmacodynamic parameters are evaluated for these architectures, their pharmaceutical utility will be established. Finally, the continued advancement of prodrug strategies to deliver these compounds more effectively and provide options to treat drug-resistant cancer cell types sets an exciting future for nucleoside analogue chemotherapeutics and the underpinning requirement of chemical synthesis in realising this.